New Frontiers on the Information Superhighway:
INTERNET TELEPHONY FORUM
White Room
National Press Club
529 Fourteenth Street, N.W.
Evolving Applications --
Demonstrations of basic Internet telephony and telephony-related applications.
Panel III -- Future Directions This panel discussed Internet Telephony applications to expect, technical challenges and market impediments, and policy approaches.
Moderated by:
Closing Remarks --Larry Irving, NTIA
141
1 AFTERNOON SESSION
2 (1:30 p.m.)
4 MS. BROWN: Once again, I'd like to
5 welcome you back to NTIA's Forum on Internet
6 telephony. I'd like to welcome our audience, who
7 is here in the room at the National Press Club,
8 and also for our audience who is on the Net.
9 For those of you who haven't had a
10 chance to see our site, it's set up in the demo
11 room. It's actually going quite well. I'm
12 pleased. I hope some folks who are out there
13 will let us know how the quality is and if it's
14 going well for you. From our end it looks okay,
15 so let us know.
16 That gives me an opportunity, if you'll
17 allow me just for a moment, to thank staff at
18 NTIA for the kind of work they put in to put this
19 day together. Our technical folks have been
20 really terrific. Dan Davis, Charlie Franz, and
21 John Mukai are the fellows who are running around
22 doing the technical stuff and have got us up on
23 the Net and did our Web page and we are pleased
24 with that. Ann Stauffer has done a terrific
25 job. Beth Austin, Cindy Bens and Charmaine
142
1 Bosswell, let me just publicly thank them.
2 As I said to you, Fred Lee has done a
3 lot of the conceptualization of the day and I
4 appreciate that. I want to also just mention
5 that the VON Coalition, particularly Sandy Combs,
6 who is someplace, and Jeff Pulver have been a
7 great help in trying to put some things together
8 here, for putting the demo room together, for
9 graciously offering services that the government
10 can't offer you, like something wet to drink.
11 With that, I'm going to turn this back
12 to Fred so we can get underway with the show and
13 tell portion of our program. Fred.
14 MR. LEE: Thank you, Kathy. Welcome
15 back. Over the next 45 minutes we are going to
16 see a demonstration of Internet telephony
17 applications.
18 We are fortunate enough to have two
19 pioneers in the field. Jeff Pulver is clearly
20 well-known to many of us and has been
21 characterized as one of the true believers in
22 Internet telephony. Shane Mattaway is the
23 executive vice president and chief technical
24 officer of NetSpeak. I think Jeff is going to go
25 first and demonstrate PC-based and also
143
1 gateway-oriented. And then Shane with Jeff's
2 help is going to demonstrate applications
3 involving Web-based call centers. Good afternoon. Thanks
5 for the introduction. I'd like to take a moment
6 to thank the NTIA for having us work with them.
7 For those of you not familiar with the VON
8 Coalition, we were set up last year sort of as a
9 response to the ACTA Petition. A group of long
10 distance phone companies got together and
11 petitioned the FCC asking for two things that
12 were fair: One was to ban the use of Internet
13 telephony software in the United States -- and,
14 oh, by the way, if you should happen to have such
15 software, it should be regulated like a telco.
16 Many people here today were very
17 instrumental in helping to make it become real.
18 Quickly, after we served the coalition, while we
19 were doing our filings with the FCC, we had a
20 common-ground meeting, and I'm much appreciative
21 to Katherine Brown for the image that she left to
22 everyone who participated in the first
23 common-ground meeting in what it's like to deal
24 directly with the government and to in fact find
25 common ground amongst everybody. We are here
144
1 today because of everyone's collective efforts.
2 It's been a real fun run for myself.
3 What I'm here today to do is provide an
4 overview of Internet telephony, not so much the
5 history, but where we are today. And I will take
6 you through hopefully a demonstration, phone to
7 phone, that is, calling a computer which will
8 then call another computer on the Internet which
9 will then place a phone call just to demonstrate
10 what the quality is. We have gone over this.
11 Just to set the groundwork, Internet
12 telephony as classic PC to PC, multimedia PC to
13 multimedia PC, became very much popular in 1995,
14 PC to phone, the first original dial-up program
15 and phone to phone. Internet telephony is not
16 about free long distance. If the International
17 Bureau of the FCC has their way, we will in fact
18 see accounting rate parity between the United
19 States and our 50 largest trading partners.
20 It's really not Internet telephony
21 per se but it's voice over IP. If you have
22 TCP/IP connectivity, that's where the technology
23 plays.
24 What's needed? What are the
25 components? Well, depending upon the software,
145
1 because there are people that claim this runs on
2 a 386, I find that a Pentium class machine is
3 good, a P100 or faster, at least a 14.4
4 connection to the Net, although as fast as you
5 can get is good. You need a microphone. You
6 need speakers. You need an ability to tolerate
7 less than ideal conversations and conditions.
8 What I mean by that is to the extent the Internet
9 is an environment that is not a production center
10 per se. Routers go around all over the world.
11 You may experience packet loss.
12 I'm a ham radio operator. I enjoy
13 speaking to somebody under any adverse
14 conditions. For me to communicate with somebody
15 on the Internet, it wasn't so much how good they
16 sounded but the fact that I was talking to
17 somebody. Over the next two and a half years as
18 we move forward, certainly the requirements are
19 much greater and the technology is drastically
20 improved. But if you want to talk to someone
21 over the Internet, all bets are off sometimes as
22 far as the quality.
23 Of course, you do have to have this
24 desire to speak to total strangers. It just
25 happens that while you can set up phone calls and
146
1 you can speak to some of your friends and
2 relatives, a large population using Internet
3 telephony on the Internet, not in the business
4 world but in the hobbyist arena, are talking to
5 people just for the sake of talking. These are
6 people using the Internet just for the purpose of
7 entertainment.
8 Again, what's needed. You need to have
9 a good computer, speakers, microphone and a
10 desire to communicate with others.
11 What's a gateway? In very simplest
12 terms, a gateway speaks to the IP networks in the
13 public-switched telephone networks. It has a
14 DSP, it has a network card, PSTN connection and
15 some software. Don't look for all of these
16 products to be scaleable yet.
17 If you happen to be a telco, you want
18 to run 10,000 ports out of your hop, well unless
19 you physically have a large amount of room, the
20 technology is getting there, but it's not there
21 today.
22 Who is buying gateways today? Research
23 labs, competitors, ISPs and selected
24 corporations. Those guys are the ones with the
25 vision trying it today. I think technology plays
147
1 very well for the Fortune 500s or areas where you
2 have a mobile sales force where you want to be
3 able to extend your ISP connectivity all the way
4 around the world.
5 1997 is the year of the trial. What's
6 happening? Among other things, lots of strategy
7 announcements including noise from AT&T, MCI,
8 Sprint, cable and wireless and BT.
9 Everybody is playing with Internet
10 telephony. Some publicly, some privately. The
11 challenge here is if you are a telco is not to
12 become disadvantaged by not playing. Everyone
13 has an Internet telephony strategy. The
14 corporation, we are starting to see the adoption
15 of a single wire theory. This is part of what
16 Dan was talking about, although not directly: the
17 idea of having a single wire, maybe a gigabyte
18 LAN to have voice and data services on there.
19 During this year, we have seen
20 interoperability of progress. We are getting
21 there. I'm looking at -- if you happen to be a
22 vendor, it's about a $150 million market this
23 year and there are about 65 consortiums that have
24 been formed to actually work together to provide
25 voice over IP services; if you are an ISP, both
148
1 fax and voice. So by midyear, about 14 companies
2 have announced these gateway products.
3 These are companies that you may or may
4 not have heard of. They are coming from the
5 telecom world: Cisco, Bayh, Ascend; telecom
6 equipment vendors like Nortel, Lucent, Ericcson
7 and VocalTec and NetSpeak.
8 If you look for strategic relationships
9 to drive the marketplace, announcements like
10 Telcom's investment in VocalTec is an example and
11 there are many others. As the market grows, you
12 will see strategic players. People not expecting
13 to work together are working together.
14 Lots of issues are remaining, issues
15 like directory services. I can call Information
16 with the telephone. In the real world, you
17 associate a telephone number with a place. On
18 the Internet, you are associating with a person,
19 per se. It's kind of hard because when you are
20 mobile, how do you track that person? You can
21 call Information and find out by phone number
22 where I live. If I'm on the Internet, using one
23 of 38 products, how do you find me if you want
24 to? It's kind of hard. There is technical
25 discussions and I won't get into that here.
149
1 There are quality service improvements.
2 The big issue as far as I know are
3 accounting and settlements. About six months
4 ago, I made a presentation at ITU in Geneva to
5 the World Telecommunications Council, basically
6 the Board of Governors of the ITU. While we
7 agree that RSVP was going to happen in 1996 and
8 we are still waiting for it to happen, the real
9 issue driving them is the ability to charge for
10 it.
11 And the big issue is how to do the
12 settlements. How to charge for it, how they
13 internationally will be able to do that is going
14 to help drive, ultimately, the quality service
15 that you, as a telephony or other data services
16 on the Net, is going to drive your future.
17 The ACTA Petition still remains. Sandy
18 Combs, most recently last week in Reno, made a
19 presentation. At the VON conference next week,
20 you may find ACTA exhibiting. The ACTA Petition
21 still remains on the books.
22 While I live in Internet years, the FCC
23 and others don't. I really think we should
24 resolve this once and for all. The VON Coalition
25 is at WWW.VON.ORG.
150
1 We have a definition of some of them
2 available on our Web site. If you are an ISP, we
3 are talking about enhanced services. There is
4 convergence going on. People used to sell
5 traditional telephony communications. They are
6 doing really neat things with virtual PBXs.
7 Is anyone here in the room familiar
8 with Free World Dial Up? I don't want to bore
9 you. If you can imagine a group of people
10 getting together trying to test new technology
11 and at the same time trying to impact the world
12 around them.
13 It was back in September of '95, I used
14 to work on Wall Street. Some friends of mine
15 were gainfully employed as well. We figured out
16 from the radio world phone patching was a cool
17 thing so that at my parents' expense during
18 holiday seasons, I would take those calls from
19 Antarctica and route them through Michigan. I
20 had fun. I was making these people happy.
21 Moving the markets forward a few years,
22 I was there on the Internet talking to these
23 people about enterprise computers. What about
24 the future? We took our computers and patched
25 them into the telephone. Is there a service? Is
151
1 it possible. Turns out it took $100 and a sound
2 card and some programming and for a few months it
3 was really cool. I had a guy in Tokyo
4 coordinating server locations and a gentleman in
5 Indonesia writing code. They all got new jobs
6 and they walked away from it and I ended up
7 dedicating my full-time efforts to Internet
8 telephony.
9 If we can provide a service which
10 doesn't compete because frankly it doesn't
11 scale -- we are talking if you have a presence in
12 New York State, it's two phone ports, two phone
13 lines. If too many people find out about it, you
14 get busy signals. After a week of operations, a
15 gateway had to be shut down because too many
16 people were abusing it. Learn about human nature
17 about this.
18 Anyway, Free World Dial Up, we have
19 locations around the world. I restarted this a
20 few months ago. This time around I'm embracing
21 telcos, government agencies, everybody and
22 anybody that wants to learn about the Internet.
23 Some of the major players are these
24 PTTs. They are playing anonymously. If you look
25 at some of the countries I'm operating in you can
152
1 guess. What they get out of it is knowledge.
2 They get a knowledge of what it's like to run a
3 long distance service over the Internet. No
4 commitments other than a dedication to the
5 cause. The idea is it's a free service just like
6 ham radio. It's open really for friends,
7 families and expatriates.
8 We are running it -- actually for the
9 last few hours Free World Dial Up, Washington,
10 D.C. has been running. What I'm going to attempt
11 to do is place a call from my gateway New York to
12 my gateway New Jersey just to hear what it sounds
13 like.
14 Again, we pick up a telephone. We call
15 computer. That computer is peering with another
16 computer over the Internet 15 hops away. Phone
17 to another person's phone through the Internet.
18 If you want information, this is --
19 someone sent me a picture of their daughter.
20 I set up these -- because I let people use the
21 system. Not only can it go phone to phone, it
22 can also go from PC to phone. Some woman sent me
23 a picture of her daughter using it.
24 So let me real quickly, before I run
25 out of time, go into, just try to do a quick
153
1 tempo if I can. I have slides here about dealing
2 with the effective voice on the Net. It
3 dovetails into what Dan Berninger was speaking
4 about earlier. If I can, I'm going to just try
5 to dial a call.
6 I'm going to make a long distance
7 call. I'm calling long distance first actually.
8 I just actually just called my gateway in New
9 York.
10 OPERATOR: Please dial the number you
11 are calling.
12 MR. PULVER: I don't want to broadcast
13 the touchtones.
14 OPERATOR: Please dial the number you
15 are calling. Include the country code and area
16 code if needed.
17 MR. PULVER: 750-8772. Now it should
18 ask me for my user ID.
19 OPERATOR: Enter your personal ID.
20 MR. PULVER: Authenticated, did a
21 hopoff. It's going from New York right now to
22 New Jersey, and placing a call to a friend of
23 mine, Tom, who works at VocalTec. If this
24 doesn't work, I'll call the New York City weather
25 so you can at least hear the quality of it. I
154
1 don't hear any ringing. This is not such a good
2 thing.
3 OPERATOR: Your call did not go
4 through.
5 MR. PULVER: Let me try again.
6 OPERATOR: We are sorry, your call did
7 not go through.
8 MR. PULVER: It may tell me the call's
9 terminated. If it does, I'll be able to do a
10 live demo. I'll wait for that to recycle just in
11 case it recycles.
12 The idea here is that the quality is
13 pretty okay. I'll hang up. All right. I will
14 invite everybody to come in the other room if
15 they would like to hear it. I might actually try
16 to get my demo going while Shane is doing his,
17 just to see if it works. The idea here is that
18 the quality is pretty okay. There is latency.
19 It's getting much better.
20 The technology is enabling, to allow
21 people like me to actually do live demos or at
22 least try them. You can hear the operator on the
23 other side complaining. That's pretty okay. I
24 run about three hours a day on my New York
25 switch, my New York switch being on that computer
155
1 which is now here.
2 Although it's amazing, people do
3 complain even though you run a free service. I
4 don't know why, but they do. If I have a few
5 minutes left, I just want to go through my --
6 well, here is who is making money on the Net
7 today.
8 It's just facts to deal with. Adult
9 entertainment industry, like it or not, they are
10 really making very much money. Next generation
11 telcos are doing it. Great business
12 opportunities for corporate consultants.
13 VocalTec, NetSpeak, and Voxware are now public.
14 Can you backtrack?
15 MR. MATTAWAY: It I can.
16 MR. PULVER: Is that -- yes. Just real
17 fast, we'll go through these. What are the
18 effects of voice on the Net? Reduced barriers to
19 entry via enabled incremental deployment of
20 infrastructure. Internet telephony is one of the
21 main motivations for countries to accept
22 cost-based settlements, and the movement of
23 intelligence out of the network to the edges
24 enables fantastic service creation
25 opportunities.
156
1 The Internet is offering new ways to
2 provision customers and deliver information to
3 them. The cost of connectivity remains
4 proportional to the duration of the connection
5 thus to the extent packetization of continuous
6 data times are possible, packet switches do it
7 most of the time.
8 Rather than viewing voice on the Net as
9 a threat, most major telcos embrace this
10 technology and will be using it to create
11 enhanced services. Migration of computer
12 industry based solutions into traditional
13 telecommunications represents a great thing for
14 end users. In other words, rather than being in
15 a business that grows at 7 percent per year, you
16 could be in a business that grows at 30 percent
17 per year. Voice services are being driven to a
18 commodity.
19 Here's the scary part. Maybe 25 years
20 is a wrong number but it's not 125 years. Within
21 a certain period of time, almost everybody will
22 be able to offer a telephone service. The
23 challenges are to Telstra and AT&T and others to
24 redefine themselves. Gateways may become popular
25 as fax servers. Is it death of the PSTN? No.
157
1 Look for leading edge corporations to begin to
2 offer VOIP services. Corporate trials are
3 already underway. If you have any questions, you
4 can E-mail at Jeff Pulver.com. With that, I'll
5 transition to Shane and if we have time at the
6 end, I'll try placing that call again. Thank you
7 very much.
8 (Applause.) I'm Shane Mattaway with
10 NetSpeak Corporation. It's visible. As our
11 colleagues have been telling you that IP
12 telephony is really a major paradigm shift in
13 major communications. One of the major
14 value adds that IP telephony provides over
15 conventional telephony is the ability to
16 communicate with more media types such as video,
17 sharing applications, things of that nature.
18 What I'm going to do for you today is demonstrate
19 for you some of the other IP telephony services
20 that can become available to you as carriers or
21 corporate customers.
