1

1	UNITED STATES FEDERAL TRADE COMMISSION
2	and
3	UNITED STATES DEPARTMENT OF JUSTICE
4
5
6
7	SHERMAN ACT SECTION 2 JOINT HEARINGS
8	UNDERSTANDING SINGLE-FIRM BEHAVIOR:
9	REMEDIES
10	THURSDAY, MARCH 29, 2007
11
12
13
14
15	HELD AT:
16	UNITED STATES FEDERAL TRADE COMMISSION
17	6TH & PENNSYLVANIA AVENUE, N.W.
18	WASHINGTON, D.C.
19
20
21
22
23
24
25

2

1	A P P E A R A N C E S
2
3
4
5	REMEDY IN THE FACE OF TECHNOLOGICAL CHANGE
6
7
8	MODERATORS:
9	Douglas Hilleboe, Federal Trade Commission
10	Ed Eliasberg, U.S. Department of Justice
11
12
13	PANELISTS:
14	Michael Cunningham, Red Hat, Inc.
15	Renata B. Hesse, Wilson Sonsini
16	Marina Lao, Seton Hall Law School
17	William H. Page, University of Florida
18	Howard A. Shelanski, UC Berkeley
19
20
21
22
23
24
25

3

1	P R O C E E D I N G S
2	- - - - -
3	MR. HILLEBOE: Good morning, everyone, thank you
4	for coming. I'm Doug Hilleboe, attorney with the
5	Federal Trade Commission, Office of the General Counsel,
6	I'm going to be one of the moderators here today for
7	this third session on remedies. My co-moderator is Ed
8	Eliasberg, he's an attorney with the U.S. Department of
9	Justice, Legal Policy Section of the Antitrust Division.
10	Before we start, I need to go over a few
11	housekeeping matters. As a courtesy to our speakers,
12	please turn off your cell phones, Blackberries and other
13	devices that make a noise, and I'll ask the speakers to
14	do the same, they actually interfere with the
15	microphones and we had a little problem with that.
16	Second, the restrooms are located down the hall,
17	through the double doors that you came through. Third,
18	in the unlikely event that the building alarms go off,
19	please proceed calmly and quickly, as instructed. If we
20	must leave the building, take the stairway which is to
21	the right, on Pennsylvania -- on the Pennsylvania side,
22	and after leaving the building, follow the stream of FTC
23	people and meet at the sculpture garden, which is across
24	from the intersection of Constitution Avenue and 7th
25	Street.

4

1	Also, we must enforce our rule that there's no
2	questions or comments that come from the audience during
3	the session. Thank you.
4	We're honored today to have assembled a
5	distinguished group of panelists that have agreed to
6	offer their testimony in connection with this hearing on
7	remedies in the face of technology change.
8	Howard Shelanski is an associate dean and
9	professor of law at the University of California,
10	Berkeley, and the director of the Berkeley Center For
11	Law and Technology.
12	Renata Hesse is a partner at Wilson Sonsini
13	Goodrich and Rosati, and formerly was a chief of the
14	Networks and Technology Enforcement Section At the
15	Antitrust Division.
16	Michael Cunningham is general counsel at Red
17	Hat, Inc.
18	William Page is a Marshall M. Criser eminent
19	scholar at the University of Florida's Levin College of
20	Law.
21	And Marina Lao is a professor of law at Seton
22	Hall Law School.
23	We plan to hear from each of the speakers for
24	about 15 minutes each and then take a ten-minute break
25	and then we'll hear from the remaining speakers. We

5

1	will then have the speakers comment upon what they've
2	heard, and then have a moderated discussion among the
3	speakers with Ed and I leading the discussion.
4	Before starting, I would just like to state by
5	way of introduction that many of the product markets in
6	which the United States enjoys a comparative advantage,
7	vis-a-vis the rest of the world, are fast-changing
8	dynamic markets, including high technology markets.
9	Some critics of the antitrust laws have claimed that the
10	laws, including Section 2, are not nimble enough for
11	effective use in these types of markets. Others
12	disagree. We will explore this issue and others in this
13	session.
14	Some commentators have suggested that the
15	potential for error in antitrust enforcement may be
16	greater in these dynamic markets; however, other
17	commentators have suggested that due to network effects
18	and other possible factors, these markets may tend
19	towards monopolization to a greater agree and therefore
20	perhaps deserve particular antitrust scrutiny.
21	We are interested to learn what these panelists
22	believe about these and other issues, and their
23	implications for antitrust enforcement in Section 2
24	cases.
25	Before beginning with the speakers, my

6

1	co-moderator, Ed Eliasberg has some words about the
2	hearing.
3	MR. ELIASBERG: Thank you, Doug. I very briefly
4	on behalf of the Antitrust Division plan to welcome our
5	panelists, thank you for coming and we look -- we're
6	very much looking forward to hearing what you have to
7	say.
8	So, with that, Ed, let me turn back to you.
9	MR. HILLEBOE: Thank you, Doug. Howard
10	Shelanski is the Associate Dean and Professor of Law,
11	Boalt Hall, University of California, Berkeley and the
12	Director of the Berkeley Center for Law and Technology.
13	From 1999 to 2000, he served as chief economist of the
14	Federal Trade Commission -- Federal Communications
15	Commission, excuse me, and from 1998 to 1999, he served
16	as senior economist for the President's Council of
17	Economic Advisors At the White House.
18	Howard?
19	MR. SHELANSKI: Thanks, Doug, and I appreciate
20	the promotion. Well, I have a few main points that I
21	want to make and the points that I am going to make I
22	hope connect to what my co-panelists are going to say.
23	We had a call a week ago and I just want to set
24	up a few ideas here about the implications of the
25	implementation of remedies for monopolization in a