22 What I'm going to show you here is
23 effectively Web-based call centers. NTIA, I
24 guess is part of the Department of Commerce, so
25 it seems appropriate to discuss how IP telephony
158
1 services can be deployed to essentially
2 revolutionize electronic commerce.
3 One, with Web-based call centers,
4 customers are able to contact and communicate
5 with sales agents, support agents while they are
6 on the Internet while they are browsing without
7 requiring an additional telephone line.
8 In addition, Web-based call centers can
9 provide the multimedia communications between the
10 agent and the customer in the form of realtime
11 voice, video, agent-initiated multimedia content
12 push, collaborating by working on the same form
13 or document at the same time.
14 In addition, Web-based call centers
15 return personalized human services back to
16 Web-based sales and support, giving the customer
17 more secure, comfortable customized interaction.
18 We started out with telephone sales and
19 then we went to bulletin board services where you
20 would buy things off bulletin boards. We went to
21 Web-based type commerce. You didn't talk to
22 anybody. You lost the human touch.
23 With Web-based call centers, now you
24 are bringing back that human touch and it makes
25 the customer, and I'm sure you all realize this,
159
1 much more comfortable in giving a credit card
2 over the Internet, to be able to provide them
3 with that customized service that they really
4 need.
5 Until 20, 30 years from now where we
6 have artificial intelligence systems with IVR and
7 speech recognition where you are going to
8 eliminate the human, this is the best we are
9 going to do.
10 In addition, the Web-based call centers
11 can deliver automatically demographically located
12 content. Over and above just pushing a Web page,
13 you can deliver forms. You can actually speak to
14 -- a Web call center can speak to the customer
15 in his own native language and the like.
16 Another major advantage of Web-based
17 call centers is that agents can be located
18 anywhere within network reach. They do not need
19 to be located in boiler rooms or in the same
20 facility as a conventional ACD. So that your
21 agents could now be at home, on the road, as I
22 say, on a boat somewhere, on a notebook computer
23 with a cellular modem, wherever you want to get
24 them. It also enables customers and agents to
25 utilize the conventional telephone system the
160
1 organization already has.
2 Lastly, it seamlessly integrates with
3 the traditional POTS-based call centers.
4 A distributor can deploy IP services and
5 essentially leverage their existing equipment to
6 provide Web-based call centers.
7 The first thing I am going to do here
8 is go to NetSpeak's home page. I want to speak
9 to a sales representative. I would simply click
10 on an icon in which it would initiate a call to
11 our ACD. Now, our ACD is nothing but a piece of
12 software running on a PC with Microsoft
13 Windows NT as the operating system. I'm going to
14 an operating agent.
15 AUDIENCE MEMBER: Hello, Shane, can you
16 hear me?
17 MR. MATTAWAY: I'm going to call you
18 back. Put yourself on Do Not Disturb for 60
19 seconds. I did not show them the IVR.
20 OPERATOR: You got it.
21 MR. MATTAWAY: He wasn't ready. He is
22 actually an agent on the ACD. Let me just give
23 you a little display here what's going on. I am
24 out here on the Web as a customer. I'm on the
25 Internet through an Internet service provider.
161
1 When I clicked on that Web page, it initiated a
2 Web phone call to this ACD which happens to sit
3 on NetSpeak's land back in Boca Raton, Florida.
4 That ACD went and looked for any available agents
5 that might be out there in that split, in that
6 queue.
7 It turns out that Erik, this is a
8 control center. This is my administrative
9 console for the VACD. Erik is now in Do Not
10 Disturb. I can initiate the callback and I'll
11 get the IVR and be placed in a queue. I'll be
12 redirected to Erik once he comes back online,
13 wherever he is. He can be in the Four Seasons
14 Hotel and be on a PC like I am and retrieve that
15 call as a sales call or support call.
16 Let's see here now, if I click on the
17 page and Erik is idle, then I'm going to be
18 thrown into a queue because there are no
19 available agents.
20 OPERATOR: Thank you for calling
21 NetSpeak Corporation. All of our representatives
22 are assisting other customers. Please stay on
23 the line.
24 MR. MATTAWAY: It's going to push
25 consent to me. It's going to direct my browser
162
1 to whatever you want to go to while I'm in the
2 queue. If I were to call support or a help desk
3 at a company and while I'm waiting for a support
4 agent, for example, I might be able to solve the
5 problem by navigating the customer through his
6 browser. The ACD could also push forms for the
7 customer to fill out while they are waiting in
8 the queue. Now, I'm on hold here. I turned off
9 the hold music.
10 OPERATOR: Thank you for continuing to
11 hold. Please stay on the line.
12 MR. MATTAWAY: It's continuing to push
13 content to me. Erik is in use. I'm being
14 transferred to Erik at this point.
15 ERIK: Hello.
16 MR. MATTAWAY: How are you doing?
17 ERIK: I'm fine. How are you doing
18 today.
19 MR. MATTAWAY: I'm fine. Do you want
20 to sell me something? It will help if I talk
21 into the phone. I'm sorry, Erik, you are waiting
22 for me to say something. Go on.
23 ERIK: Hello. Everything is fine. Are
24 you just not talking into the mike?
25 MR. MATTAWAY: I was talking into a
163
1 microphone. So what are you going to sell me?
2 ERIK: I'm going to sell you a couple
3 ACDs and a couple dozen gateways. I think there
4 is a special this week on gateways.
5 MR. MATTAWAY: Are these Dialogic-based
6 gateways?
7 ERIK: Oh yes.
8 MR. MATTAWAY: Give me a smile. I want
9 to capture your face. Erik, I'm done with you.
10 I appreciate it.
11 ERIK: All right. No problem.
12 Good-bye.
13 MR. MATTAWAY: What happens is when
14 agents aren't available, you get thrown into a
15 queue. I'm actually running an administrative
16 control center remotely here so that I can see
17 the agents that are in the split. Erik is on Do
18 Not Disturb. I'm an agent but I'm offline.
19 There are other splits with other agents, and you
20 can simply drag and drop agents from one split to
21 another to, if you are monitoring your automatic
22 call distributor.
23 Now, I'm going to come back here to the
24 original page and I want to call a support line.
25 Before I do that, let me bring up this. This is
164
1 where the ACD was deployed on the company site as
2 a piece of software. For a Web phone to agent,
3 Web phone call like I made through an ACD, an ISP
4 or IP carrier could host the services and
5 generate revenue by charging their corporate
6 servers for the equipment. It's Centrex. In
7 this example, we just moved the ACD and the
8 connection services to an ISP and just gave agent
9 Web phones to the customer in a control center.
10 They essentially view their own ACD but it's not
11 on their premise. They're charged a monthly fee
12 by the ISP.
13 The next example is I'm going to
14 effectively treat an agent, use a regular phone
15 for an agent. A company that has a switch
16 already and has regular phones could simply
17 deploy the ACD and a NetSpeak gateway to
18 essentially utilize their existing telephone
19 equipment for agent phones instead of using the
20 computer.
21 I'll demonstrate that by calling
22 support at NetSpeak and that's going to ring that
23 phone. That telephone sits right here where I'm
24 pointing. It's on the public switch telephone
25 network. The gateway on the corporate LAN is
165
1 configured to act as a virtual agent phone on
2 behalf of the ASD on behalf of the line going out
3 to the PBX so that an inbound call comes into the
4 ACD, it sees the line from the gateway or the
5 virtual agent line on the gateway as being in use
6 and busy. I'm leveraging existing telephones.
7 At this point in time, that agent, he
8 is not busy. I'm not going right to AVR. I
9 wasn't put in a queue. Can you hear me all right
10 over there?
11 MR. PULVER: I hear you just fine.
12 MR. MATTAWAY: Do you want to sell me
13 something?
14 MR. PULVER: How about some gateways.
15 MR. MATTAWAY: That's good. I'll take
16 a dozen of them.
17 MR. PULVER: Sold.
18 MR. MATTAWAY: Thanks, kid. What that
19 demonstrates is the ability to leverage your
20 existing switch and phones for a Web-based call
21 center.
22 Now, another example, and I'm going to
23 become an agent. I'm going to show you how a
24 regular telephone caller, the traditional call
25 center environment, someone gets on a telephone,
166
1 dials an 800 number, comes into an existing ACD
2 to an agent on a regular telephone. Here you can
3 take a regular telephone call, go through the
4 telephone network, go to gateway and the ACD will
5 look for agents anywhere on the planet and then
6 connect that caller from the regular telephone to
7 that agent, whether he is on the corporate LAN or
8 remotely, as I'm going to show you an example.
9 So I'm going to be an agent here. I'm
10 coming online. As I come online, I'll show you
11 through the control center that I'm no longer
12 offline here. I'll go back to the demo sales
13 split. I am no longer offline -- I am offline.
14 I'm idle. There I am. See, you don't see it. I
15 have popped up. Here I am.
16 What's happened is the ACD as an agent
17 phone has delivered to me a form that lets me put
18 my status in as I'm handling the call so that the
19 ACD knows the status of the agent.
20 I'm going to go over here as the
21 customer and make a call. I need you to come
22 with me. I'm calling in over the network. It's
23 going in through a gateway into the ACD and I'm
24 going to get IVR. Wait a minute.
25 OPERATOR: Thank you for calling the
167
1 NetSpeak Corporation. For technical support,
2 press 2.
3 MR. MATTAWAY: You can just talk.
4 There I am, I have a customer coming in from the
5 gateway. NetSpeak support, can I help you?
6 MR. PULVER: Yes. I'm having some
7 trouble with a gateway.
8 MR. MATTAWAY: That's not unusual in
9 this state of the game, but I appreciate your
10 call, how can I help you?
11 MR. PULVER: Can you just tell me how
12 to do that configuration at line 27, please?
13 MR. MATTAWAY: I'll E-mail it for you.
14 Thanks for calling.
15 MR. PULVER: Thank you.
16 MR. MATTAWAY: I as an agent could have
17 been anywhere on the planet fielding calls from
18 what normally would be a customer out there on
19 the PSDN making inbound calls to a call center.
20 MR. PULVER: You really could be on
21 your boat, too.
22 MR. MATTAWAY: Definitely. I could
23 have done it on the boat. This could be deployed
24 by carriers as well, by IP carriers as well to
25 generate revenue by selling the services to their
168
1 corporate customers, where in this case, their
2 corporate customers again have agents either on
3 their corporate intranet or on the Internet
4 remotely and their services are deployed on the
5 IP's carrier network. It could be a cable
6 company. It could be a company that just has an
7 ATM data network or satellite network, any
8 physical network that supports IP. That's pretty
9 much the demonstration.
10 In conclusion, you can see that IP
11 telephony is more than just picking up a PC or a
12 telephone and carrying voice over the Internet or
13 over corporate data networks to somebody else on
14 the other end. There are value-added functions
15 just like in the regular PSDN which are being
16 deployed in the IP realm. Thank you very much.
17 (Applause.)
18 MS. BROWN: Thank you. We have about
19 one or two minutes. Does anybody have any
20 questions they need to ask before me move on?
21 That was neat. Anything? No. Thank you so
22 much, gentlemen, we are going to assemble the
23 next panel. Again, thank you.
24 While people are getting situated,
25 we'll just give them a second. Let me now
169
1 introduce to you Becky Burr, who was formerly at
2 the FTC and has been very much involved on
3 Internet side on the content side as well as on
4 the government side. She did an enormous amount
5 of work on privacy issues. She is involved in
6 the domain name registration issue. She also
7 traveled with Ira over the summer and Diane so
8 she has done that globetrotting trip with the
9 Administration's commerce policy. Let me
10 introduce Becky Burr.
2 MS. BURR: Thank you, Kathy. You could
3 tell I was the least smart member of the world
4 traveling team with Ira because I was the only
5 one who actually had to go actually all the way
6 around the world.
7 I think I probably have the easiest job
8 today. I get to introduce a panel to show us
9 sort of what's coming down the pike, and to spend
10 some time talking about what we should do or what
11 we shouldn't do to make this technology available
12 to us as quickly as possible.
13 We are going to start out with Joseph
14 Rinde, the director of switch network
15 architecture in MCI's data architecture
16 department. Mr. Rinde spends an enormous amount
17 of time working on, among other things, it
18 appears Web-based services and IP phone and fax
19 services. I hope we are going to get a good
20 sense from him, and maybe some pictures of what
21 they are thinking about.
22 Next we'll hear from Steven Marcus, who
23 is at AT&T Labs. Dr. Marcus is working on the
24 application of speech technology for Internet
25 services. Before coming to AT&T in 1986,
171
1 Dr. Marcus was the principal investigator in the
2 speech research group at the Institute of
3 Perception Research in the Netherlands. He may
4 have some theoretical.
5 Then we'll turn for the how is this
6 going to affect us view to Danny Weitzner, the
7 deputy director of the Center for Democracy and
8 Technology. Danny is probably familiar to most
9 of you. He focuses on the impact of new media on
10 law and Democratic values. I suspect he will
11 have a few things to say about what we should or
12 should not do.
13 Finally, we'll turn to Robert Pepper,
14 chief of the FCC's office of policy and
15 planning. Mr. Pepper has been in that position
16 for a long time and has had a tremendous impact
17 on telecom in the last decade. He had his roots
18 here at NTIA in the Annenberg program.
19 I'm going to turn the microphone over
20 and let's get started. I'll plug in to make sure I
22 don't go blank too many times. Thank you very
23 much. Let me start out by saying that the
24 technological implications of voice -- the future
25 of voice over IP is a new way to communicate.
172
1 We are talking about new ways to
2 communicate and not cheaper voice. Things
3 started out in that vein; that's where all the
4 hype started out. That's really what it's not
5 all about. We are talking about a way to
6 communicate that's as different from the
7 telephone today as telephone was from telegraph a
8 few years back.
9 We are not talking about all or
10 nothing. We are talking about future uses which
11 will include both the packet switching world and
12 the circuit switching world. Both are going to
13 be used to provide integrated services and the
14 user is going to benefit by receiving the
15 services he wants at the lowest possible price.
16 The question that keeps coming up is,
17 is this a single line or is this multiple lines?
18 The answer is sometimes. What we are really
19 dealing with here is provision of services
20 independent of the network.
21 What we need to do is provide the most
22 cost-effective service in any given situation.
23 Sometimes that will mean a single line. The case
24 of the road warrior in a hotel room, he only has
25 one line. That may be the case for a consumer.
173
1 On the other hand, it may mean multiple lines.
2 In any business you walk into, you have
3 a PBX and a LAN. Let's use our facilities in the
4 best possible way. The classic view of
5 telecommunications is computers on the Internet,
6 phones on the PSTN and there is an impenetrable
7 barrier between them. If we are to understand
8 how the future is going to look, we have to break
9 out of this kind of view.
10 The new view of telecommunications is
11 still computers connected to the Internet but
12 also connected to the telephone system. You can
13 buy yourself a modem and your multimedia medium
14 becomes a speakerphone. Telephones are getting
15 more sophisticated. We are looking at phones
16 that connect directly to the Internet. Such
17 phones already exist.
18 To this end, MCI made an announcement
19 earlier this year of a vault technology. This is
20 a marriage of packet switching and circuit
21 switching and all of that under a canopy of
22 intelligent network services to provide the end
23 user services. Looking at using whatever network
24 is appropriate at the time because users should
25 not be concerned with networks. The only thing
174
1 they should be concerned about are services.
2 That's what they need to focus on.
3 The way we propose doing this is to
4 take the circuit switch network and take the
5 packet switch network, gateways between them, of
6 course. We have had a lot of discussion of
7 gateways earlier today. We take intelligent
8 services and put that over everything signaling
9 to both networks. The thing that ties it all
10 together are profiles.
11 What does this customer subscribe for?
12 What kind of services does he want? How does he
13 want those services customized? Kinds of
14 applications that we can address now are command
15 and control, where you can use your Web to
16 compare your telephony services or vice versa.
17 Fax over the Internet, voice over the Internet or
18 the most interesting one of all, that is
19 collaborative multimedia.
20 There has been a lot of discussion here
21 today about cost savings, what's cheaper, what's
22 more expensive? Frankly a lot of comparisons
23 that have been made or presented have been very
24 confused.
25 What frequently happens is that the
175
1 price of PSTN gets compared to the cost of IP.
2 It's a lopsided view of the world and generates a
3 very warped understanding. Very intelligent
4 people are making incorrect conclusions.
5 For example, if you look at the recent
6 Forester report on the Internet voice coming of
7 age, they make the statement that telephony costs
8 $10 million and an IP hop is only half a million
9 dollars, so look at how much cheaper it is.
10 That's true. But how much capacity
11 does that voice hop have versus that data hop.
12 When you start doing those comparisons, you find
13 that the differences are really not that great.
14 There are, of course, differences in how things
15 are tariffed, how things are priced. PSTN is
16 sold by the minute, Internet sold by the month.
17 What does it cost you? Cost of a minute is low
18 if you compile a lot of minutes. There are of
19 course the regulatory issues that we talked about
20 earlier, accounting, access charges, accounting
21 rates.
22 Getting back to the price versus cost
23 comparison. That's something we have to be
24 careful of when we try to really understand
25 what's going on. These discrepancies are going
176
1 to normalize over time.
2 In terms of voice over IP, phone to
3 phone over the Internet is something that's not
4 been proved to be economically on a dial basis
5 domestically. That's where you do a second
6 dialtone, the Free World Dial Up, for example.
7 There was no benefit in calling New York to call
8 New Jersey. Even if you were New York and even
9 if you are calling San Francisco. It turns out
10 the economics aren't there.
11 A little over a year ago, a company
12 called LANIC made an announcement they were going
13 to offer domestic dial service over the
14 Internet. They were supposed to start service
15 September of last year. Between D.C. and San
16 Francisco, no services have started. Their
17 pricing that they had proposed was 5 cents to 8
18 cents a minute. 5 cents a minute, they would
19 have lost their shirts. It's only 2 cents less
20 than you pay for telephony today. Who is going
21 to go through a double dial and reduced quality
22 to save 2 cents a minute. Some people will, not
23 a lot.
24 Internationally, it's a different
25 story. Domestic long distance is already priced
177
1 low enough that this is simply not a competitive
2 offering. Internet telephony requires gateways.
3 We have also talked about. These are a major
4 cost source for that kind of service and we heard
5 from VocalTec earlier today how these are going
6 to come down. They have to come down a lot.
7 This is the thing that makes Internet telephony
8 on a par with PSTN in terms of actual costs,
9 darned near.
10 Internationally, we have the high
11 prices with the arbitrage opportunities the
12 greatest. Of course, the WTO agreement that's
13 been referenced several times today is going to
14 bring those prices down. It's going to minimize
15 that arbitrage opportunity. Who is going to use
16 international voice over IP? Bandwidth is
17 limited, so the delay is far greater. No only
18 are the distances greater, but the bandwidth is
19 not there. When you are talking about making
20 those calls you are going to have delays.
21 I don't know how many people noticed
22 the delays that took place during the demos just
23 now. We were getting during the call center
24 delay, a two and a half second delay roundtrip.
25 I wasn't using my watch, but I was counting. As
178
1 soon as I hear someone stop counting, I start
2 counting.
3 I know it's possible to get delays
4 down. Internationally, it's extremely
5 difficult. Alphanet was targeting that
6 particular market. They were supposed to have
7 started service late September, and they haven't
8 started service.
9 There are folks who tolerate delay.
10 There is the example of Delta III, a start-up
11 company in Israel that started up company between
12 Tel Aviv and the United States. People had lots
13 of people they had wanted to talk to in Russia.
14 This was a target market that was willing to
15 trade price, quality. They made a growing
16 business. They were competing with another
17 company that charges outrageous rates for long
18 distance.
19 Israel has opened their long distance
20 competition with two licensed carriers and rates
21 have come down. The arbitrage opportunity was
22 there and I'm sure they made a lot of money in
23 the interim. On the business side, most people
24 can't tolerate the delay.
25 It's bad enough on an one-to-one call,
179
1 again on a conference call with a two and a half
2 second delay. It just won't work. Who is going
3 to pay for additional bandwidth? Today, all the
4 bandwidth coming in is paid by the foreign ISPs.
5 That's not going to last forever. But for the
6 moment, we are in a prima donna position. We say
7 we have the Web sites. You want to get to us,
8 you pay for the access.
9 The point is someone has got to pay for
10 that international bandwidth. That's going to be
11 reflected in everybody's price because we don't
12 charge for specific usage on the Internet. I pay
13 the same amount whether I'm going to France or
14 New Jersey.
15 Intercorporate calling over intranets,
16 as was made reference to earlier, changes the
17 situation considerably. We are talking about
18 using infrastructures of LANs with WAN
19 connections. We can interface to existing PBXs.
20 Certainly the PBX manufacturers are coming out
21 with interfaces that will get you onto the
22 Internet. NEC recently made an announcement they
23 are going to use Vienna Systems. They make a
24 gateway box, a relatively inexpensive one. They
25 are going to provide that as an exterior box.
180
1 Other ones are talking about putting it under the
2 skins. This allows users to use regular
3 handsets. Very important. You don't want to
4 change user variance if you can avoid it. The
5 PSTY or IP is made dynamically by the PBX.
6 For intercorporate PC calling, that's
7 another possibility. One of the things that was
8 mentioned earlier was calling a PC. That's fine
9 if he is sitting on a corporate LAN. That's not
10 so fine if he is a dial-up user. It's very
11 expensive to stay online even though they are
12 selling unlimited usage. They are losing money
13 on you today. That won't last. Nobody wants to
14 blink first. NETCOM has already blinked. PCs
15 are always online, no extra charge.
16 Single call, whether a LAN or on a
17 dial-up is about 10 kilobits per second, if you
18 dot a T1 connection I can throw a voice call on
19 there at very marginal incremental costs. I can
20 throw on a gateway and I have got a very robust
21 capability. That's good for the business place.
22 At home, well, it's possible that cable or XDL
23 technology may bring this into the home. It
24 depends on how much it costs and how many people
25 are going to go for that.
181
1 Fax is something that's a big sleeper.
2 Not too many people talk about it. UU-Net made a
3 big announcement about that. Those require a
4 change in the way people behave, and that will in
5 fact limit how many people will actually do this,
6 but it certainly will attract a certain amount of
7 traffic. Realtime via the Internet it, was
8 referenced earlier today, it's not as easy as it
9 sounds.
10 Yes, it's true, it's not a person. If
11 you have a little delay, it doesn't hurt things
12 too much, except sometimes. If you generate too
13 much delay at the wrong times, you'll drop the
14 call. Store and forward can be used as a
15 personal mailbox which makes it a mail-me,
16 follow-me kind of service.
17 Let's look at the multimedia. I stole
18 this cartoon from the ITU Web site and the
19 caption on there reads, Now that we are all here,
20 we can get started.
21 Kind of applications, well, we
22 mentioned the road warrior, the guy with one
23 line. Think of the salesman, talking to his
24 customer. The customer wants to know how can he
25 get 1,000 widgets. The salesman needs access to
182
1 his database at the same communication. He needs
2 two lines but only has one. Internet telephony
3 comes in there.
4 I think it's even better for tech
5 support and for sales. Imagine if that tech
6 support guy could read your autoexec.bat, he
7 would find the problem a lot faster. It's the
8 killer application in the industrial world. I
9 call it a marriage of media.
10 Videophone, I have been hearing about
11 videophone since the 1950s. It's still not here
12 but it's close. Shared applications, whether
13 VocalTec or Microsoft, the key to all of this is
14 multiway conferences where I can bring in not two
15 parties but a whole raft of parties to have my
16 conference call on the Internet using all these
17 facilities. This is going to generate more
18 effective meetings and believe it or not,
19 generate more meetings.
20 This is going to generate a net
21 increase in the usage of communications and the
22 question that's open to this group is how will
23 voice be transported in this environment? Now,
24 there is no single answer because both circuit
25 and packet switching will be used. There is no
183
1 reason why a PC can't coordinate multimedia
2 physically as well as on your screen. Nowhere is
3 it written that voice has to pass over the
4 Internet, right.
5 You can see a PC with two sets of wires
6 coming out the back. One's an Ethernet
7 connection and one's a PSTN connection. There is
8 no reason why the voice can't go over that
9 high-quality low-latency connection. If the
10 voice can be passed over the Internet in today's
11 environment and current pricing cheaper, that's
12 fine, as long as you can achieve the quality
13 required.
14 I foresee applications where you will
15 be able to dial in the kind of quality of service
16 you require, no more than a half second delay,
17 whatever it is. The application will check, can
18 I deliver that kind of delay, characteristics
19 over the IP network and if I can, fine, if not,
20 I'll put it on the PSTN.
21 Whichever network I monitor to see if
22 the conditions have changed, maybe I couldn't do
23 it on IP to begin with. Maybe somebody has
24 stopped using the LAN and now I can move it over
25 transparently to the user. I'll hang up the
184
1 phone and put his voice on the IP network. So if
2 a PBX can make these kinds of decisions for us,
3 certainly a PC can do the same.
4 Ease of use is critical. If it's not
5 easy to use, people simply aren't going to use
6 it. What I think is critical for the future of
7 this is the ability to have impromptu meetings.
8 That is, no reservation required.
9 I get up and I knock on Fred's door and
10 I said Fred, you got a minute? Maybe he is
11 there, maybe he is not. I want to be able to
12 drop in on my colleague across the globe and
13 knock on his door for a virtual chat. My boss
14 said you can't do that. I said why not, he said
15 if you are on the other side of the globe, he is
16 asleep. You probably limit this to eight time
17 zones. You want to knock on his virtual door and
18 say Fred, can I talk to you, I don't understand
19 the spreadsheet. You got the shared application,
20 you can both be working on it at the same time.
21 You need a third colleague brought in.
22 You reach an impasse, call the one who knows.
23 You want to bring him in. That's why the
24 multiway conference is so important. Today we
25 are limited to a building, sometime to a floor.
185
1 If we can go across the globe and hit our offices
2 wherever they are, we are going to have more
3 informal meetings resulting in more efficient
4 communication within our organization and the
5 impact of all of this, I believe absolute usage
6 is going to increase. It's not a shift of
7 usage. We are not going from the circuit switch
8 network to the packet switch network.
9 We will have an absolute increase in
10 usage. Circuit switching will be continued to be
11 used very often on the fallback when the quality
12 of the IP isn't adequate. What we are talking
13 about is a new capability, not a replacement
14 capability. IP and conventional telephony will
15 live in harmony for years to come. The future of
16 voice over IP is not cheaper voice.
17 I complain to my vendors, gateways,
18 that their gateways are too expensive. They said
19 Joe, it's not a cheaper voice. And I said yes,
20 but it better not be more expensive. Right now
21 it is. So the future is going to be feature
22 function. Facilitation of the collaborative
23 computing to fit into that PC-centric workplace
24 that we are in right now. We are going to create
25 opportunities for those who are prepared to
186
1 exploit them.
2 Here we are on a high note. Before we
3 all get carried away with it, let me mention that
4 there are a few rocks on the road. The biggest
5 rock is delay. Again, in the corporate world,
6 dealing with things like conference calls, delay
7 is going to be a very serious issue.
8 Need for absolute reliability. We are
9 talking about voice now. This isn't your LAN
10 going down and you can't use your computer for
11 half an hour. This is your voice. If that's the
12 only place your voice is, you better be
13 reliable.
14 Got to make it easy to use. Got to
15 keep it in the same patterns that everybody is
16 used to today. The costs have got to be right.
17 In the area of delay, the first
18 question is who is going to tolerate it? We
19 talked about that a little bit. One-on-one
20 calls, much more tolerable than in a conference
21 call environment. Delay sources, well, they come
22 from all over. They come from the PCs. They
23 came from the gateway if you are using the
24 gateway. They come from digitizing time.
25 When you talk into one of these things
187
1 it's converting analog speech to digital. It
2 takes about 30 milliseconds of analog speech.
3 You got a 30 millisecond piece to send. You know
4 what, they don't accepted it. Almost every
5 application gets three of those, puts it in a
6 packet and sends 90 milliseconds of voice.
7 That's a 90-millisecond delay right there. At
8 the other end, you have got to have a buffer
9 because things get delayed and sometimes it takes
10 a little while because things get there. You
11 have got to start doing some kind of
12 interpretation and the speech doesn't sound
13 right. That's a lot of delay right there.
14 Now, you could reduce that delay if you
15 sent each 30 millisecond sample in its own
16 packet. I triple the number of packets I have to
17 send. That may work in the Internet
18 environment. In the public where we already have
19 so many packets flying around, we would exhaust
20 the packet switching capacity of the routers and
21 the backbone.
22 Then there are the congestion issues.
23 The edges of the public network tend to be the
24 sources of congestion. We have direct
25 connections to the backbones. The backbones, all
188
1 of them, UU-Net, Sprint, all of them have
2 relatively low delays.
3 The Internet service providers, that's
4 one of their biggest expenses. They tend to load
5 those lines up the most. This introduces
6 congestion. Also, when you go between networks,
7 especially if you are going from one of the NAPS,
8 as long as you are staying on the same backbone
9 you tend to do well. As soon as you start going
10 through gateways, you start running into those
11 congestion issues.
12 Well-managed internets? We tend not to
13 have those problems. Reliability. Voice is
14 mission-critical. It's bad enough when your
15 computer network is down. Who is going to take
16 this risk? Telecom managers, are they shy of
17 risk?
18 I remember a poster. It was a great
19 poster. A scene looking down the street a couple
20 stories down. You can see the window ledge
21 around the edges. You saw two hands up here and
22 two feet down here and the caption is we lost
23 Pittsburgh again.
24 Where is our telecom manager? How are
25 we going to minimize that risk during the
189
1 transition period? The question is ease of use.
2 You can't change how people behave, at least not
3 very quickly.
4 Voice over IP has to work on a
5 handset. If you put it as an integrated package
6 for multimedia collaborative computing, if you
7 got to dial, it's got to be the way you dial
8 today. You can all recall what happened with one
9 plus dialing. Suddenly is the aim to use MCI,
10 AT&T, how many people wanted to do it? My mother
11 won't do it, my wife won't do it, a lot of
12 people. Even 10 XXX is a market inhibitor. It
13 cuts down your market tremendously. Multimedia
14 has got to be easy to use. Delay is another
15 factor in that.
16 Let me just briefly mention costs.
17 The question is, is it really cheaper to do voice
18 over IP? The answer is, it is, but not by a
19 lot. If I can lock a penny or two off your calls
20 as a carrier, your actual cost, if I can reduce
21 it that much, is it going to matter much?
22 Fundamental switching costs are actually close.
23 Today, cost is almost on a par. Bandwidth use is
24 lower, not by as much as most people think.
25 The cost of the gateways eats up an
190
1 entire advantage today. For an incumbent
2 carrier, there is no great push for an immediate
3 switch. A new carrier may want to go that way.
4 In fact, will probably want to go that way, but
5 today has got to be very careful about the
6 quality of service he can deliver.
7 I have got a couple of generic Internet
8 considerations that go beyond this and --
9 MS. BURR: If we can just wrap it up.
10 I have got a few generic considerations.
11 MR. RINDE: RSVP was mentioned.
12 Reservation protocol. It's been pointed to as
13 the quality of service savior for Internet
14 telephony. There is no way to signal that.
15 There is a problem with settlements between the
16 networks. What's worse, Kennedy supported it
17 all, the public network.
18 MCI. MCI's backbone, we handle 4,000
19 flows per second. RSVP requires you to monitor
20 them all in order to deliver the RSVP capability.
21 That's not very practical.
22 There are some other considerations
23 about settlements. It turns out that in this
24 case the yellow cloud will carry most of the bit
25 miles because everything coming back from the
191
1 server where the volume is, is going to travel
2 long distance through the yellow cloud. How is
3 he going to be compensated appropriately for
4 carrying all that extra traffic.
5 Internationally, we this have problem in spades.
6 Then there is the issue of security and
7 privacy. You can put encryption but you can't
8 export it. There is directory services. With
9 that, I'll quit.
10 (Applause.) You introduced a number of
12 the points that I'll be speaking on. That will
13 help me to cut down on the length of my
14 presentation.
15 MR. RINDE: Glad to help.
16 DR. MARCUS: It has been interesting to
17 see how many points of convergence we have had.
18 I'd like to briefly introduce some things that
19 have really struck me about how much the Internet
20 and technology solutions that we are a part of
21 have been shrinking the globe.
22 If I have a touchtone phone in front of
23 me it takes about the same amount of time for the
24 message to get from my brain to my fingertip as
25 it takes to get from here to the other side of
192
1 the globe. Actually, a voice signal takes a few
2 tenths of a second. I remember in the early
3 '90s, there was a tongue-in-cheek prediction
4 that all calls will be a local call. In fact,
5 from my home in New Jersey today, I can make a
6 plain old telephone call to anywhere in New
7 Jersey for 8 cents and to anywhere in the United
8 Kingdom for 12 cents a minute. This doesn't take
9 into account we have local calls for free. Joe,
10 thanks for making that point for me.
11 Internet communications is about a lot
12 more than inexpensive phone calls. I'm going to
13 talk about the exciting vision we have or the
14 possibilities that the Internet is bringing us in
15 terms of new communications we'll be seeing over
16 the next five to ten years. As we look towards
17 2,000 and beyond, faster access technology,
18 improved voice coding algorithms or to send voice
19 using more bits for voice quality.