7

1	high-tech or technologically dynamic markets. And I
2	think my main point, my overall point would be this:
3	Remedies are hard in the best of circumstances, and I
4	think they become more complicated in technologically
5	dynamic settings, but I also think that innovation and
6	the presence of ongoing innovation in a market may
7	affect remedies in somewhat unpredictable ways, and may
8	create opportunities along with the challenges.
9	In particular, I think while innovation makes
10	structural remedies more difficult, it may in some cases
11	make conduct remedies particularly valuable. So, I
12	think while innovative markets are cause for agencies
13	and courts to be more cautious about remedies, I think
14	innovation is not cause for systematic retreat from
15	enforcement or from behavioral injunctions.
16	So, let me explain a little bit why I think this
17	is the case. You'll hear, and I think one often hears
18	that structural remedies are preferable to conduct
19	remedies or behavioral remedies in monopolization cases.
20	But, there are some caveats to this. First I would say
21	that structural remedies are not always available.
22	Where a firm is so integrated that there are not obvious
23	divisions, it's very hard to know how to implement a
24	structural remedy. Just as a classic example, the
25	District Court's second opinion in the United Shoe

8

1	machinery case would be an example.
2	The second caveat I would have is that
3	structural remedies are not always easier than conduct
4	or behavioral remedies, and in fact must often include
5	some supporting behavioral remedies, and as an example,
6	I would talk about the AT&T vertical divestiture that
7	had to be implemented by open access regulations
8	enforced by the FCC and overseen by the District Court.
9	And then, finally, I would say as a general
10	caveat, the effectiveness of structural remedies in
11	Section 2 cases is not assured and there's certainly
12	quite a bit of debate of effectiveness historically over
13	structural remedies. I'll give you a couple of
14	examples. One early quotation, "In administering the
15	antitrust acts, a number of great and powerful defenses
16	against them have been dissolved. So far as is possible
17	to judge the consuming public has not yet greatly
18	profited by their dissolution." That's Judge Rose in
19	United States against American Can in 1916.
20	Okay, now, we haven't had a lot of experience in
21	enforcing Section 2 by 1916, so maybe things have
22	changed, at least some people disagree. Bob Crandell in
23	2003 writes, divestitures are "costly exercises in
24	futility," but I would point you to the excellent work
25	of John Baker and Greg Werden in 2003 providing some

9

1	counter arguments. Just a way of saying effective
2	remedies structurally offer no guarantee of success.
3	Now, I think the structural remedies may
4	actually be even harder in technologically dynamic
5	markets, and let me offer a couple of reasons. First,
6	where a firm or industry is driven by R&D, it may do no
7	good to divest a given division or to leave a company in
8	two without sending the R&D operations with the divested
9	portions of the entity, but R&D operations are often,
10	perhaps even likely, to be more integrated and
11	inter-dependent within the firm and not susceptible to
12	clean lines of separation.
13	The second reason why I think the presence of
14	ongoing technological change may make structural
15	remedies difficult is that even if divestiture is
16	possible, high-tech firms may require more monitoring of
17	conduct during after the divestiture, because key assets
18	in such divestiture are likely to be intellectual
19	property, IP that in some cases may provide joint uses,
20	uses across the lines of the new or divested entities,
21	disputes are likely to be offered over what items to
22	transfer and whether all IP has been disclosed to the
23	new entity.
24	Moreover, because of the cooperative nature of
25	research and development, and in production, in markets

10

1	where product life cycles are short, some post
2	divestiture monitoring of relationships between newly
3	distinct entities may be needed because there may be a
4	natural incentive to favor each other as business
5	partners, and that was something that came up in the
6	wake of the AT&T divestiture, for example.
7	The third reason I think that fast technological
8	change renders structural remedies more challenging is
9	that firm and market structure may be less of an issue,
10	in some technologically dynamic markets. To the extent
11	that the so-called Schumpeterian School is correct, that
12	dynamic markets often display competition that occurs
13	sequentially, through periodic waves of creative
14	destruction, rather than concurrently, through
15	simultaneous production, divestitures may be less
16	effective or necessary such markets, although this is
17	probably more true for horizontal than for vertical
18	divestitures.
19	Okay, and my final reason that structural
20	remedies are tough in technologically dynamic markets,
21	is that where network effects are at issue, structural
22	issues might harm consumers by dissipating positive
23	network externalities. The fact that it might have been
24	better not to have monopoly in the first place does not
25	always mean it is better to break up the monopoly later,

11

1	and if such divestitures are to preserve network
2	externalities, they may have to be accompanied by
3	conduct remedies related to interconnection and
4	interoperability, doing away with those clean properties
5	of structural remedies.
6	Okay, let me turn now to conduct remedies, talk
7	a little bit about how they might work in high-tech
8	markets. As a general matter, we often hear that
9	conduct remedies are difficult, but there are some
10	caveats here as well. Not all conduct remedies are
11	created equal, and as many people have pointed out,
12	negative prohibitions, thou shalt not have exclusive
13	deals, for example, are probably easier to implement
14	than affirmative obligations, thou shall deal with your
15	rivals. In part because the negative prohibitions
16	entail less involvement of courts or agencies in
17	regulating terms of trade.
18	The second caveat that I would add is that
19	conduct remedies can have beneficial prospective impact,
20	even if they cannot roll back illegally accumulated or
21	prolonged market power. Some people say, look, conduct
22	remedies are closing the barn doors after the cows are
23	out, but if there are still some cows inside the barn,
24	it's not a bad idea to shut the door.
25	Third, even if a conduct remedy is ineffective

12

1	or weak in a given case, I think conduct remedies can
2	have important deterrent effects on others contemplating
3	the illegal behavior, and it's -- in a point that's
4	often made, some people say, if you can't be sure that
5	your conduct remedy is going to be effective, why bring
6	the case? Another reason to bring the case beyond
7	deterrence is I think as we get more experience with
8	different kinds of conduct, it can become clearer what
9	is good and what is bad, and it enables agencies to move
10	more quickly in subsequent cases, and perhaps get a
11	remedy implemented while the harm is still able to be --
12	to be nipped in the bud, so I would not let lack of a
13	clearly successful conduct remedy -- I think one needs
14	to be clearly articulable at the start of a case, but if
15	you can't be sure it will be implemented in time or it
16	will be successful in remedying the market power, there
17	may be some reasons to go ahead with the case anyway in
18	terms of establishing precedent and creating deterrence
19	effects.
20	And finally, just an observation, I think that
21	the effectiveness of conduct remedies are likely to --
22	the effectiveness is likely to be tied to the precision
23	with which one can define the cause of anticompetitive
24	harm, and in some cases, this can be done quite clearly,
25	and in those cases, I think behavioral injunctions can