20 We send something that's almost close
21 to CD quality right now, certainly to FM
22 broadcast stereo quality at that bandwidth.
23 We'll be seeing a convergence of voice and data
24 networks. That's something about multimedia
25 communications. Joe has used a similar term.
193
1 The option of better voice quality and low-cost
2 basic telephony, not free but certainly low
3 cost.
4 What's new on the Internet? Speech
5 coding. We started in the 1960s developing that
6 and it's used in the POTS telephone network.
7 When you make an international call, the very
8 minimum is that you fit two phone calls onto one
9 phone line. There is more dense coding that is
10 used on top of that. The coding that's used and
11 developed on the telephone network have really
12 come from the same source.
13 The computer network protocols were
14 developed in the 1970s. Client/server
15 applications and graphical user interfaces both
16 appeared in the 1980s. What was new? Around
17 '92, '93, Netscape came along with a Web
18 browser. It was this user interface that brought
19 things together. Who remembers what the Internet
20 looked like before the Web browser? Here I
21 scratch my head and tried to work out.
22 Rather than click to go get a file,
23 this is about the first half of going to Apple's
24 Web site to retrieve a file.
25 Telephony before the Web:
194
1 OPERATOR: Hang on a moment while I try
2 to conference Don.
3 DR. MARCUS: That is familiar to some
4 people.
5 OPERATOR: You have deleted this
6 message.
7 DR. MARCUS: The world is getting more
8 complicated with as well as realtime
9 conversations, we are going to be having
10 multimedia interactions, sharing voice data and
11 graphics.
12 Are we going to be able to share each
13 other's applications remotely? We have messaging
14 and messaging will include not only voice but
15 various sorts of multimedia content. We will
16 have agents that we can interact with as well as
17 people. So well-designed applications and user
18 interfaces are going to be absolutely essential
19 here.
20 You can look at basic communications,
21 people communicating with people. How can we
22 solve this problem of trying to conference Don in
23 and not losing the call? What we have been doing
24 is looking at using the Web browser as an
25 interface for setting up just standard voice
195
1 teleconference calls.
2 No special software set up? You can
3 use a Web browser anywhere. The interface was
4 designed in AT&T research, and we did a small
5 technical trial. You had a much richer set of
6 features than a touchtone interface. Today we
7 use conventional phone lines because in the
8 business environment, people aren't using their
9 main phone line, at least not in their desk as a
10 way of getting data.
11 They got a LAN. They get a second
12 phone line with a minimum on it. Here's the page
13 where I can go in and I can either select one of
14 the regular conferences I have got set up or I
15 can go to my list of frequently called people and
16 click on one of their boxes to check them off or
17 click on them to go and find alternate locations
18 that I have stored for them like home or office.
19 When I'm actually in the call, I then
20 have a status display that shows me each of these
21 people, allows me to add extra people and extra
22 phone numbers and shows the status of each
23 person. So the host is connected. Greg Blonder
24 can be dialed by pressing this button. Doris
25 Miller has been muted. We are able to mute our
196
1 own people, hang up specific people without
2 hanging up the whole conference call. We have
3 been doing some studies on this and we found that
4 people are really enjoying using this interface.
5 There is an extra participant at the
6 bottom here called Conference Scribe. This will
7 allow you to record conference calls and for
8 people to go in and access archives of conference
9 calls later on, or to join the call later and
10 catch up on what's been happening. This brings
11 me to the other class of applications which are
12 people interacting with machines and I think we
13 are not really talking about these sorts of
14 machines. There will be a question at the end if
15 anybody can tell the connection between this
16 slide and the next one. We are talking about
17 this sort of connecting with machines. My main
18 topic is speech recognition research.
19 All I'm going to do is play this clip
20 from Babylon 5:
21 MAN: Any messages for me?
22 OPERATOR: One from Maintenance, one
23 from Jeffrey Sinclair.
24 MAN: Top. Display one from Sinclair.
25 OPERATOR: Unable to comply. Use
197
1 password for access.
2 DR. MARCUS: This is 2159. I believe
3 in the next ten years -- if you come to our lab
4 in the next five to ten years, we are going to
5 see you able to interact with your communications
6 device which is some combination of telephone,
7 videophone, Web TV, entertainment and educational
8 source in the household in this way.
9 Moving on to other interactions of
10 people with machines over messaging. Searching
11 and browsing stored multimedia data is going to
12 be a major topic, and what do I mean by
13 multimedia? Data multimedia archives. These are
14 sequences of image, text, speech, music and sound
15 and it's correlated in time-align. Some of this
16 might be tightly time-aligned like sound tracks
17 on films. Some of it might be loosely aligned.
18 For example, some are news clips or if you have
19 got a conference call. Somebody who may have
20 typed in various points along the call.
21 Bigger, better, and faster isn't the
22 answer. We need to use the technologies we have
23 in video processing, text processing, and audio
24 and speech processing, for example with imaging
25 to identify scenes, people, places in text to
198
1 take actual transcripts, also closed captions on
2 TV. These are approximations to what's actually
3 been said by people that have been typed in as
4 fast as people can type in and fit that much
5 information on the screen and I think you are
6 doing a wonderful job there, by the way.
7 In the case of speech, we may want to
8 recognize exactly what's been said or to identify
9 the general topic so that we can go in and index
10 speech and find out where we are talking about a
11 particular subject. We may want to identify a
12 particular speaker with identifying tunes,
13 composers, the mood of the piece, and to answer
14 questions like did the dog bark in the night.
15 There is a system we developed in our
16 speech and image processing services lab which we
17 have termed pictorial transcripts. What this
18 does is to generate condensed multimedia
19 representations of film, television, radio
20 programs. I'll be showing you a short videotape
21 which shows this actually in action.
22 What we are doing with video is
23 something called content-based sampling where we
24 segment the video into sequences of frames that
25 are closely related. You don't have to send 30
199
1 frames a second or even one frame a second. You
2 send information when the information is
3 changed. The closed caption text in the TV is
4 all in upper case. We do a lot of processing on
5 that to restore upper and lower case. Also at
6 the same time, as an essential part of that
7 process to identify key words and topics. When
8 you put all of this together it can actually be
9 transmitted at modem rates.
10 The content-based sampling. Here is an
11 example of how that works. This will actually
12 show you the video. You'll see each of these
13 frames comes out of that.
14 How far do you have to go for great
15 tasting water? As far as your faucet. The BRITA
16 water filtration pitcher has the special BRITA
17 filter that reduces chlorine, lead, the things
18 nature never intended. And the taste --
19 wonderful. BRITA, tap into great taste.
20 Available at Pathmark and Shop-Rite.
21 The best thing right now would be to
22 run the videotape. We can put together a
23 storyboard like this with sequences of images,
24 the closed caption text. Maybe that wasn't the
25 best thing to do. There is an explosion of
200
1 multimedia information available in the form of
2 television and video programs. Pictorial
3 transcripts makes this data accessible over
4 communications networks by applying techniques
5 that automatically analyze, condense, and index
6 multimedia information. Content-based sampling
7 selects representative frames from video
8 programs.
9 Closed caption information undergoes
10 lexical and linguistic processing to provide the
11 textual component of the transcripts. The
12 resulting is to create a hypermedia transcript of
13 the program. It can be accessed within seconds
14 of a program's conclusion. This is how the final
15 transcript looks to the end user. An index page
16 presents a textual and pictorial table of
17 contents for the user. Selecting a word or image
18 on this page will take the user to the
19 corresponding point in the program. Key words
20 and phrases are automatically linked to
21 additional information about that topic. Audio
22 information is also available.
23 Now, the timber industry begins to
24 change the way it does business. This compact
25 rendition of the program can be transmitted over
201
1 phone lines. Pictorial transcripts utilize many
2 technologies available within AT&T Labs' research
3 and combines them into a fully automated system
4 that delivers a level of speed and convenience
5 never before available for indexing and accessing
6 the vast informational resources generated by
7 television.
8 I have touched briefly on a number of
9 topics that we are working on in AT&T labs
10 research. If it seems like I have come a long
11 way from telephony and a conditional phone call,
12 that's because I believe that's where the
13 Internet is going. We didn't synchronize our
14 talks but Joe really made the same point as I.
15 In time, when we speak about communications, it
16 will be nothing like what we know as a phone call
17 today. It's Internet telephony that's providing
18 enough of the infrastructure for that. User
19 interface design is going to be actually crucial
20 in providing services that are easy to use, that
21 people are willing to use and that people are
22 actually excited about using. Thank you. Good afternoon,
24 everyone. I'm going to have to speak from here
25 on the theory that I might be faster. But let me
202
1 first thank the NTIA for this forum. We
2 obviously hear an awful lot about Internet
3 telephony in our respective work in these related
4 fields and it's actually nice to spend all day
5 really focusing on it. So thanks to Larry,
6 wherever you are, there you are, and everyone
7 else for organizing this.
8 I want to just bring some very brief
9 reactions from the user perspective. I have to
10 say that I'm not the happiest of users today.
11 That's because on Sunday night, over Labor Day
12 weekend, my office's Internet connection
13 mysteriously evaporated. We were actually in
14 sort of a billing confusion with our service
15 provider and had thought they had terminated the
16 connection. Although they promised us they
17 wouldn't, the fact is they hadn't.
18 As it happens, the local exchange
19 carrier, who will remain nameless, terminated the
20 connection incorrectly. They had a termination
21 for someone's relay circuit and transposed the
22 address of that circuit with our circuit and we
23 ended up on Monday morning with no service. I do
24 have to say that if we had anywhere else to go to
25 get frame relay service, I would have gone there
203
1 on Tuesday morning. We don't, so here we are.
2 So I'm less than happy a user today. It does to
3 some extent color my views.
4 As bad as that experience was for my
5 organization which depends quite heavily on the
6 Internet, it would have been all that much worse
7 had that also been our telephone line, our voice
8 telephone line. We couldn't even have called our
9 service provider to complain to them. I don't
10 know what we would have done, sent a carrier
11 pigeon?
12 It does, I think underscore a lot of
13 the concerns that people have been talking about
14 today, that these are very real issues. If this
15 sort of thing happened in the middle of a
16 legislative debate that was going on or a
17 grassroots campaign that we were hot and heavy in
18 the middle of that we do over the Internet, I
19 probably would have suspected foul play on the
20 part of some government agency or another, but it
21 doesn't even take that. It just takes these
22 sorts of things which do happen. So I think that
23 the reliability issues and the competition issues
24 really are just critical from a user
25 perspective.
204
1 I have to say that I think that a lot
2 of the questions surrounding Internet telephony,
3 the questioning surrounding the transition of our
4 telecommunications regulatory environment from
5 where it was and is to where it's going,
6 whatever, wherever it's going between these two
7 presentations are staggering questions, as you
8 know, and I don't pretend to have particularly
9 good or specific answers about where they ought
10 to go. But I do think that we have in the last
11 couple of years learned some things about how to
12 treat the Internet as a matter of public policy.
13 A number of the people in this room have I think
14 contributed to that thinking.
15 I want to just begin with one, I think
16 a relatively simple idea that actually emerged
17 out of the debate out of the Communications
18 Decency Act and ultimately out of the Supreme
19 Court's decision on the constitutionality of the
20 Act.
21 I think the Supreme Court, what the
22 White House has just said is that we have to
23 really focus first and foremost on the primacy of
24 users and on giving users control over their
25 experience of this new media.
205
1 Efforts such as the Communications
2 Decency Act such as encryption regulation that
3 I'm not going to go into because this part of the
4 Commerce Department doesn't handle it, I think
5 are real examples of what not to do when trying
6 to understand how to treat the Internet.
7 I think that we really have to keep
8 very focused on the fact that what I think for
9 many of us are really truisms today, that the
10 Internet is a global medium and that it's a
11 decentralized medium and that we cannot apply
12 traditional top-down regulatory models to
13 Internet issues. It just simply will not work.
14 It also leads us into huge policy conflicts which
15 lead us to the Supreme Court and hopefully not to
16 the Federal Communications Commission, but that's
17 going to happen, as I'll mention as well. That's
18 I think what we are all trying to avoid so that
19 all of these innovations have a chance to get out
20 there and actually be tested with those of us who
21 are consumers and users, at least potential
22 consumers and users.
23 I want to make just two very, very
24 simple points along the line of how to give users
25 more control, how to respect the control that
206
1 users ought to have. First of all, in the whole
2 discussion that has been going on beginning with
3 the ACTA Petition, I think that from a consumer
4 standpoint, the notion of a price discriminatory
5 policy that distinguishes the way I as a user of
6 telecommunications services use my bits and use
7 my package distinguishes between whether I happen
8 to be using my packets for Internet telephony or
9 for accessing a Web site or for sending E-mail or
10 for making a voice telephone call simply makes no
11 sense. I think that there are clearly very
12 substantial issues involved in the future of
13 universal service, involving who pays for the
14 cost of all the various network services that we
15 are worried about.
16 I very strongly believe that price
17 discrimination based on what an individual user
18 chooses to do with a particular packet is
19 absolutely the wrong road. It is an effort or a
20 suggestion that regulators can control what goes
21 on, on a user's personal computer, on a user's
22 information appliance, whatever it is. I think
23 it's a huge mistake and it is I think, it goes
24 against the lessons that I think we are learning
25 from the CDA debate from other Internet policy
207
1 debates.
2 I really strongly believe as a matter
3 of sensible policy, and I think also as a matter
4 of individual user sovereignty there has got to
5 be a limit somewhere about how far up the
6 protocol stack, if you will, regulators think
7 they can and should go. Once you get to the
8 application level of the stack, I think that is
9 the business of the consumer, not of the
10 regulator.
11 The second point I want to make
12 involves some recent and to most people
13 relatively arcane developments in the
14 implementation of the digital telephony bill
15 which is legislation, as most of you know, that
16 was designed to assure that law enforcement had
17 status quo, electronic surveillance capabilities
18 as the public switch network transformed, became
19 a more digitized network, and also became a more
20 competitive environment.
21 In the process of implementing this
22 statute, there is a relatively Byzantine standard
23 process that is underway in the private sector in
24 which the private sector, the telecommunications
25 carriers and telecommunications equipment
208
1 providers attempt to make collective decisions
2 about how they will comply with the statute.
3 That process has input and participation from the
4 FBI and other law enforcement agencies, the goal
5 of which is, is to yield a standard which vendors
6 and equipment manufacturers and service providers
7 can comply with in order to say that they are in
8 compliance with the statute.
9 One of the interesting developments in
10 this standard is, has to do with the treatment of
11 various surveillance requirements on packet
12 networks, whether they be ATM networks or native
13 TCP/IP networks provided by public network
14 service providers by telecommunications
15 carriers.
16 The FBI has indicated a concern about
17 continuing to be able to do what they call pen
18 register requests for trap and trace requests.
19 That is the ability to conduct surveillance which
20 gives them the phone number dialed or the phone
21 numbers which are making contact with a
22 particular target under surveillance.
23 In the packet network context, that
24 would be providing the FBI access to header
25 information on packets, whether they be TCP/IP
209
1 packets or ATM cells or whatever they are. The
2 FBI has interestingly enough decided that it's
3 not possible to give law enforcement when they
4 have a legitimate surveillance request like this
5 only the addressing information. That is, only
6 the equivalent of what would be the phone number,
7 the address in the header. They have asked to be
8 given the entire packet, both the signaling
9 information and the content.
10 This doesn't necessarily sound so bad,
11 but what it means is a substantial degradation of
12 individual user privacy any time a user happens
13 to use, whether they know it or not, a packet
14 network instead of a circuit switch network.
15 This is an issue that has just come before the
16 FCC. Congratulations. And it's an issue that I
17 expect will receive quite a bit of discussion.
18 But I think it's another example of essentially
19 trying to fit a square peg in the round hole as
20 we move from the communications, the
21 telecommunications environment that we were used
22 to operating in, into a new environment where we
23 expect consumers to use all these new services to
24 experiment with them, to develop them into new
25 markets but yet we seem to have some proposals
210
1 before us which penalize users either by charging
2 them more for using these same services or by
3 severely compromising their privacy when they use
4 these services.
5 My plea here is to think carefully
6 about the impact on the ultimate users of these
7 technologies. I think they are all very
8 exciting.
9 I will admit that I'm not a regular
10 Internet telephony user. I would be happy if it
11 made sense for me to be one day. I'm looking
12 forward to working this all out. Thanks very
13 much.
14 (Applause.) Thanks for that new
16 assignment. I won't be talking about that. This
17 panel is entitled Future Directions. I think the
18 only thing that we know for sure is that there is
19 going to be a lot of change in development.
20 Companies, including MCI, AT&T, Lucent, Motorola
21 and others are making investments in new
22 technology. That means that what we have on the
23 Net in 12 month, 18 months, two years is not
24 going to look like what people are doing today.
25 We already saw maybe a glimpse into that.
211
1 But think back two years ago, three
2 years ago, four years ago into the Net. There
3 currently are 4,100 ISPs in the country. The
4 average ISP employs 16 people, multiply that out,
5 that's something like 65,000 people work at
6 ISPs. This is an incredible industry. That is a
7 lot of jobs, a lot of payroll checks. There are
8 a lot of people betting that this thing is going
9 to grow even larger.
10 The early consolidation, demise, of all
11 the projections of consultancies. It's always
12 fun to poke fun at the consultant groups who
13 claimed maybe we wouldn't have maybe five, six
14 ISPs at the end of this year, well, sorry about
15 that. So as we look forward to the next couple
16 of years, I think we can expect things to change,
17 to continue to change, and to grow but in ways
18 that we can't really predict but I subscribe to
19 the school that chaos is good, not bad. In fact,
20 that kind of change leads to a lot of
21 creativity.
22 Internet telephony is really on the
23 edge of this sort of shift in paradigms. If you
24 think about it, you know, today we have a voice
25 network that's struggling to carry data in the
212
1 future. We are going to have a data network that
2 easily can carry voice plus lots of other stuff
3 that we were hearing about. Voice is becoming an
4 application, not a distinct network. We have to
5 stop thinking about it as a distinct network.
6 In fact, you know, I was thinking about
7 the name of the conference the entire day,
8 Internet telephony, and yet most of the day was
9 talking about voice on IP. In fact, Internet
10 telephony is not just telephony if you think
11 about it.
12 What do you do on the telephone today?
13 Just think. It's not necessarily realtime
14 voice. How many people here today made telephone
15 calls where they left voicemail? Of the people
16 that have made calls, how many people when they
17 got a break went to their phone and retrieved
18 voicemail? Right? That's all storing forward.
19 There is no reason why you need
20 realtime circuit switched no-latency voice
21 networks for that. In fact, it's a stored
22 forward medium. If you think about what people
23 use the long distance network for today and do
24 numbers, probably about half the revenue, maybe
25 40 percent of the revenue is generated by
213
1 services that are not realtime voice. So you
2 begin thinking about it in that perspective.
3 You begin to think about applications
4 over the Internet that in fact in some respects
5 are substitutes but will go beyond just being a
6 substitute from a people who use the traditional
7 telephone network for today. That's by the way
8 really the low-hanging fruit of the way the
9 Internet can provide very important -- we have to
10 come back to this -- a competitive alternative to
11 the traditional telephone networks, and it's that
12 competition that's very important because those
13 are all services that don't require again
14 realtime voice.
15 I think the notion that this is
16 Internet telephony and having a discussion about
17 that is absolutely the right perspective, but
18 let's not mistake that for being Internet voice
19 telephony.
20 Now, what this also means is things are
21 growing and changing. The regulatory system, as
22 well as the industry organization and the
23 business models that were built around networks
24 optimized for voice that those are going to lead
25 to change as well. So it's the regulatory
214
1 structures that are going to be under pressure as
2 the business cases, business plans and industrial
3 organizations change as a result of the shift in
4 the uses and applications, some of which we heard
5 about earlier. With high bandwidth networks,
6 voice may be, may be able to ride, even realtime
7 voice, especially if it's not realtime voice, we
8 may be able to ride very, very easily over these
9 high bandwidth networks that are put out there
10 for other kinds of applications. Somebody told
11 us recently. There is somebody here from Cisco.
12 Something like 40 percent of the service calls
13 are on the network and not necessarily through an
14 800 number call center. I don't need to make a
15 realtime voice call for that.
16 What's interesting is that, mentioning
17 earlier, intranets, if you really want to have
18 control of your quality of service and the
19 business applications of Internet telephony, it's
20 going to be over very high-quality, very closely
21 managed intranets where you control things like
22 the congestion, where you control the quality of
23 service.
24 So this new world that we are talking
25 about is going to require new ways of thinking
215
1 and it's also going to mean that we should not be
2 imposing or just taking legacy rules and imposing
3 them on new things. I think it was Congress,
4 among the really good things they did in the '96
5 telecom Act, one of the wisest things was they
6 said, wireless is treated different than wire
7 because in fact of its very different
8 characteristics in terms of the underlying
9 economics. You want the regulations and the
10 rules of technologies to be competitively neutral
11 but that also means when there are fundamentally
12 different underlying economics of technologies,
13 you don't want to treat them the same.
14 That leads us to the question about the
15 legacy rules that may not fit in this world of
16 packet networks versus the traditional telephone
17 or even broadcasting world. You know, the
18 broadcasting world is a world in which you have a
19 one to many paradigm. The constraint on
20 competition and the constraint on doing things,
21 the scarcity of spectrum where the lack of enough
22 spectrum being out there, is the scarcity real or
23 artificial, but the point is that not everybody
24 can be a broadcaster.
25 Because it's a one-way medium from a
216
1 consumer user perspective, there is very little
2 control by the user, very little choice. It's
3 one way. Somebody else decides how you see
4 things and when you see it. The telephone world
5 by contrast is one to one. The constraint is the
6 interconnection. The capacity through that
7 circuit switch, how many circuits can we
8 complete, interconnecting those communications.
9 It's very user initiated and controlled. I
10 decide who I want to talk to, when I want to talk
11 to that person, and we can control that. The
12 Internet is different. The Internet is both.
13 It's any to any, the constraint if any is in the
14 statistical multiplexing, the speed and it is
15 both user controlled and provider controlled.
16 It's both push and pull so it's going to change
17 the way we need to think about it.
18 Just picking up and taking legacy rules
19 and plopping them down on top of this new medium
20 is not going to work.
21 I have the red light so let me just say
22 a couple of very quick things to wrap up. We
23 have already heard about expectations. If you
24 are going to be using the Internet for
25 mission-critical communications, it has to be
217
1 reliable for today's communications created by
2 today's telephone network. It works and we all
3 really love the fact that you pick up the phone
4 and it works. It's always there. So reliability
5 and mission critical, quality of service
6 standards I think are just going to be demanded
7 in the marketplace.
8 The real questions here I think are how
9 do you increase bandwidth, how do you increase
10 packet networks. How do you create incentives
11 for new entrants to make those investments, and
12 as I have already said, we see the Internet not
13 just as new applications and uses but also as a
14 very significant new competitive force across the
15 telecommunications transmission markets,
16 transmission services and technology and it's not
17 just in the telephony aspect of it but the
18 Internet provides enormous potential for
19 competition in essentially every segment of the
20 incumbent telecom industries. I'll stop there.
21 (Applause.) Well, we have either just
23 solved or just uncovered all of the problems of
24 Internet telephony. I think we'll move quickly
25 into questions because I imagine that there are
218
1 quite a few. Any questions from the audience?
2 AUDIENCE MEMBER: If we can put on our
3 universal service hats for a moment, Section 254
4 of the Act sets forth formal criteria for
5 defining a universal service, such things as
6 being applied over a network, essential to the
7 public health, safety. And this is the one I
8 want to get to, something that's demanded by the
9 substantial majority of residential users.
10 Looking down the road, say five years
11 ahead, what would be your vote for Internet
12 communications, a specific application that you
13 think would fill that bill? I'm just sort of --
14 to exhort you on, you can be remembered as a
15 great seer or a great prophet. This is your
16 chance to go down in history.
17 MS. BURR: Anybody like to make some
18 predictions here?
19 MR. RINDE: I really don't want to
20 comment on that only because I think over the
21 next several years this is mainly going to be in
22 the commercial environment and not in the
23 consumer environment. I don't see any killer
24 applications in that area.
25 MR. WEITZNER: I guess that I think
219
1 that the most important thing for universal
2 service, for the future of universal service is
3 to continue to provide access to higher quality
4 digital bandwidth. I actually am relatively
5 confident that at least in the residential
6 market, and I put aside public institutions like
7 schools and the Snowe-Rockefeller side of the
8 world, but as to the individual residential
9 consumer market, I think that the Internet has
10 shown an extraordinary capacity for figuring out
11 on its own more or less what applications people
12 need. The critical thing is for people to be on
13 the Net so they can be part of that process of
14 figuring out. I think it's a mistake.
15 If you asked that question five years
16 ago, it's pretty clear that the answer would not
17 have been the World Wide Web, and there are a lot
18 of other things it would not have been. So I
19 think we should just accept that we are not going
20 to be able to be profits of that sort. I do
21 think that what we have learned is that open
22 digital infrastructure is very important because
23 it gives people access to all the different
24 applications that the Internet has, and will
25 have.
220
1 DR. MARCUS: I don't think it takes any
2 prophet to see that today we have the Web TV box
3 for $300, not much more than a mid range VCR and
4 what's going to drive down what's going to be in
5 that box that's going to be more than the
6 multimedia PC today that you are going to be able
7 to buy for what people are willing to spend on
8 entertainment devices. The way that people use
9 that is something we will see.
10 AUDIENCE MEMBER: Would you say the
11 cost and ease of use would be major drivers?
12 DR. MARCUS: I already said ease of use
13 is a major driver. In order to get these devices
14 anywhere, cost is always major. Probably as this
15 device takes on, is multifunctional and included
16 entertainment in one of its functions, then
17 people will be willing to buy these devices.
18 AUDIENCE MEMBER: Thank you.
19 MS. BURR: Any other questions? I
20 think you all must be just stunned or in that
21 after-lunch mode. We are going to take a break
22 now. I'm happy to tell you that VON is again
23 hosting our break so there is cookies, soda,
24 coffee, et cetera, in there and we'll be back in
25 15 minutes.
221
1 (Recess.) I'm always impressed and
3 grateful to those of you who stick it out to the
4 last. Thank you. The last session of the day we
5 hope to have it be a little more informal and
6 allow you to speak and not just us to speak.
7 The concept here is really a sort of a
8 reporting back on what we heard today and then
9 some comment from folks from various parts of the
10 industry with us the NTIA staff, particularly
11 represented by Larry there, really listening to
12 all of you and trying to get a full feel for what
13 it was we heard and perhaps what it is we should
14 take away.
15 I first met Michael Spencer about a
16 year and a half ago, two years ago, I guess, in a
17 teeny little room in New York when the very
18 infant VON Coalition was just coming together.
19 They were talking about just the beginnings of
20 these issues that we heard about today. I was so
21 impressed with Michael because he was able to
22 take a room full of Joe Rindes, people very
23 excited about all different parts of this issue
24 and how to deal with it and he was able to
25 conceptualize it sort of into a map forward.
222
1 I asked Michael if he could do that for
2 us today, to listen to what was going on today,
3 to think about what he heard and then to feed it
4 back to folks who are on this roundtable and
5 you. We hope we will wrap up. We will publish
6 the proceedings of today. This last session can
7 be like a conclusion to the day. So with that,
8 I'll turn it over to Michael, and then at the
9 close, I'll just give you some administrative
10 facts and then we'll wrap up the day with Larry.
11 So for now, Michael, go.
223
1 ROUND TABLE DISCUSSION
2 MR. SPENCER: Thanks, Kathy. Well,
3 those of us presenting are even happier that you
4 stuck it out to the end of the day.
5 I'm joined on this panel by five other
6 stalwart presenters who I would like to
7 introduce. On my left, Mark Coblitz, VP for
8 Comcast Corporation, John Curran, GTE
9 Internetworking Services, Frank Gumper, who is
10 vice president for long-range public policy at
11 the now Bell Atlantic combined with NYNEX. David
12 LaPier, who is product manager at Cisco,
13 responsible for new access network service
14 providers, Robert McDowell, who is the deputy
15 general counsel to the America's Carrier
16 Telecommunications Association, ACTA. We have
17 finally, Larry Irving, who you all know.
18 I'm going to try and do something
19 infinitely more risky than Jeff Pulver did with
20 his live demonstration. I'm going to try a live
21 presentation, one that I have written over the
22 last hour or so having listened to the talks this
23 afternoon so it could quite possibly crash. It
24 may make no sense whatsoever.
25 I would encourage you to throw your
224
1 questions up immediately and make this whole
2 session fairly interactive. You are all no doubt
3 dying to exercise those vocal chords by now. So
4 the first technical hole is the presentation.
5 There we go.
6 The first point that I heard, and I was
7 really quite surprised by the degree of unanimity
8 on this point which certainly did not exist in
9 the little room that Kathy describes a year and a
10 half ago in the Grand Hyatt, but really nobody
11 who stood up today said anything other than
12 Internet telephony is part at least of a
13 radically redesigned communications platform or
14 architecture.
15 It's quite impressive that in the
16 course of a year that belief has entered our
17 industry, but I think it's true. I have tried to
18 characterize that different architecture with
19 some very crude block diagrams.
20 On the left what I have called the
21 traditional telephone network, it's basically
22 made up of three blocks, not three actual boxes
23 in the network, but three kinds of things, Telco
24 infrastructure, copper, fiber, et cetera of
25 course, that's spread out, is the whole voice
225
1 switching network. Sitting on top of that are
2 phones.
3 On the right-hand side we have a crude
4 block diagram representation of this new Internet
5 architecture which takes a broader range of
6 infrastructure potentially, wired, wireless,
7 satellite perhaps. Certainly we hope it will be
8 broader bandwidth than the voice grade
9 infrastructure on the left, applies an IP data
10 communications wire on top of that, communicates
11 with multiple devices. It can be my PDA, phone.
12 It could be a telephone, computer.
13 Layered on top of these applications,
14 these services that we have become so fond of
15 pursuing, voice, satellite, fax, Web. Going hand
16 in hand with those two different architectures,
17 traditional voice and the Internet architecture
18 are a set of beliefs held by the proponents of
19 those architectures. On the left-hand side in
20 the phone architecture computers, if they are
21 discussed at all, belong on the Internet.
22 Data is the exception. Data is a
23 special kind of voice. On the right-hand side,
24 we want to provide services that are independent
25 of the network, the way we access the network and
226
1 independent of the devices that we use to get on
2 to the network and we believe that computers
3 ought to interoperate with telephones.
4 Interestingly, across both sets, there
5 is a broadly consistent belief about what the
6 Holy Grail is.
7 In each case, we wanted to drive
8 towards mixed media, collaboration, in whatever
9 flavor, so this last point which is where we all
10 want to drive to is not a difference in belief to
11 architectures. Instead we need to ask which of
12 these architectures is likely to get us to that
13 point. Realtime telephony is real now. We can
14 handle phone to phone with good quality and
15 sometimes acceptable quality on public networks.
16 A point Joe reinforced and others have
17 made, the cost of the PSTN is not great.
18 Practical applications particularly for phone to
19 phone include low-cost international calling
20 where the price of that service due, to the
21 artificial settlement rates, is much greater than
22 the actual cost. And now the kind of application
23 today where there is high value, so call
24 centers. Not economic for U.S. domestic
25 application.
227
1 Now, reflecting those beliefs, major
2 players, including incumbents are at least taking
3 positions at the trial level. Another important
4 learning over the last few months, this isn't,
5 although it has some of the character of a
6 religious war, one versus the other, it isn't in
7 fact an either/or. This isn't independent of the
8 PSTN.
9 Several people including Larry Fromm
10 from Dialogic this morning pointed out many
11 useful applications. We heard the widespread
12 belief, this isn't simply an arbitrage game. We
13 would have an Internet telephony market. A
14 question or just a scratch? The view was, and
15 this again, I was surprised how widespread we
16 heard these sort of opinions.
17 A dynamic view of the Internet
18 architecture, so look at the Internet
19 architecture less at what it is and more of the
20 dynamics of it. So it's technology
21 characteristics made up of standard components.
22 Independent service, meaning I can take any
23 service and support it with an IP network.
24 This leads us to a set of benefits.
25 For example, Moore's law, innovation
228
1 improvements. If we would have started the clock
2 of Moore's law and that has been something like a
3 doubling of price performance every 18 months or
4 two years. We had started that clock ticking and
5 applied it to bandwidth to the home.
6 We can take that innovation and do many
7 things with it. If we had sent bandwidth to the
8 home increasing at Moore's law rates, we would
9 for the price of our telephone access be sitting
10 on something like a fractional T3 circuit. Who
11 has the fractional T3 circuit at their home.
12 Mark points out that actually I do.
13 MR. COBLITZ: $39.95 a month.
14 MR. SPENCER: Here is a big problem.
15 You had to design up front what mode of media it
16 was designed to carry. A lot of work was spent
17 optimizing telephone switches and their supposed
18 replacements to carry particular kinds of
19 traffic. The Internet arguably doesn't carry any
20 traffic particularly well, but it does an
21 acceptable job of any kind of media and it is
22 proved to be acceptably acceptable. The kind of
23 standardized components and distributed
24 intelligence that support this architecture work
25 very much better at low scale than traditional
229
1 telecommunications plant, enables many, many new
2 entrants to get in, and we have had an explosion
3 of players in the telecommunications markets
4 providing you include the ISPs and Internet
5 customers. Customized services explained by the
6 capability of the PSTN, call waiting, call
7 forwarding. Not very exciting services. Useful,
8 but not thrilling.