13

1	be quite effective.
2	So, the overall lesson about conduct remedies, I
3	think that it is right to be weary of behavioral
4	remedies, particularly those in which the enjoined
5	conduct has ambiguous welfare effects, or in which
6	courts or agencies will have to become involved that
7	were doing terms of trade, but in the right context,
8	conduct remedies can work and can send valuable
9	deterrent signals.
10	I would just say that inability to articulate a
11	structural remedy therefore should not be decisive in
12	whether or not to prosecute an argument that is
13	sometimes heard.
14	Okay. Well, I think that technologically
15	dynamic markets create both challenges and opportunities
16	for implementing conduct remedies. The first challenge
17	is this: If one accepts that remedies may deter
18	marginal innovation, and I'll assume for the moment that
19	all innovation is good, because private returns are less
20	than social returns to innovation. Let's just take that
21	as a working assumption, it need not be true in all
22	cases, but if one accepts that, and one accepts that
23	remedies can marginally deter innovation, then the
24	deterrence risk and the costs of such deterrence may be
25	much greater in dynamic markets. It needn't be the

14

1	case, but I think innovation deterrence becomes a more
2	salient issue and a more salient concern in
3	technologically dynamic markets.
4	The second challenge is that in fast-changing
5	markets, it is more likely than it is in more static
6	settings that the conduct at issue in the case will be
7	moot by the time antitrust liability is established.
8	And in such cases, neither conduct nor structural
9	remedies are likely to be effective, and perhaps
10	something else like disgorgement might be called for if
11	such a remedy can be created.
12	But there are also opportunities in high
13	technology settings, I think, for conduct remedies to be
14	particularly effective. In some cases, technological
15	dynamics can render conduct remedies effective where
16	they would not be in more static markets.
17	In some cases, monopoly once obtain may not be
18	easily eroded, even if exclusionary or predatory conduct
19	that contributed to that monopoly is stopped. Whether
20	because of brand recognition, economies of scale, or
21	customer switching costs, new entrants will be slow to
22	appear or succeed, even when other barriers to entry,
23	such as the exclusionary or predatory conduct at issue
24	in the case, even when those barriers are eliminated,
25	you might not see competition arising.

15

1	But I think where competition is more innovation
2	based and where product life cycles are short, an
3	injunction against the behavior that led to the
4	establishment or maintenance of monopoly power may prove
5	very effective, as it is the latter set of barriers,
6	rather than any brand or economic advantage, that might
7	have kept the incumbent dominant.
8	As new waves of innovation come forward, how did
9	they stop someone else from being the innovator who came
10	in with the new product? Well, through the exclusionary
11	or predatory conduct, and branded here and switching
12	costs, other things like that, may be very, very
13	different in the high-tech environment. So, merely
14	eliminating the harmful conduct may open the door for
15	new entry and the conduct or remedy, particularly
16	negative injunctions, I think, can be very successful
17	and very helpful.
18	I would like to just raise an additional point
19	about the overall question of whether or not the cycles
20	of innovation move so quickly and the innovation process
21	moves in such different a way from the standard
22	competitive process that we should step back generally
23	from antitrust enforcement, and this is an argument that
24	one hears quite often.
25	I think when one looks at the kinds of behavior

16

1	that limit innovation, and that stop people -- that stop
2	competitors from innovating, it's very unclear to me
3	whether or not monopoly has anything particular to
4	recognize it, nor is it clear to me that new waves of
5	innovation are always going to be sufficiently powerful
6	to overcome artificial barriers to entry like
7	exclusionary -- exclusionary kinds of behavior like
8	exclusive deals when it is a monopolist that has that
9	exclusive deal, contractual terms that bar competitors'
10	products from ever being used, tying that prevents
11	consumers from ever having access to products.
12	It's unclear to me no innovation will always be
13	so great that it can overcome those barriers, those
14	barriers can lead to slower product life cycles, and
15	greatly harm consumers, and I think that there's a lot
16	of evidence of benefits from antitrust enforcement in
17	high-tech areas. And when one looks at the studies that
18	have said there are no benefits to Section 2
19	enforcement, or in a more nuance way, no benefits to
20	Section 2 enforcement in technologically dynamic
21	markets, there's a counterfactual, all of these papers
22	acknowledge the counterfactual, and we can't tell what
23	would have happened absent the antitrust enforcement, we
24	can't tell what would have happened in other markets had
25	there been antitrust enforcement, and then those

17

1	arguments are sort of dismissed, tucked under the
2	carpet.
3	I wouldn't dismiss them so easily. And, so, my
4	overall argument would be, be very cautious, be very
5	case-by-case in the application of Section 2 remedies in
6	high-tech markets, I think structural remedies are
7	likely to be harder to implement, but there may be good
8	opportunities for conduct remedies to be very effective.
9	Thanks.
10	(Applause.)
11	MR. HILLEBOE: Thank you very much, Howard. Our
12	next speaker, excuse me, is Renata Hesse, who is a
13	partner at Wilson Sonsini Goodrich and Rosati. Prior to
14	joining Wilson Sonsini, Renata served as the chief of
15	the Networks and Technology Enforcement Section at the
16	Antitrust Division and oversaw much of the division's
17	technology litigation, including the Oracle/Peoplesoft
18	and First Data/Concord matters. In addition, Renata
19	worked extensively on both the American Airlines and the
20	Microsoft case.
21	Renata?
22	MS. HESSE: Getting myself around is a little
23	harder these days.
24	So, Howard covered a lot of ground which I think
25	fundamentally I agree with almost everything he said.

18

1	In fact, I think I probably agree with everything he
2	said, but wanted to pick up where he was leaving off,
3	which was I think in talking about the notion that you
4	shouldn't back away from Section 2 enforcement in high
5	technology markets, and the main reason why I think
6	that's true is that despite all of the innovation and
7	the fast pace of change in those markets, there is an
8	opportunity for durable market power to exist in them,
9	and you do want to make sure that you're not overlooking
10	that possibility and potentially addressing it.
11	So, I wanted to start with just a few basic
12	points about Section 2 remedies that I think are
13	important, and some of these overlap with some of the
14	things that Howard said and I'm sure that will happen as
15	we go along down the line of speakers, but the first
16	thing that I wanted to talk about is the importance of
17	focusing on remedy early, and the main reason -- there
18	are several reasons for that, but the biggest reason is
19	that it helps you try to figure out what your goal is.
20	What's the violation that you're really thinking about,
21	what do you think has really happened that's harmful,
22	and how can you address it? That isn't to say that if
23	you can't come up with a perfect solution to the problem
24	that you shouldn't go ahead and try and do something
25	about it.

19

1	I think Howard is right that there's a good
2	deterrent effect in enforcing the law, even if you're
3	not 100 percent sure that the way that you think you can
4	fix it will be successful, but I do think it will -- it
5	helps you focus your investigation, and here again, I'm
6	speaking as if I were a government lawyer, but focus
7	your investigation and theories so that you can really
8	figure out whether or not you've got a case that is
9	worth allocating resources to, and pursuing.
10	And I just think it gives you a much better
11	sense of the definition of the harm that you're trying
12	to alleviate.
13	The second point is that I think when you start
14	with thinking about remedy, or at least you think about
15	remedy relatively early in the process, you can get a
16	better sense for whether or not you actually can come up
17	with a remedy that is really going to leave the
18	marketplace in a better place than it was when you
19	started.
20	And I would sort of call this the first do no
21	harm rule, and it is one of these things which you
22	always need to bear in mind, which is that you don't
23	always want to make things worse, you don't want to
24	deter innovation or take an action in the marketplace
25	which stifles productivity, and I think in technology

20

1	markets, that's something that you really need to keep
2	in mind.
3	But if you were stepping back and thinking about
4	that early, you can think about whether or not there are
5	ways to achieve the goal that you want to achieve
6	without having at least a large countervailing harmful
7	effect.
8	The third point is related to the resource
9	allocation point that I made. I think fundamentally
10	it's just a basic responsibility that particularly
11	government enforcers have to think about how you're
12	going to fix the problem, and whether or not the problem
13	is subject to a fix that's worth the investment of
14	resources in not only the investigation and prosecution
15	of the matter, but also the compliance and enforcement
16	activities that will happen post judgment, and those
17	are, I think, much more complicated when you're talking
18	about conduct remedies and structural remedies, but,
19	again, Howard correctly notes that when you do a
20	structural remedy in these markets, very often there are
21	going to be conduct remedies associated with it in any
22	event.
23	But I think you really do want to have in your
24	mind whether or not the consumption of the resource is
25	likely to result in some improvement to the competitive

21

1	conditions in the marketplace.
2	And then there's a fourth point which is that
3	sort of the question of if you have a good idea of what
4	you think the remedy that you want to put into place is,
5	then I think you'll have a better idea of whether or not
6	the -- again, the pursuit of the investigation or
7	prosecution is worth while, and by that I mean that
8	there are some kinds of Section 2 violations that are
9	easier to remedy than others.
10	So, one example might be you can think of
11	exclusive dealing or vertical foreclosure, for example,
12	where you have fairly easily identifiable concrete types
13	of conduct that you can undo. I think monopoly
14	maintenance, to a certain degree, monopoly acquisition
15	cases are much harder.
16	So, if you're in the situation where you're
17	balancing these things out, and you've got a choice
18	between two matters that you want to devote your
19	resources to and one of them has a reasonably good
20	likelihood of being able to be fixed, and the other is a
21	little tougher, then you've got to figure out how to
22	allocate your resources, then you might want to think
23	about going towards the one that actually has a solution
24	that you can identify and that you think will be likely
25	to result in an improvement in the competitive

22

1	conditions.
2	And this just goes back to something that I
3	think people often think about in the context of -- of
4	the -- when you're trying to come up with a remedy, what
5	is it that you're trying to achieve, are you looking at
6	a monopoly that you believe has been illegally created
7	and are you trying to undo that, or are you looking at
8	conduct that has maintained a monopoly and are you
9	trying to restore the conditions of the competitive
10	marketplace to the pre-exclusionary conduct state? And
11	depending on which of those two things you're looking
12	at, you're going to have a pretty different, I think,
13	idea about what's the right way to go about recommending
14	the harm.
15	The second thing I wanted to talk about was just
16	the point that Howard started with, which is structural
17	remedies and the general point that generally I think
18	structural remedies should be preferred. I think it's
19	clearly true that they are not always possible, and
20	that's certainly more true in Section 2 cases than in
21	other kinds of cases, but I wouldn't advise sort of
22	ignoring them as possible ways of recommending harm,
23	because I think they do have a number of benefits.
24	One of the benefits is that developing a
25	functional set of conduct restrictions that are likely

23

1	to have a beneficial effect, without having this sort of
2	countervailing, potentially negative effect on the
3	marketplace is an extremely complicated and resource
4	intensive process. It took a really long time to come
5	up with the conduct restrictions that we developed in
6	the Microsoft case, and I think, you know, you can --
7	it's open for debate whether or not those were worked
8	well or not well, but it took a long time to figure them
9	out, and to just evaluate all the different
10	possibilities and try to develop language that's
11	concrete enough and understandable enough in a legal
12	document for people to actually then be able to
13	implement it and understand it and understand what the
14	rules of the road are. It's just an inherently
15	difficult process to do, and I think that isn't just
16	Microsoft, that's any time when you're trying to come up
17	with a set of conduct restrictions where you're dealing
18	with complex technology.
19	It's also hard to judge their success, I think,
20	and that's also true in structural remedies, in some
21	situations, but it's very hard to know when conduct
22	restrictions have succeeded. I think you can know when
23	they've failed, but I don't think you can know as easily
24	when they've succeeded. How do you measure success with
25	conduct restrictions?