9 My idea of a customized service may be
10 very different from Mark's and very different
11 from John's and I have a very realistic way
12 because I can take my computer, PDA and interface
13 with control software and dial Joe. That might
14 be valuable to me and Mark would say I don't want
15 to talk to Joe and I certainly don't want to do
16 it that way and he may very well have a very
17 different set of customized services.
18 You can do this in the Internet arena.
19 It just proves to be administratively and
20 logistically impossible in the PSTN arena. For
21 industry structure, it says well, if barrier to
22 entry is much lower we would expect to see lots
23 of new entrants. There were 5,000 ISPs in the
24 last month from Austria talking about
25 consolidation to ISPs, but in terms of actual
230
1 aggregate numbers of ISPs, this hasn't happened.
2 Barriers to entry to succeed as an ISP
3 may be different but it's a superficial
4 explanation. Lots, lots more innovation. If I
5 can take this whole communications problem up and
6 divide it into a set of layers or modules, I can
7 get to work on that problem with myself or ten of
8 my closest software engineering graduate friends
9 and I can realistically make an impact.
10 A good example of that would be a
11 company, IONA, which few of you will have heard
12 of which I suppose is an eight-year-old company
13 started up by college graduates who now have, by
14 many people's reckoning, the best implementation
15 of a piece of middle-ware which is important.
16 For IONA, it's made a bunch of people real rich.
17 For customers, I can look to lower
18 lifecycle costs. Why? If I'm using standardized
19 components, I don't have to worry that every time
20 I upgrade a service I can throw away my phone. I
21 can progressively update this thing and expect to
22 see modular replacement rather than catastrophic
23 replacement, and high flexibility. I can have my
24 services customized to my mode of operations.
25 What did we hear about the future,
231
1 richer services. Lots of hurdles on the way to
2 those rich services and in particular intranets.
3 Primary benefit, not low cost voice. If low cost
4 voice is what you want, get rid of settlement
5 rates, re-jig access charges, get some additional
6 competition at the PSTN. Instead, richer
7 services. The architecture can do very much
8 more, will do very much more.
9 By the way, not necessarily low quality
10 voice either. It was pointed out you could have
11 more or less whatever quality you want because I
12 don't need to specify ahead of time when I design
13 the service how much bandwidth I commit to it. I
14 can at least in principle have that flexibly
15 selected even on a per call basis.
16 If I want to pick on Joe once again to
17 sing my latest song to him and I want to
18 appreciate it in stereo sound then we may pay for
19 it on the CD quality link. Talking to my
20 grandmother, I would prefer a weaker link.
21 Breaking up, got to go granny.
22 Increasing blurring between the PSTN
23 and the Internet. What does that mean? It means
24 that at some point in the future, no one is
25 really sure. Anyone who tries to put a date on
232
1 it is a consultant. We won't talk about the PSTN
2 and the Internet. We'll call it the Net. We
3 will call it PST-Net. We'll call it something.
4 The good aspects of the PSTN will not be thrown
5 away and discarded but will be deployed and
6 deinvolved, probably not dramatically upgraded,
7 not by belief but certainly used within a larger
8 encompassing network. Evidently didn't talk fast
9 enough. The screen disappeared.
10 High value in controlling the PSTN with
11 data networks. The PSTN in practice has a lot
12 more flexibility in it than most people are able
13 to access because the 12-key DTMF keypad is not a
14 user friendly device.
15 If you are able to bring a screen-based
16 view, a Web-based view into a control panel we
17 are able to see this afternoon how quite
18 sophisticated conference calling applications,
19 for example, were managed. You don't know except
20 on the word of the presenter whether what was
21 going on in the background was a PSTN activity or
22 a Web-based activity. All you could see was
23 there was a Web-based control point. Carriers
24 will make the decision on the back end whether it
25 makes sense to carry this over a circuit switched
233
1 or a packet switched. That's not something an
2 end user ought to be worried about.
3 Big hurdle, still no clear path towards
4 quality of service on the public network. If you
5 want to do this Holy Grail application, realtime
6 collaboration, mixed media, whatever it is you
7 want to call it, you are going to do it on a
8 subnetwork. Where are they? Corporate LANs or
9 WANs, logically segmented from the public
10 Internet to provide more privacy, more assured
11 performance.
12 Corporations also happen to be the ones
13 who have the most to gain. They have the
14 spending dollars. Put these two together, that's
15 where you will see these applications happening
16 first.
17 What does this all say about the
18 regulation agenda, the policy agenda? Well, at
19 least at first sight, this seems to be good
20 news. Even if all this were about arbitrage,
21 price arbitrage and nothing more, there would
22 seem to be value of it from a regulatory agenda
23 in the sense of putting pressure on artificially
24 inflated prices, but there is more. This
25 industry even in its relatively nascent form
234
1 promises to achieve this goal of competitive
2 entry and flexible entry/exit and high degree of
3 integration and customer service and that's
4 broadly speaking the direction that we have been
5 trying to go for years and regulation has been
6 prodding us in this direction and either because
7 of or despite regulation, we seem to have arrived
8 at it here in at least one part of the market and
9 very happily, demonstrates that for some
10 instances, market competition can replace
11 regulation in many instances.
12 Capacity deployment is an issue which
13 according to the speakers today will be resolved
14 in the market. Solutions to shortcomings around
15 standards. One example is a good news, bad news
16 example would be around the voice coding
17 standards. Part of those have been
18 satisfactorily resolved by market forces.
19 Another part which is paying for the
20 darned things have not really been adequately
21 addressed, but no serious belief that
22 regulation's role is to get involved in that.
23 The market will sort it out and only the market
24 will pay if it doesn't happen quickly.
25 On the less clearcut side, there seems
235
1 to be a lot of regulatory open-ended issues. How
2 would you know parity has been achieved and how
3 will you know it? Are you going to measure it by
4 the volume of traffic? That was a question we
5 had this morning to which there is no
6 satisfactory answer at the moment. Under which
7 scheme will parity be achieved? Is it under the
8 current scheme modified or more like the Internet
9 scheme.
10 Does this enormous growth of the
11 Internet threaten the PSTN by overloading it with
12 traffic? If it does, there would be a policy
13 implication because the policy agenda includes
14 protecting the PSTN.
15 How do you deal with less
16 pro-competitive and particularly overseas
17 agencies and how do you deal with the fact that
18 the Internet is everywhere and nowhere and as
19 legitimately in New York as in D.C., and if this
20 is as big as our speakers seem to think and we
21 are heading towards universal data tone access,
22 how do we make sure everybody in society has that
23 in the form of universal access?
24 One thing seems pretty clear, the
25 regulatory environment doesn't fit. We can't
236
1 look at this new architecture in terms of
2 application, individual applications or
3 individual services.
4 They were, if you remember, read back
5 to the first chart, just applications that sit
6 upon a large common infrastructure. The way that
7 infrastructure is used may change very, very
8 unpredictably from month to month. Nobody
9 seriously tries to forecast the nature of traffic
10 on the Internet. Today the best you can do is
11 try to get an aggregate forecast and stay two
12 months ahead of it.
13 What is a voice call? If I have video,
14 if I have text chat, if the voice includes a
15 bundled voicemail, is that parity, or is this a
16 different service? So any attempt to try and
17 create an apples-to-apples based scheme here
18 which applies equal treatment to Internet voice
19 as PSTN voice seems doomed to fail, collapse
20 under the weight of its own logic.
21 Second point, if we try to continue the
22 general idea of paying for universal services
23 with a tax on players, too, looks increasingly
24 problematic. If one of the key effects of this
25 industry transformation is dramatic new market
237
1 entry and much more flexible role definitions,
2 how do we recognize a telecom player when we need
3 one? How do we decide whom to tax? How do we
4 decide on what basis to tax? It's going to get
5 really, really hard to do this.
6 Finally, per-minute transfer pricing
7 whether between players, between network
8 operators or even out to end users perhaps, it
9 makes sense only for session-based traffic.
10 Well, we also pointed out, several
11 people have pointed out that this whole Internet
12 structure is a combination of session-based
13 traffic and occasional traffic and transactional
14 traffic, et cetera so why pick one kind of
15 traffic and price the whole, and try and recover
16 costs on the basis of that.
17 Currently, you would have to make very,
18 very bullish forecasts about Internet voice for
19 its total volume of traffic to even to register
20 in the double digits amid the total Internet
21 capacity. So it doesn't look like a good way to
22 recover costs going forward.
23 So what might you do instead, and some
24 of these comments were from the floor and I
25 editorialized here a bit as well, I have taken a
238
1 broadly sort of stack-based approach here,
2 starting from the top of the protocol stack.
3 Services will be created in software,
4 likely a highly fragmented industry and service
5 provision could be pretty fragmented too.
6 Standard software residential rules should apply,
7 plus beefed up user protection around privacy,
8 security-type issues at the service provision
9 level.
10 IP networking, if our collective sense
11 and prediction is right, becomes the dialtone of
12 the next millennium, therefore data tone becomes
13 actually critical and one might imagine that
14 that's an immediate red policy flag. How do we
15 ensure that this happens?
16 Well, the evidence, if we look at both
17 the theory of the industry structure and what's
18 actually happening, there are going to be a lot
19 of IP providers. This feels to me like light
20 touches, business practices regulation
21 predominantly, not anything dramatically
22 customized to telecommunications.
23 Long distance infrastructure, a
24 commodity. It's going to be traded. In a few
25 limited cases today it is on a form of limited
239
1 futures market. Well, the kind of regulation
2 which would seem to make sense there is the kind
3 of regulation which you apply to a commodities
4 market.
5 High density buildings. These will
6 have two or more local access connections and by
7 and large the market will control price and
8 quality. But low density and small and local
9 residential access is likely to remain a problem
10 for a long time.
11 That pipe has to be layered with the IP
12 networking and the services on top to give it any
13 value. This feels like an utility kind of
14 business. Our experience with utilities probably
15 is what's relevant here. Anyway, I think -- a
16 question, my very last bullet point.
17 AUDIENCE MEMBER: Inasmuch we really
18 don't know what this new architecture is evolving
19 to it, it occurs that this may be the chicken
20 versus the egg situation. Given that we don't
21 know what that new architecture is, there is no
22 kind of regulation to apply to it, so consistent
23 with the thought there, though, perhaps, and I
24 think what I heard is that regulation should
25 reflect and promote the evolution, regulation if
240
1 required should reflect evolution.
2 MR. SPENCER: That's a good point. Any
3 other thoughts on this?
4 MR. GUMPER: I would suggest that most
5 residents' homes already have two pipes, a cable
6 and telco. From what I heard, it seems like,
7 with that remark, both could be used in the
8 future to access this network.
9 MR. SPENCER: I think that's the ideal
10 opportunity for us to bring our panelists into
11 debate. We heard the view expressed this morning
12 that the IP architecture ought to be highly
13 useful in bringing new infrastructure players
14 into play. Mark from Comcast represents one of
15 those potential sources of infrastructure. I'd
16 like to hand over the microphone to comment on
17 how he plans or whether he plans to use his
18 infrastructure in this way.
19 MR. COBLITZ: First let me thank Larry
20 and Kathy for holding this session. I think the
21 whole education process is a very important thing
22 to do, and I'm very pleased to be here.
23 What I thought I would comment on
24 because I just want to take a couple of minutes.
25 I think that's all we are going to have up here.
241
1 Dan made a statement this morning, who said there
2 are lots of ways for people to get access to the
3 Internet. And there was also a comment that Joe
4 made that I thought was insightful in the way we
5 talk about this, which is, is this a single line
6 or a multiple line. It's the whole concept of
7 line.
8 It's sort of when you are in a
9 packet-based network, it sort of changes and from
10 the cable industry perspective, what we are
11 looking at is to be a part of that infrastructure
12 that really implements the packet network and
13 implements it all the way to the home and
14 implements it in the backbone kinds of delivery
15 mechanisms that are necessary as well, so maybe
16 60 seconds on Comcast for those of you who don't
17 know.
18 We are the fourth largest cable
19 operator. We are also a cellular operator as
20 well. We are an ISP delivering high speed data
21 access which was my comment about Michael because
22 he is a subscriber.
23 We own QVC which interestingly has one
24 of the largest call center activities taking 80
25 million customer service calls a year to sell
242
1 product. We happen to be completing for local
2 telephone service in the United Kingdom. We sort
3 of cross a lot of the gamut in this. My role is
4 to bring new businesses into play. The one that
5 we are focused on is how do we take our local
6 network providing an alternative access mechanism
7 to everyone going over the LEC's line with
8 whatever sets of issues there happen to be there,
9 and then connect it either into a private
10 Internet, maybe small-I Internet, or a packet
11 network which for us would be at home, and
12 working with the rest of the cable industry, to
13 be able to have interoperability across the home
14 partners, but people like Time Warner and
15 MediaOne, and use that as an infrastructure to
16 deliver all of these kinds of services that we
17 talk about.
18 We'll take that as a jumping off point
19 and then people can ask specific questions
20 later.
21 MR. SPENCER: Thank you, Michael.
22 John, to some extent, you would seem to be in a
23 great position to decide from a somewhat general
24 field what kind of architecture you would like to
25 build with a combination of GTE's resources, IPN
243
1 networking with a lot of fiber from Quest. Can
2 you comment on what sort of architecture you are
3 planning to deploy or are deploying.
4 MR. COBLITZ: Let me also thank
5 everyone for the opportunity to be here.
6 Certainly I actually sat in a debate very similar
7 to this just several weeks back, but it was
8 entirely internal. It was someone dealing with
9 being an Internet provider, competitive local
10 exchange company, an incumbent local exchange
11 company. We are actually in the process now of
12 straightening out, sort of what makes sense from
13 a business and a policy perspective going forward
14 on the same types of questions, Internet over
15 voice, a wide range of issues, many of which show
16 up in dockets, of course.
17 And on the Internet over voice, one of
18 the, I guess we have actually already made the
19 internal transition, I guess. You threw up a
20 slide of our architecture. We have a scenario
21 already where if you take a look at the
22 exponential growth in data, the network is a data
23 network and voice is the application. We are not
24 talking about a circumstance where you are
25 building voice infrastructure. You are building
244
1 data transport infrastructure and voice is just
2 one of those pieces.
3 For some of us, it may run native over
4 the sonnet. It may run over the AP layer or the
5 switch layer. There are different places in the
6 infrastructure where you can carry voice. To
7 some extent it makes sense to segment the
8 problem. People are going to be in circumstances
9 where they are going to get told, you have a
10 choice.
11 There is a wide range of how your voice
12 is going to get from point A to point B. Do you
13 want that choice? There are lower and higher
14 quality of service available? Shall we make that
15 dynamically for you? Would you like to fix it on
16 one particular transition method.
17 In the network, there will be the
18 capability of doing that tradeoff and a number of
19 us will have infrastructure to support that.
20 Sure, there are some gateways involved and there
21 is certainly a cost. You want to make sure you
22 gain enough on reduced bandwidth and shared
23 multiplexing, but frankly, I know multiple
24 providers are working that space.
25 From a pure transport perspective, it's
245
1 quite possible that a lot of that is transparent
2 to the user. There are people today who are
3 using voice over the Internet. That's
4 principally voice as the application on their
5 current Internet connection, a very different
6 circumstance.
7 It might be worthwhile in terms of
8 working on the problem to really partition it.
9 We see the possibility of transparent voice
10 transmission being something where we can change
11 how we get it across the network as being
12 something that's going to be inherent and
13 something that can just be rolled out.
14 Now as a much harder questions about
15 what you do on the desktop, you get many more
16 complicated features if you actually want to talk
17 about voice over IP and you are actually playing
18 with that handset that the person is using. You
19 are giving him something other than that keypad.
20 Someone pointed out earlier that that
21 keypad isn't friendly. Today's handsets are
22 pretty solid. We need to think about when we
23 stop talking about transport of IP, voice over IP
24 and we instead talk about interfacing to the
25 network using computer-based applications, you
246
1 know, there is a wide range of issues in terms of
2 reliability.
3 Someone talks about the Internet being
4 reliable, I'm not as concerned about that.
5 Certainly we know how to do the IP transport of
6 voice in a manner that's as reliable that we can
7 do it over an infrastructure that's switched.
8 That's not a problem.
9 What we have difficulties with is
10 conceiving of the fact that the person trying to
11 make the emergency call may have a software vault
12 on their machine. We have a problem if they only
13 have access to that beautiful Windows interface.