24

1	I think structural remedies generally eliminate,
2	although not entirely, the need for ongoing enforcement
3	in compliance activity, which also can be an extremely
4	time consuming and resource intensive process. It can
5	require, and this is something else I can talk about a
6	little bit later, but it can require a lot of assistance
7	from people who know more about technology and business
8	and licensing and all these things that come up in
9	technology markets work, and structural remedies tend to
10	need a lot less of that.
11	I think structural remedies are generally less
12	easy to evade. It's pretty clear what you're supposed
13	to do, and you've either done it or you haven't done it.
14	You've either divested the plant or the asset or
15	whatever it is, or you haven't. You know, there are
16	issues associated with those kinds of things, whether or
17	not you found an adequate buyer and all of those other
18	sorts of issues, but at least there's a very clear line
19	about what you are supposed to have done.
20	I think they have a potentially greater
21	deterrent effect, because they have the capability at
22	least of really restructuring a business in a way that
23	most businesses don't want to have happen. So, that can
24	discourage people from engaging in conduct that folks
25	think violates Section 2.

25

1	And I think generally, again with some of the
2	caveats that Howard laid out, they're more likely to
3	work. The lines are clearer, and if you've actually
4	proven a violation where you can support imposition of a
5	structural remedy, I think the likelihood of that
6	structural remedy having an effect is probably higher.
7	So, those are some kind of basic points. A few
8	points that are more directly connected, just to sort of
9	the technology markets, and the first is, you know,
10	everybody always talks about technology markets are fast
11	changing and innovation changes everything, and as
12	Howard said, sometimes people say, maybe you don't need
13	to worry about them because they're just going to be
14	self correcting. I tend not to agree with that latter
15	viewpoint, for the reason that I started with, which is
16	that it's clear that there's a possibility for the
17	existence of durable market power in these markets, so I
18	think just leaving them alone and hoping that the
19	exclusionary conduct somehow magically stops and things
20	correct themselves is not likely to lead to a lot of
21	success.
22	I do think that the fact that they can sometimes
23	be slow and that the antitrust enforcement process can
24	sometimes be slow is a down side in these markets, a
25	greater down side in these markets than in other

26

1	markets, because sometimes you feel like you get to the
2	end and you're addressing the problem when it's actually
3	a little bit too late.
4	As a consequence, I think you need, when you're
5	thinking about conduct remedies in technology markets,
6	to be a little bit more flexible about how you think
7	about them. And to address categories or types of
8	conduct relating to types or categories of products or
9	services as opposed to saying, well, this -- you did
10	this particular thing with this particular kind of
11	product, and you should do that -- you shouldn't do that
12	anymore. This is the negative prohibition point versus
13	an affirmative obligation point.
14	If the conduct remedy is too narrowly focused,
15	it runs the risk of being ineffective, and I think in
16	most cases is likely to be ineffective, particularly,
17	again, if you're talking about undoing some sort of harm
18	that has occurred.
19	You know, Microsoft is a simple example of this,
20	the consent decree doesn't just talk about browsers,
21	which was the primary focus of the case, but it talks
22	about other products which were potential platform
23	threats and has some construct restrictions in it that
24	are designed to try to go after those particular -- or
25	not go after them, but to try and make sure that the

27

1	conduct relating to those other kind of potential
2	platform threats were restrained.
3	There's a possibility in technology markets that
4	they should be of shorter duration. Again, Microsoft is
5	another example, it was a five-year consent decree, it's
6	now been extended in some pieces for longer than that,
7	but I think there's a reasonable basis for at least
8	looking at the question of whether or not you really
9	need something to last ten, 20, some decrees in the past
10	have lasted for hundreds of years, some of them very
11	perpetual, and whether or not that makes sense
12	particularly in the context of technology markets is I
13	think something that people -- it's worth looking at.
14	I also think if you're going to think about
15	decrees of shorter durations, or remedies of shorter
16	durations, that including some mechanism for revisiting
17	that question before the term of the decree expires is a
18	good idea. I think it's just these markets are
19	inherently unpredictable, and given the complication of
20	structuring conduct provisions in them, that giving
21	yourself an opportunity to take a second look and having
22	a standard for how you would be able to convince a court
23	that you need to extend a decree in these kinds of
24	markets is something that should be given some
25	consideration.

28

1	And the final point on this area is that I think
2	conduct remedies in Section 2, Section 2 remedies in
3	technology markets may need to be more forward looking,
4	and this is a little slightly basically the same thing
5	with a slightly different pitch on it, but you do have
6	to think about what it is that you can predict about the
7	marketplace and changes in the marketplace going forward
8	and whether or not what you've devised in the context of
9	the conduct remedy is adequate to address the changing
10	technology in the marketplace.
11	The last piece about technology markets that I
12	think makes them different is that they're hard, and
13	it's hard to understand them, and they're particularly
14	hard for people who are not educated in technology.
15	And, so, compliance monitoring enforcement can be a
16	difficult thing to do.
17	As a consequence, I think if you're looking at
18	these markets and you're looking at behavioral
19	restrictions, particularly ones that relate to licensing
20	of intellectual property or access to technology or
21	just, you know, you're requiring a company to stop doing
22	a particular activity with a particular type of
23	technology, that you really need to anticipate getting
24	some technical help, and when I think of technical help
25	in this context, I don't think just of software