14 How is it going to be moved around the
15 backbone networks. Someone mentioned the
16 settlement issue. One of the reasons we haven't
17 reached a decision is because of the settlement
18 issue. It becomes international. When this
19 problem is resolved, it will probably be resolved
20 on an order of magnitude and scope that is
21 enormous. You'll be talking about providers who
22 are working on networks on the side that we are
23 looking at. You are looking at multiply OC/48,
24 40 times OC/48 on 24 strands.
25 The IP settlement issue will be
247
1 resolved. It has to. We have already solved it
2 for usage-based. A long time ago everyone paid
3 the same rate for Internet services. Some
4 brought up a session from the Internet because
5 you had domain names.
6 The fact of the matter is that in 1989
7 we took the total cost of the year divided it by
8 the number of uses, that gave you a fee. No
9 marketing was involved here. We solved that. We
10 now get into a situation where instead of being
11 flat based, we have usage based. Your average
12 business is paying based on how much they use.
13 The Internet is usage-sensitive. The
14 average business is paying for its T1 usage. The
15 settlement, international reconciliation and
16 introduction long-term of distance sensitivity
17 will change the arbitrage question, will make
18 dramatic changes in the transport.
19 I think you'll see voice over IP
20 happening. It will be just another way of
21 transport. It will have its own way of economy.
22 It may actually work out a little cheaper in some
23 circumstances and it will be an option to the
24 user.
25 All of that discounts the access side
248
1 of IP. We spent a long time trying to figure out
2 how we have universal access and ensuring that we
3 have very reliable access. We can't dismiss that
4 quickly and go from computer-based access to
5 voice and ignore the issues that we have evolved
6 all these years.
7 MR. SPENCER: Frank, can you perhaps
8 give us a perspective from the Bell Atlantic
9 incumbent LEC point of view, both threats and
10 opportunity sides to this. How do you see it?
11 MR. GUMPER: Well, obviously we have
12 been talking all day about the threats and it's
13 nice to hear that they are going to be resolved
14 pretty quickly. All we need to do is get rid of
15 some of these high regulatory prices which are
16 substantially above cost.
17 We look on this as a great opportunity,
18 obviously, and I think there is some agreement as
19 to this convergence, although I'm not quite sure
20 how fast it's going to occur. I do think that
21 the issues that -- and I deal with public policy
22 issues, I'm not dealing so much from a marketing
23 business viewpoint -- but the issues of
24 reliability, the safety factors, the privacy
25 factors, I think we tend to assume that those are
249
1 going to be taken care of. Hopefully they will.
2 It's not obvious that they are going to be taken
3 care of that quickly.
4 I addressed a group of Internet people
5 back in Berkeley earlier in March and was asked
6 to take a look at some of the long-range policy
7 issues. I pointed out to them because they had
8 spent the whole day talking about access charges
9 and should they pay them or not pay them?
10 Years ago we took care of our own
11 network. By and large, the incumbent LECs were
12 responsible for deciding how their networks would
13 operate and how they would be interoperable.
14 One of the System Seven systems went
15 down in this area. Congress was up in arms. We
16 had Congress creating a reliability council.
17 They said you guys have lost your flexibility.
18 Your network went down. We said it was a
19 software glitch. They said it can't happen.
20 Your networks can't go down.
21 At some point, the Internet is going to
22 face that same problem. You have outages and
23 people tend to ignore them. The last one did hit
24 the papers. These are going to be very, very
25 important issues and how will this affect an area
250
1 that doesn't want to see government involvement.
2 As far as a utility, we have to address
3 that. We in the cable industry, we like to see
4 our pipe, hopefully expanded bandwidth pipe, to
5 the home be the source where people jump on to
6 this fabulous new world. When we look at these
7 businesses, and we have the same debates that you
8 are having internally as to how we want to play
9 and how we want to enter this market. We have
10 this other overhanging structure and that's
11 regulation.
12 Right now our primary concern, I think,
13 goes along the line, we want to be in this area
14 but we are seriously concerned that the present
15 regulatory structure, the present regulatory
16 rules that exist are going to be significant
17 inhibitors. Therefore, in the near term, our
18 approach is probably going to be more aimed at
19 trying to fix that situation and once we get
20 that, some of those things removed -- and we
21 can't be a major Internet provider in our own
22 right because, quite frankly, we are not allowed
23 to cross a LATTA boundary on this stage. We
24 can't have that extensive Intranet or Internet
25 capability. We are going to be focusing in on
251
1 those regulatory type things. We just hope that
2 it doesn't take too long to fix that regulatory
3 scheme.
4 MR. SPENCER: Thank you very much,
5 Frank. Dave from Cisco, you have heard quite a
6 bit of skepticism expressed today about the
7 scaleability of quality of service and total
8 ability to handle traffic, particularly in the
9 wide area of public Internet. Cisco is the
10 largest supplier of some of the components of the
11 public network. What have you got in the works
12 to address this.
13 MR. LaPIER: There are a lot of areas
14 where addressing bandwidth is appropriate. There
15 are the backbone of providers who are growing
16 their infrastructures, and are deploying larger
17 and larger routers and switching capabilities,
18 either running their routers on ATM networks,
19 running their routers on sonnets to carry their
20 application space. They have interconnects to
21 each other and then to the whole tier of
22 Internet, of the smaller local regional access
23 Internet providers.
24 Now, the biggest challenge in terms of
25 voice and many of these applications has to do
252
1 with the fact that for that, for each tier to
2 make money, they have to oversubscribe. They
3 have to utilize that subnetwork to the greatest
4 extent possible. So that means that there are
5 challenges that are introduced to support
6 queueing and the ability to manage certain
7 traffic types.
8 It isn't so much that the Internet now
9 in many respects is you get what you pay for,
10 anything goes. There is not necessarily a
11 service level agreement for residential Internet
12 customers, so we are looking at ways to allow
13 business customers, especially, to be able to
14 have different grades of service, quality of
15 service, and then have the packet data treated in
16 different ways based on who they are so on the
17 one hand, we throw horsepower, we throw bigger
18 pipes on our boxes and we let networks grow.
19 On the other hand, we have to create
20 new capabilities that our customers are very
21 interested in that let them create grades of
22 service. By creating grades of service, they can
23 differentiate on prices and they can treat
24 different types of traffic differently and they
25 can enable the multimedia applications. They can
253
1 enable the voice interconnect to use excess
2 bandwidths when it's available. Some of these --
3 so it's important to realize that there are tiers
4 of providers who are going to have different
5 business strategies in terms of how they are
6 going to offer their services.
7 And then we need the network equipment
8 vendors have to improve speeds and feeds and
9 densities. They also have to create ways of
10 differentiating services and creating more kinds
11 of networks.
12 One way in which we are doing that is
13 to be able to apply separate routes and separate
14 queueing criteria to different grades of traffic,
15 literally mark a different user's streams of
16 packets with different priority and when there is
17 a congestion condition, treat the queues
18 differently so you create multiple queues over
19 the same networks, and that's a way of creating
20 subnets that have different qualities of
21 service.
22 The next big problem is you always can
23 joke of whether you can measure the traffic or
24 forward the traffic. We finally have enough
25 horsepower that we can both measure and forward
254
1 traffic. We are getting closer. I don't think
2 it might meet John Curran's needs in his core
3 backbone but at the edge of the network, it is
4 beginning to do so.
5 The idea here is that we just add
6 processor tricks and merge layers so that we
7 literally can count the packets by class of
8 service and treat them in a way that will let the
9 providers charge differently.
10 And then the hard part, though, is now
11 we have all these tweaks in the boxes. They are
12 great for enterprise. How do we make that
13 applicable to a Bell Atlantic or a GTE in terms
14 of service management. How can we turn a service
15 order, move, add, or change into something that
16 actually happens automatically for customers, and
17 that's probably the biggest gap.
18 To complete the answer on the backbone
19 side, we have speeds and feeds. We have the
20 inherent reliability of the Internet protocols,
21 and then on the services side, we have the notion
22 of creating the capability to differentiate and
23 add value in the network and eventually that will
24 go all the way to the ability to turn services on
25 and off on a per-customer basis. So it's a
255
1 further degradation and a further blurring and a
2 further movement towards voice as an application
3 over an unified set of networks. That is the
4 direction that I see.
5 MR. SPENCER: Thank you very much,
6 David. Finally, Robert McDowell of ACTA.
7 Robert, you and ACTA have been quite vocal in
8 pointing out that Internet telephony highlights
9 broader inequities, shall we say, in the current
10 pricing and tariffing arrangements. We heard
11 Frank echoing the fact that he was concerned
12 about Bell Atlantic's ability to compete was
13 constrained by current regulation.
14 What specifically do you think are the
15 regulatory hot spots that ought to be addressed
16 to enable you to compete the way your members
17 would like to in this Internet telephony arena.
18 MR. McDOWELL: First of all, thank you
19 very much to NTIA for inviting ACTA here today.
20 I'm not sure if I should be on this panel or the
21 policy panel. How many people in the room today,
22 prior to today, had heard of the ACTA Petition?
23 ACTA actually files many petitions, but this
24 actually became the ACTA Petition. We filed one
25 yesterday on access charges actually, all
256
1 related.
2 ACTA, first of all, is an association
3 of over 200 telecommunications service
4 providers. It was founded in 1985 in the wake of
5 the breakup of AT&T. Competitors who had goals
6 in common still have goals in common here in
7 1997. There are Internet service providers,
8 interexchange carriers, CLECs and a whole host of
9 other folk us who support those industries.
10 I'm glad I have an opportunity to speak
11 today because obviously when the ACTA Petition
12 was filed about 18 months ago, the mainstream
13 press, in particular, wanted to polarize. The
14 press likes conflict -- I guess it's somewhat
15 interesting that we are here at the National
16 Press Club here today -- but polarize these
17 issues as a fact, it was opposed to Internet
18 telephony. That's not the case at all. I won't
19 go into all the nuances. Folks are growing weary
20 after a very rich day.
21 One thing that Internet users in
22 particular need to keep in mind is that crazy
23 aunt locked in the basement which nobody wants to
24 talk about which is universal service.
25 On February 8th, 1996 when the
257
1 Telecommunications Act was signed into law, the
2 definition of universal service changed
3 dramatically. Prior to that date, universal
4 service was basic telephone service.
5 Basic voice service conveyed the old
6 classified circuit system. Now it has been
7 expanded. We have the new world, and the statute
8 is open-ended in terms of defining what universal
9 service is. We have $2-1/2 billion being
10 allocated to the wiring of schools and libraries
11 to the Internet with a 90% or up to a 90% usage
12 discount for those entities, health care
13 providers as well. That is a noble goal, but
14 that changes the whole situation.
15 We upped the price of what services we
16 consider to be universal, yet we still are
17 classifying one set of companies in an old
18 mindset, regulatorily speaking, to be the
19 subsidizers of those services.
20 The ACTA Petition, I wanted to point
21 out several things. Three of them I'll mention
22 now. One is that the regulations governing
23 telecommunications are antiquated. Even after
24 passage of the Telcom Act of 1996 they did not
25 keep pace with the blending or the blurring, as
258
1 was mentioned earlier, of technologies. We have,
2 thanks to the Internet, voice, video, and data
3 all blurring together. A bit is a bit is a bit.
4 It's all data.
5 Also the ACTA Petition wanted to point
6 out that access charges were too high. That in
7 the current environment where we do not have
8 robust or even any real competition at the local
9 level, that last mile of wire to the house or to
10 the business, we have a monopoly. That gives a
11 captive audience which is extracting overinflated
12 access charges. IXCs and IXPs are configured
13 identically into the PSTN.
14 How do we differentiate in the law and
15 from a policy perspective in those entities and
16 do we want government to have a thumb on the
17 scale. Policy should encourage the use of the
18 Internet and ACTA agrees 100 percent. Most, if
19 not all, of ACTA's members are pursuing many
20 applications of the Internet, including voice
21 telephony.
22 Regulation cannot inhibit the
23 development of other technologies whether it's
24 the Internet or not. If we are going to rely on
25 the market, we have to be careful to have
259
1 government not artificially encourage or
2 artificially discourage certain technologies with
3 its thumb on the scale. We don't have that
4 opportunity right now.
5 But something that's very, very
6 compelling -- back to universal service, and this
7 is a statistic that came out in the context of
8 the WTO negotiations and it was reinforced during
9 those, I should say.
10 There are roughly six billion people on
11 the planet earth. Half, one-half have never even
12 made a phone call, let alone own a phone, let
13 alone own a PC for Internet telephony, video or
14 data. We have to keep this in mind because we
15 are talking about the PSTN or the Internet. We
16 are not talking about what's within the borders
17 of the United States. By definition, we are
18 speaking about what's beyond. That is something
19 that regulatorily speaking, Michael, to answer
20 your question, is something we are going to have
21 to consider defining universal service
22 differently and who is going to pay for it.
23 Right now it's the IXCs.
24 We cut big checks. We have a bunch of
25 them coming due at the end of the year which some
260
1 of them are just now noticing. This is something
2 that we have to keep in mind and which we'll come
3 back hopefully in a year or so and discuss that
4 topic maybe in a little more detail as things
5 changed.
6 MR. SPENCER: Thank you very much,
7 Robert. We have some time now. We have five
8 very different perspectives. I'm sure you have
9 questions for them.
10 MR. McDOWELL: If you are still awake.
11 MR. SPENCER: Mr. Berninger with a
12 planted question.
13 MR. BERNINGER: So half the people in
14 the world have never had access to telephone
15 service. After 125 years, that's kind of an
16 indictment on the current way of doing business
17 and that really should tell us, okay, let's try
18 something else. You are right, you can't have a
19 particular group try to reinforce or move money
20 around. We need to try something else.
21 MR. McDOWELL: If I could respond to
22 that, the answer to that is competition which
23 ACTA strongly endorses, if not rapidly. That
24 also to point out that one-half figure is outside
25 the United States. The penetration inside the
261
1 United States is extremely high. But it wouldn't
2 be if it weren't for current universal service
3 laws that have been in effect for a while.
4 We are talking about the wonderful
5 applications of the Internet. Let's remember, we
6 are talking about 3 percent or less, 1 percent of
7 the world's population are able to use this on
8 expensive PCs. Bill Gates is very vehement about
9 having telecommunications access to every single
10 human being on earth. Let's get food in their
11 stomach. We have more immediate and more urgent
12 problems to solve. We can't feed the six billion
13 people on the planet.
14 MR. SPENCER: That might be a bit of
15 scope creep for today's session.
16 AUDIENCE MEMBER: I hope I'm not going
17 to add to the scope creep. I am struck by what
18 an interesting cross-section of the whole
19 Internet industry's past, present, and future we
20 have here. I want to ask a question that does
21 take us a little bit beyond Internet telephony.
22 It's really about what I think a lot of people
23 would be is the foundation of the Internet, open
24 standards and agreement to carry traffic, to
25 carry packets wherever they are supposed to go.
262
1 We have heard over the last couple of
2 years concerns about where HTML standards are
3 going, where Web standards are going, are we
4 really staying open and standardized there. In
5 the early days back home, there were some
6 questions about whether some tricky packet
7 prioritization techniques were going to be used.
8 John raised some of the settlement questions and
9 the peering questions.
10 I'm curious for anyone who wants to
11 answer, where are we going? Is the Internet
12 going to be as open as it has been as you all who
13 were, some of whom are in the business, get more
14 into the business and some of you who are just
15 getting in come in with full force?
16 I have always regarded the openness of
17 the Internet as one of its greatest assets and
18 also one of its greatest bulwarks against
19 excessive government involvement. I'm curious
20 from all of you where you think we are headed in
21 terms of the openness of this network. Thanks.
22 MR. GUMPER: I think certainly from our
23 perspective, we have strived in our network to
24 make it open. One of the problems that was
25 alluded to as we have gone through this
263
1 discussion, we have long suffered that the LECs,
2 our switches were not open -- they were our
3 switches but we couldn't do anything with them
4 because we had to go back to our manufacturers
5 and beg and plead with them every time we wanted
6 to make a change. They would say maybe in the
7 next generic, we'll make that capability
8 available.
9 When you look at the Internet, it's
10 extremely open. Apparently that's what has led
11 to this phenomenal growth. The reason why it was
12 open I think was because of the history. It was
13 basically a fairly closed community that wanted
14 open communications between them and there really
15 wasn't a commercial endeavor there. I think that
16 is a real issue that as the Internet becomes
17 commercialized and really becomes, if in fact it
18 becomes a global electronic commerce vehicle, our
19 companies for commercial gain are going to try to
20 put closed applications on it in order to extract
21 higher profits. It's an interesting subject.
22 You may need government intervention in the
23 future to keep the Internet open.
24 MR. COBLITZ: I'm going to respond on
25 two fronts. I need to throw a little disclaimer
264
1 as has been a tradition today. I actually am the
2 co- area director for the Internet engineering
3 task force operations and network management
4 area, which means I supervise the horde of
5 network engineers who are working on standards in
6 the IETF, in particular in the area of operating
7 the network and managing it. This is such things
8 as SNMP and a whole bunch of other colorful
9 specs.