29

1	engineers or hardware engineers, but I also think of
2	licensing expertise, business expertise, you know,
3	trying to figure out whether a royalty ran is a
4	difficult problem, and it's not a problem that most
5	antitrust lawyers deal with on a day-to-day basis.
6	And having the ability to have access to people
7	who actually do that kind of work for a living, who know
8	what particular types of technologies, what kinds of
9	royalties particular types of technologies command is, I
10	think, critical to the ability to actually do an
11	adequate job of monitoring and enforcing compliance.
12	Again, I started with sort of a more broad
13	definition of technical assistance, but a narrow
14	definition of technical assistance, which is just
15	actually having somebody who knows how software code is
16	written, and what to look for and how to evaluate
17	whether or not something has been done in the code is
18	very important. I think one of the really unusual and
19	innovative things that was in the Microsoft decree was
20	the technical committee provision, which allowed the
21	Department of Justice and the states to have access to
22	basically a full-time group of technical consultants who
23	were hired to work for those people and the cost of
24	which was borne and continues to be borne by Microsoft.
25	I think it was an unusual idea, but it really

30

1	has become, I think, a key component to the United
2	States enforcement and monitoring, compliance monitoring
3	efforts of the Microsoft decree, and it was essentially
4	copied by the European Commission in the work that
5	they're doing in Microsoft as well.
6	And it had not been done before. There were
7	lots of times where in complicated markets people had
8	used monitoring trustees, I shouldn't say there were
9	lots of times, but there were examples of monitoring
10	trustees being used, usually they were in things like
11	prison condition litigation, where there was some pretty
12	complicated oversight that was needed, but hiring
13	technical experts to help out was an innovative thing to
14	do and I think has proven to be a pretty successful
15	component of the Microsoft decree.
16	Now, you also may need technical assistance when
17	you're trying to figure out whether or not somebody has
18	violated the decree and you actually want to go after
19	them for contempt. I think the Microsoft model doesn't
20	quite fit so well in that context, because it's a little
21	hard to see how you can justify the party who you're
22	going to be pursuing in contempt actually paying for the
23	expert that you're going to be using, to go after them
24	in contempt, but it's something that people -- you want
25	to think about, and at least have the resources and

31

1	capability to get that kind of help on board.
2	So, I have probably 30 seconds at this point
3	left. The last thing I would say is that licensing
4	remedies are incredibly common in technology markets.
5	They can be useful, and I think can work well, but I
6	think they work particularly well in the context where
7	you know or have a very good idea of what the
8	intellectual property is or what the asset is that needs
9	to be licensed, are there particular patents who needs
10	them, and again, if you go back at the very beginning,
11	to those are things that you can think about early on
12	and figure out and they'll help you determine whether or
13	not a licensing remedy is likely to be successful.
14	And of course when you're doing that, you need
15	to think about the policy issues that are associated
16	with compulsory licensing of intellectual property,
17	which is a hot topic these days.
18	(Applause.)
19	MR. HILLEBOE: Thank you so much, Renata, for
20	those comments.
21	Michael Cunningham is general counsel at Red
22	Hat, Inc. Prior to joining Red Hat, he served as
23	associate general counsel at IBM, where he had legal
24	advisory responsibilities for the Business Consulting
25	Services Division for Europe, the Middle East and

32

1	Africa. He was also a partner and associate general
2	counsel at PricewaterhouseCoopers.
3	Michael?
4	MR. CUNNINGHAM: Thank you, and good morning.
5	I'm pleased to have the opportunity to participate in
6	this important consideration of Section 2 remedies, to
7	do so before distinguished representatives of the
8	government, as well as with this particularly
9	knowledgeable panel.
10	I'm the general counsel of Red Hat. I'm going
11	to make a little disclaimer, I'm a technology lawyer,
12	I'm not principally an antitrust lawyer. I hope that I
13	can offer some comments, however, as an executive of a
14	technology company that are relevant to these inquiries.
15	With your indulgence, I would like to describe a
16	bit about our business that I think is relevant
17	innovation, given the debate about antitrust remedies
18	stifling innovation, I think it's particularly
19	appropriate this morning.
20	The software solutions that Red Hat offers, and
21	for which we provide services, are developed by very
22	broad horizontal communities that are without
23	geographic, organizational or political boundaries. The
24	community of innovators that unleash the value of open
25	source are not contained within Red Hat. Some of its

33

1	contributors are, but it's not.
2	The contributors include the customers and
3	vendors of hardware and software. It includes
4	academics, it includes many, many motivated individuals
5	that we call hackers, it includes persons from every
6	continent and from multiple political subdivisions.
7	The development environment is also not
8	controlled by any single individual company or political
9	entity, it is instead a free, meritocratic marketplace
10	of ideas. Individuals take these ideas and they place
11	these ideas with their individual name and reputation
12	into the marketplace in a particular software
13	development project to which their idea is relevant.
14	There are literally thousands of these projects
15	out there. In one of our offerings, Red Hat Enterprise
16	Linux, hundreds of projects are represented. These
17	ideas are then reviewed by that development community,
18	for that project, and only those ideas that can handle
19	the open scrutiny of this open source community are then
20	adopted.
21	In this way, the best ideas and the bets bits of
22	ideas bubble up. Moreover, if there happen to be a
23	serendipitous discovery that is made in one of those
24	projects that's relevant to another project or might be
25	an entirely new approach, the contributor or any other

34

1	person is free to contribute it to that project or
2	indeed to go out and start a new project to take the
3	technology in a new direction.
4	This model has produced and continues to produce
5	copious innovation. It also accelerates and multiplies
6	innovation, I would argue, by providing tools of
7	innovation, such as information ideas to a broader and
8	more diverse community than development within any one
9	firm is possible could provide.
10	The open exchange of information and ideas is an
11	innovation force multiplier. For example, sophisticated
12	business and other users of software frequently take the
13	modular pieces of well crafted software that's developed
14	in the open source community, cobble bits and pieces of
15	it together, modify it, append to it and create
16	solutions for problems that heretofore were not solved,
17	or new problems that arise in their business.
18	Similarly, the creative juices of the lone
19	teenager in North Dakota in some remote location can
20	contribute to that process, so can a Cal Tech physicist
21	who is wondering why there hasn't been a software
22	development that would help in his or her research. And
23	so are many, many others unleashed in the creative
24	process through this open development and collaboration
25	model.

35

1	The modular and open nature of open source
2	software has fueled much innovation, but it is by no
3	means limited to software. It is not a software-only
4	phenomena. No, I would submit to you that the relative
5	ubiquity and low cost of the Internet, and collaboration
6	tools like email and dedicated web sites portends for
7	joint collaboration that is unleashing all sorts of
8	innovation across the world.
9	If you've read the best selling book by Tom
10	Friedman, The World is Flat, you will get a very good
11	sense of some of these trends, I think. I would also be
12	happy to comment on some other areas where that
13	innovation is being unleashed in the questioning, if
14	that's helpful.
15	With that bit of an introduction, maybe I should
16	turn my attention now more directly to remedies. First,
17	I believe that in the software space at least, the
18	relevance of the antitrust law hangs on the issue of
19	remedies. I can think of no way as a practitioner and
20	an executive in a company in the industry to more
21	starkly illustrate that point than to disclose my actual
22	advice to my client in pursuing whether to participate
23	in or pursue any monopoly-related case, whether that be
24	in a government-related case or in private litigation.
25	I would tell my client, it's too expensive for

36

1	you to fully embrace and do that. You cannot do it.
2	You don't have enough money to pursue it, it's certainly
3	over $10 million, it will be a long time, and it is
4	likely, I would submit to you, at least this would be my
5	advice, it is likely and substantially likely that the
6	remedy that will result will be of limited utility. So,
7	therefore, those sorts of expenditures would not be
8	justified.
9	And guess what? Those that the government
10	representatives seek to regulate know this, and they
11	know it well. By way of illustration, a high-ranking
12	representative, indeed a very high-ranking
13	representative of a party found to have market power by
14	multiple international competitive authorities has
15	aggressively and indeed smugly advised Red Hat that
16	there is no competition authority in the world that this
17	firm will not outspend, outlast, and seek to thwart.
18	In short, the system seems broken in terms of
19	speed, cost, and effectiveness of remedies, at least
20	from my little corner of the world. You know, why is
21	this the case? Well, as others have said, technological
22	change is very rapid and litigation is not. The rate of
23	change at least in information technology is in very
24	short cycles, three to five years, maybe six to eight
25	years, certainly not longer than that in many, many

37

1	areas of information technology.
2	Remedies that only address a particular market
3	complained of, and established at great expense, will
4	often be too late to provide meaningful relief. A
5	remedy focused on future conduct would address some of
6	those limitations and in many instances I think is
7	necessary.
8	I also am intrigued by the idea of smaller
9	simpler cases with speedier trial times that would focus
10	on future contact to make the law more relevant.
11	Clearly cost and delay undermine the perceived and
12	actual effectiveness of the antitrust laws in our
13	competitive zone.
14	In that way, some of Professor's Lao's writing
15	on the role of the intent in finding liability seem a
16	fruitful avenue for further inquiry to me.
17	Second, technology can be manipulated. The
18	speed with which information technology moves and can be
19	molded provides real opportunity for conscious
20	manipulation by the monopolist away from the market
21	complained of. The government enforcement actions
22	against Microsoft are an example of the timing
23	challenges, I'm thinking now about the European Union,
24	even the most aggressive threats by the EC are mired in
25	delay, seemingly extended without limit.

38

1	According to the most recent statistics we've
2	seen, Microsoft continues to gain in the operating
3	system worker group server market, meanwhile the market
4	continues its very rapid evolution, probably reducing
5	the relevance of any remedy that may eventually be
6	enforced and/or issued.
7	I guess I should also point out that private
8	enforcement actions have not solved the problem either,
9	this won't be a surprise from my earlier comment. The
10	antitrust law, like the Ritz Carlton, is open to the
11	rich and poor alike. The most entrepreneurial and the
12	most innovative firms, the small fledgling ones are
13	without means to mount private antitrust cases.
14	Let me turn my attention for a few moments to
15	innovation. Protecting competition does not mean
16	stifling innovation, I don't believe. While there is an
17	inevitable tension between the intellectual property law
18	and the antitrust law, competition law cannot achieve
19	its purpose if regulators and courts are preoccupied
20	with a concern that remedies affecting some intellectual
21	property rights will necessarily stifle innovation.
22	That focus on IP, that is intellectual property,
23	a legal concept, is misguided. The focus should be on
24	true innovation, not patents and copyrights, public
25	grants of a monopoly.

39

1	Why is that the case? Well, first I think
2	equating innovation to the accumulation of intellectual
3	property is suspect, at least in the software world.
4	The software patent approach in the United States is
5	being broadly questioned, and that's the case for at
6	least two or three different reasons.
7	First of all, the software industry in
8	particular survived for almost 20 years with very
9	limited forms of software patents, not the broad range
10	that we now see following State Street and other court
11	decisions.
12	Second, I would submit to you the relationship
13	of software patents to innovation is suspect. I
14	regularly review the academic literature in this area
15	and I am aware of no convincing argument that software
16	patents have unleashed -- and no empirical study --
17	that they have unleashed and spurred additional
18	innovation.
19	Third, the news is regularly filled with stories
20	of highly suspect software patents, patents that are not
21	new and innovative, ones that are anticipated by prior
22	art and ones that common sense tell us lack sufficient
23	novelty to warrant 20 years of protection.
24	Of course that shouldn't be surprising, there
25	are well publicized challenges in the Patent & Trademark