10 We have seen over since 1991, 1992
11 where many of the networks that compose the
12 Internet began to switch from nonprofit to
13 for-profit. We began to see entrepreneurial
14 initiatives forming. We began to see the growth
15 of the early Internet. The same community that
16 was there in the beginning is still there, even
17 if they are working for for-profit firms.
18 The level of energy put into the
19 standards effort, if anything, has gone up. We
20 now have, where we used to have meetings of 70
21 people somewhere in a small hotel where no one
22 knew, we now have events of 1,500 taking place
23 where we had over 100 working groups meeting over
24 the course of a week. It is, with the energy
25 level that has been contributed to the Internet
265
1 by commercial firms coming in has helped advance
2 this growth.
3 I'm not too concerned on the standards
4 track. There are times when it gets rather
5 entertaining to watch the divergence of multiple
6 approaches to a problem and rein them down to in
7 the Internet.
8 One standard is what we try to get for
9 all problems. We don't like having multiple
10 standards unless we can avoid it. The actual
11 standards process, the openness is there. The
12 IETF remains open. All of its communications are
13 conducted online. It's available for everyone to
14 avail themselves. Many of the entrepreneurial
15 firms in this room have taken advantage of that.
16 Let me talk about settlements and
17 interconnection very briefly. There is no
18 openness of the Internet which entitled people to
19 send traffic for free. It is true that smaller
20 ISPs, it's possible to connect to the Internet
21 and originate traffic. But there is a spectrum
22 between a backbone providers and an end user
23 customer and at some point someone has to pay for
24 the infrastructure.
25 The backbone providers are installing,
266
1 we are all seeing dramatic growth doubling,
2 depending on who you ask, for the ISP is doubling
3 from a four-month to a six-month cycle. The ISPs
4 might see it every 12 months. We are seeing
5 doubling on very, very short cycles. That means
6 a lot of investment that has to be recovered.
7 Backbone providers put it in. They have
8 traffic. It has to be charged back.
9 There has been an absence of the
10 settlement at the top of the Internet hierarchy.
11 Many said it's more important to keep the network
12 working than it is to figure out the exact
13 details of the cost. As things get larger and
14 the order of magnitude increases, particularly as
15 we now go truly global and we see Internet
16 infrastructure being operated by companies that
17 span the globe, there is a need to ensure that
18 there is adequate cost recovery. We have already
19 seen efforts in the past.
20 I don't think we really have to worry.
21 There is a robust mix of backbone operators.
22 There will be a market in providing transit
23 services to ISPs. There already is a market. As
24 long as that market exists, I think we can rely
25 on the industry to work that out.
267
1 There is certainly a case to be made
2 that as in all industries, we want to be sure we
3 have robust infrastructure, we have players who
4 can support the reliability that we need and can
5 be there on an ongoing basis, but we do need to
6 have multiple competitors in the market as the
7 Internet scales, that we have multiple
8 competitors at every layer, whether that's access
9 or backbone or in the transit in between.
10 MR. LaPIER: I think the first point is
11 that to remember that notion that was introduced
12 earlier of a platform and that these are
13 platforms, and therefore standards make economic
14 sense in that context so the macroeconomics of
15 standardization is still in place. The second
16 point is that because of the standardization,
17 that's what allows -- it's one of the competitive
18 advantages of voice over IP, because everyone's
19 got IP, whereas all of these PBXs all around the
20 world have their own proprietary signaling
21 interfaces that are very difficult to
22 interconnect with. We can bypass these legacy
23 PBXs through -- that bypass is enabled through
24 standardization. That all fits together in this
25 framework.
268
1 MR. COBLITZ: Particularly since you
2 mentioned at home, I'd like to at least address a
3 little bit that type of issue, and Robert made a
4 comment that I'm going to take issue with, which
5 is he said a bit is a bit is a bit, and I have
6 heard that for years.
7 All of us who are thinking about
8 Internet telephony that at some point in time has
9 got to be turned into something where the stream
10 of those bits happen in a particular order in a
11 particular timing, recognize that not all bits
12 are equal in that system. Whether the solution
13 is RSVP to give quality of service to a
14 particular kind of product or whether it's some
15 other mechanism of doing that, I think those
16 systems are going to have to exist.
17 John's comments which I absolutely
18 agree with on the settlement issue, that there
19 may have to be some kind of economic
20 justification and mechanism for this kind of
21 service to be delivered through the network.
22 Do I agree that it should be a
23 standards-based mechanism? My answer would be
24 yes. We have to cross the network boundaries.
25 Things are going to be different. I just don't
269
1 think it's going to be the endgame.
2 The second piece is that some
3 customers, and I think for a long time are going
4 to want in effect virtual private networks where
5 they, working with whoever the provider is, will
6 engineer a solution for themselves and because
7 they want that particular solution, get a certain
8 value out of the network in being able to send
9 things across the network.
10 Open Internet, open standards, but
11 still now something that's different for that
12 person, that set of people or that provider
13 that's different from other people because they
14 paid for it. The transition of this network into
15 something that is as ubiquitous as a packet
16 network will eventually start to say, well, how
17 do I use a packet network for all these different
18 services and what are the ways that I
19 economically get a return for the different
20 things that I do? I think all those are going to
21 play out.
22 The one comment that we are thinking
23 about in the at-home environment and may happen
24 is what if you are a customer of at-home and we
25 are doing the Web hosting and you want to get to
270
1 our customers who, now we have an engineered
2 circuit all the way down to the home without an
3 issue of anyone in between, totally monitored and
4 engineered, could you provide that to somebody on
5 a rapid basis, and the answer is yes. Does that
6 disadvantage the other person? No. It puts them
7 on an equal footing of everybody else who is out
8 there. It might advantage the person who may
9 want to come to us. If that were to happen, I
10 don't think that that's a bad thing.
11 It's not a standards-related set of
12 issues. I think we are going to see that kind of
13 thing. I think it's part of a virtual private
14 network that that Web site is trying to get.
15 MR. McDOWELL: Briefly, first of all,
16 when ACTA filed the petition that a bit is a bit
17 is a bit and that we shouldn't be differentiating
18 between voice bits and data bits.
19 Obviously, from a regulators'
20 perspective, ACTA was arguing that they are
21 different. Telephony is telephony under the
22 law. Whether it's two tin cans on a string, the
23 law doesn't differentiate that. It may be bad
24 law. That's when the ACTA petition becomes the
25 Internet petition.
271
1 Real briefly, in terms of the openness
2 issue, what we have now related to that is an
3 artificial government incentive of maybe $30
4 million to $40 million to hype traditional
5 handset to handset telephony over the Internet.
6 In the short term, handset to handset,
7 there is not going to be a big market because of
8 liability. But Internet years are sort of like
9 dog years and technologists are doing wonderful
10 things with improvements that occur
11 exponentially.
12 In the midterm, I do see the Internet
13 being subjected to, for good or bad, plain old
14 handset-to-handset voice transit because cost
15 will go down. We are still in the innovative
16 stage where things are expensive like gateways.
17 There are a lot like VocalTec, who are investing
18 millions of dollars in handset-to-handset voice
19 over the Internet gambles and they probably know
20 a little bit better than I do.
21 MR. SPENCER: That turned out to be a
22 very rich question. I do want, given the
23 richness of the answers, quickly to summarize
24 that.
25 We heard a spectrum of reviews from,
272
1 it's open but open doesn't mean one size fits
2 all. We heard a view that says commercial
3 interest is not making it any less open, a view
4 that said commercial interest is jeopardizing
5 openness and another point of view which said it
6 is not in the economic interest of the players,
7 commercial players, to get in the way of
8 openness. Now, these are all from people in the
9 industry so we have an interesting spectrum of
10 view.
11 Shall we move on to the next question.
12 AUDIENCE MEMBER: We have a question
13 from somebody via E-mail, Dale Hoover. He asks
14 in regards to Internet fax traffic, will it still
15 be profitable if prices fall, especially if there
16 is a price war amongst IXCs and IXPs, who will
17 come out the winner?
18 MR. SPENCER: I think we should aim for
19 quick answers to that. Who would like to take
20 that on?
21 MR. McDOWELL: ISPs.
22 MR. LaPIER: Customer.
23 MR. GUMPER: I think the customer is
24 going to win but it's going to be via the ISP.
25 BY MR. CURRAN: I think the customer
273
1 will win and in the long run it will just reflect
2 the fact that to offer some of these services is
3 going to take a long period of time for the
4 pricing to settle out and players had better be
5 prepared to be in the market for many Internet
6 years, maybe even many realtime years.
7 MR. GUMPER: Let me make a comment on
8 that. I'll take it from personal experience. My
9 office is in New York. I'm down in Washington
10 quite a bit. If I need something and I need it
11 quickly, I ask my secretary to fax it to me. We
12 have an internal E-mail system. If I wait up to
13 a half-hour, I will E-mail it but if I need five
14 minutes, it gets faxed to me because sometimes
15 the E-mail will be here in two minutes and
16 sometimes it will take a half-hour.
17 MR. SANDY COMBS: I know we are up
18 against a time element. I would like to say
19 thank you to NTIA. Thanks to all the panelists
20 all day long, and especially the people that
21 stayed here in Washington. The Voice on the Net
22 coalition started based on ACTA's petition 18
23 months ago. We were born then. It's turned into
24 a multibillion dollar question mark and it's also
25 turned into a great conference producing entity
274
1 for Pulver.com.
2 MR. McDOWELL: When are we getting our
3 commission check?
4 MR. SANDY COMBS: We did ask you to
5 attend, and I'm sure someone will be there. More
6 importantly, it's turned into a dialogue of
7 intellectual entities from across the spectrum
8 that normally wouldn't come about.
9 I live in Vermont. I work out of my
10 house in Vermont as director of the VON
11 Coalition. I couldn't have done that five years
12 ago. Yet we were able to help NTIA. We were
13 able to help many, many carriers bring about what
14 has now become the Internet telephony dialogue.
15 For that I truly would like to thank NTIA and all
16 the panelists today. Thank you.
17 MR. SPENCER: Thank you, Sandy. Thank
18 you to the VON Coalition for, in conjunction with
19 NTIA, bringing us all here today. I'd like to
20 add my word of thanks to the panelists here and
21 pass the microphone to Larry Irving for a
22 wrapup.
275
1 MR. IRVING: I'm going to do a wrapup.
2 I'm going to begin my ending as I began my
3 beginning, by thanking the VON Coalition. I also
4 want to thank the folks at NTIA, particularly
5 Kathy Brown.
6 It's amazing how many things can happen
7 in 18 months. When I listen to Sandy, none of us
8 were not thinking about how we were going to
9 address this issue because none of us were
10 thinking about it until our friend at ACTA put
11 that forward. There are some issues that I
12 thought were important. Somebody asked me as I
13 went over back to my office whether there was a
14 definition for Internet telephony and others were
15 saying we should never use the term Internet
16 telephony because when we use it, people want to
17 regulate voice on the Internet as if it's a
18 telephonic service. I agree that we should stop
19 regulating a tin can on a string.
20 When you start talking about this as a
21 telephone, if you set the two cans on either end
22 is a telephone and somebody would want to
23 regulate it. We'd have 50 states and 156 nations
24 with regulators.
25 The definition that I heard that I
276
1 think was the best definition was that this just
2 isn't plain telephone service, but that it's
3 multimedia. I don't think there is or will be a
4 perfect definition. There are some other things
5 that made an impression on me.
6 I tried to be quiet because my mom said
7 it's important every now and then to open your
8 ears. Some of my colleagues, I think are shocked
9 not saying anything to the end of it. I really
10 wanted to hear what people were saying. One of
11 the things that I was talking about, this is
12 about more than just telephone calls. I think
13 that was the first thing on our mind. We'll have
14 a different opinion on that going forward.
15 I think the point that Danny made that
16 this, we have got to keep in the front of our
17 mind the privacy of the user. That's why the
18 last answer, who is going to win, the consumer.
19 That's what we want to have happen every time,
20 have the consumer and user win.
21 We want to stop thinking from the
22 traditional top-down regulatory models because
23 they don't and won't work in this arena.
24 None of us 18 months ago could have
25 figured out where we were going to be today. Not
277
1 one of us can figure out where we are going to be
2 months from now. If we use traditional top-down
3 regulatory models we are not going to get this
4 right.
5 There is a role for government. One of
6 the roles for government is to educate, make sure
7 people understand what this is about. I think
8 for the Federal Government, the U.S. Federal
9 Government too, one, we need to educate domestic
10 partners, the states and the cities about what
11 this is, and I think by using the Net -- and a
12 lot of folks here were from those particular
13 jurisdictions -- it's also important to regulate
14 our friends and colleagues overseas. There is a
15 need to make sure internationally we don't do
16 things to create barriers or obstacles.
17 As I was listening to Robert and I
18 don't mean to pick on him, but it was
19 interesting, but we talked about the investments,
20 Nortel, AT&T, and to my mind, that is a good
21 thing.
22 Premature government involvement which
23 stifled that investment, we wouldn't see the
24 hundreds of millions of dollars that's going to
25 be invested if in fact government started getting
278
1 into this and regulated right or wrong because
2 the fear of regulation will stifle some of that
3 investment. We see it time and time again in
4 telecommunications where people did not make
5 investments. We are trying not to get involved.
6 I want to talk to my friends in
7 Finland. I want to talk to my friends in Iraq,
8 Asia, who have concerns about what Internet
9 telephony means for AT&T. Among the benefits of
10 staying out of it are increased efficiencies in
11 economy. Among those benefits are how they are
12 going to enter the 21st Century. This new
13 information age will become information
14 societies.
15 There are more than 65,000 people
16 working in ISPs and there are hundreds of
17 thousands of people who are working in the
18 industries that have been created because of what
19 Internet telephony means, not Internet telephony
20 by itself, but the bundle of technologies
21 surrounding it.
22 You can create hundreds of thousands of
23 jobs on this planet and those countries that are
24 going to be part of this are going to have to
25 focus on that, but one thing that is absolutely
279
1 right, and I can give kudos before I criticize,
2 Robert, is talk about universal service. He is
3 right. It is probably the most important issue
4 that we don't have even closer answers for, how
5 we are going to change the old model and go to
6 universal service model.
7 Yes, we have universal service
8 precisely because we have subsidies in this
9 country. We have the same rates as Canada
10 because we say we are going to take explicit and
11 implicit subsidies.
12 We are moving away from that while we
13 move from a monopoly model. 80 percent of the
14 world's households don't have a telephone.
15 50 percent don't have a telephone at home.
16 99 percent of the people in this world do not
17 have a computer in their home. So we have got to
18 figure out how to lower the prices and lower the
19 barriers. You can talk about Internet telephony
20 but you don't have either a computing device.
21 The debates we are having here today aren't going
22 to change those numbers.
23 As you see things like PCs and Web TVs
24 and those kind of devices, and I think Bill Gates
25 is probably right, we do have the capability in
280
1 this world to get most of the people connected to
2 some type of communications network.
3 It shouldn't have to be food or
4 telephony -- food, shelter and clothing most of
5 our lives. I think most people believe that
6 communications, while not one of the big three,
7 is probably the next one after those three. And
8 the types of these we are seeing are going to
9 include those.
10 This was discussed a year ago when
11 Kathy went to an Internet conference in New
12 York. Kathy started talking to the people from
13 the VON Coalition today. There are a lot of
14 people that put this together, but because of
15 Kathy, we pushed a lot further.
16 We know more about each other, we know
17 a lot more about where this technology is going
18 to take us. We have to continue to talk to each
19 other.
20 I don't know about you, but I learned a
21 lot today. I'm going to leave here today with
22 more questions and answers. I am going to leave
23 here today with more questions than I walked into
24 this room with.
25 When you look at the aggregation of
281
1 minds and intellects out on the Net, if we keep
2 talking to each other, we'll get this thing
3 right. Voices on the Net needs to be voices on
4 the Net. The dialogue began today. We have got
5 to keep going. Thank you very much for coming.
6 I look forward to talking to you all either on
7 the Net or in the next months and weeks. Thank
8 you.
9 MS. BROWN: I just had one piece of
10 business. If all goes well, we will post the
11 proceedings of today on our Web site, I hope
12 within a week.
13 Our transcriber has done an incredible
14 job. I just watched her listening to Larry.
15 We'll do our best. Those people who had
16 presentations, we would like to post them as
17 well.
18 I would like to personally thank all of
19 you that came and participated and another
20 commercial advertisement, we hope on November 6
21 to hold a forum on wireless local loop, which I
22 hope will be just as invigorating as this
23 discussion. Thank you again
24 (Whereupon, at 5:10 a.m., the Forum
25 adjourned.)
Back to Top
Morning Session Transcript
Internet Telephony Forum Page