40

1	Office, there's no effective and searchable database on
2	prior art for software. There's also serious challenges
3	in retracting and retaining the kinds of experts that
4	Renata talked about to actually evaluate what is seeking
5	to be patented.
6	I say that just to suggest that the innovation
7	reflected in software patents is questionable at times.
8	Therefore, giving, you know, complete deference to
9	intellectual property in that context seems misguided.
10	Even more important to this debate, as my
11	opening remarks sought to illustrate, there are broad
12	communities of collaboration that are massively
13	innovative. Please note that their style of
14	collaboration is not readily or naturally susceptible to
15	patent protection, given the open and collaborative
16	nature of their exchanges.
17	Thus, innovation of the firm is not the only or
18	even the most effective form of innovation to be
19	considered or protected when facing the market
20	disruptive effects of monopolists. Powerful new
21	innovation paradigms are upon us now and they're growing
22	and they need to be considered and measured in balance.
23	But even if we were to assume that the firm is
24	the epicenter of innovation, the smallest and perhaps
25	most innovative are without the means to challenge the

41

1	innovation of the monopolist that is purported to be
2	reflected in intellectual property. The combination of
3	suspect software patent quality and the disparity of the
4	cost to acquire a patent versus the cost to defend
5	against it skew IP protection in favor of larger
6	enterprises with market power.
7	Cost of acquiring a patent, let's say, is
8	$25,000 to $35,000. It absolutely pales in contrast to
9	the cost of a proper infringement defense. That is
10	variously $3 to $5 to $7 million, and by all accounts is
11	growing at present.
12	Moreover, the monopolist can disrupt the
13	business of smaller competitors merely by suggesting to
14	consumers that its IP is infringed, without any proof
15	whatsoever. If you consider Steven Bommer's recent
16	statements that the users of Linux have an undisclosed
17	off balance sheet liability to Microsoft, which were
18	offered without any substantiation whatsoever. And the
19	SCO litigation that is ongoing I think offers some
20	interesting and vicarious variance on the same theme,
21	which I would also be happy to comment on in the
22	question and answer period.
23	Keeping on the intellectual property theme, an
24	effective remedy needs to prevent the extension of
25	market power. A company who has acquired market power

42

1	through anticompetitive conduct shall not be permitted
2	to be able to hide behind intellectual property
3	protection to reinforce and extend its market power. I
4	think there is an interesting lesson in history on this
5	that deals with data formats.
6	In particular, I would like to contrast how
7	Microsoft came to compete in word processing, versus how
8	it now competes. The background is as follows:
9	Software products manipulate and ultimately store
10	customer data after that manipulation. To the extent
11	this data is then placed into storage formats, that are
12	claimed as either proprietary or protected by
13	intellectual property of the software vendor, then the
14	ability of a competing product to make effective use of
15	the stored customer data and break into and compete in
16	that market, which is likely reinforced by very strong
17	network effects, can be precluded.
18	Take, for example, Microsoft's word processor
19	competition against the then-important market position
20	of the WordPerfect product in the 1980s. Because the
21	data format's inability to represent the data with
22	substantial fidelity was possible, Microsoft could
23	compete at the enterprise level by saying, give me a try
24	in parallel with WordPerfect. If I do better, then
25	incur the cost of switching out your old technology and

43

1	taking on our technology.
2	In contrast today, I would submit to you the
3	formats of Microsoft alphus data have been and are
4	increasingly being obscured by Microsoft and cannot be
5	presented, that is the data cannot be presented with
6	true fidelity by any competitor, like OpenOffice, which
7	thereby extends the time of their dominant position and
8	permits extension of power into adjacent markets.
9	It is the case that Red Hat cannot effectively
10	compete with open source personal productivity
11	applications, like word processors and other things, at
12	the enterprise level against Microsoft, it can't get its
13	foot in the door. If a client wants to give someone a
14	try and you can't render their existing data in a
15	meaningful fashion, that prevents anyone from entering
16	into that market, I would submit to you, or doing so
17	easily, anyway.
18	Microsoft controls, I would submit to you, a
19	facility of competition through the extension of IP and
20	proprietary formats that is needed to meaningfully
21	render and manipulate customer data. I have no doubt
22	that's why you're seeing states like Massachusetts
23	aggressively consider the open document format, a truly
24	open standard in format in its procurement processes.
25	The mono type litigation of Red Hat is another

44

1	example that illustrates that that I would be happy to
2	comment on later.
3	In summary, I guess I would say that innovation
4	does not equate to intellectual property, and therefore
5	greater focus on preserving and promoting true
6	innovation in the marketplace is warranted. Further,
7	there are numerous ways in which the use and assertion
8	of intellectual property rights can be a pretext that
9	chills competition and extends monopoly power.
10	Thank you.
11	(Applause.)
12	MR. HILLEBOE: Thank you very much, Michael, for
13	that, and I think we will take about a ten-minute break
14	now.
15	(Whereupon, there was a recess in the
16	proceedings.)
17	MR. HILLEBOE: Thank you, everyone. William
18	Page is a Marshall M. Criser eminent scholar at the
19	University of Florida Levin College of Law and he is
20	also an alumnus of the Antitrust Division, where he
21	served as a trial attorney in the 1970s.
22	Bill?
23	MR. PAGE: Thank you. Rather than speak in
24	generalities about Section 2 remedies in high-tech
25	markets, I want to zero in on one highly technical and

45

1	seemingly obscure provision in the final judgments in
2	the government's Microsoft case that has turned out to
3	be the most difficult and the most problematic in its
4	enforcement.
5	The provision requires Microsoft to license to
6	software developers communications protocols that
7	Microsoft uses in its Windows Client operating systems
8	to interoperate with Microsoft server operating systems,
9	either in corporate networks or over the Internet.
10	Communications protocols are the rules for transmitting
11	information between different devices.
12	So, in a computer network, the protocols allow a
13	user of a client computer, for example, to store
14	information on a network drive or send an email or
15	display a web page, among many other things.
16	This sort of interoperation is relatively easy
17	when the client computer's operating system and the
18	server operating system share a common base in code.
19	It's like they speak the same language, so they can
20	interoperate easily.
21	Where the client computer, usually a Windows
22	client, has to interoperate with servers from other
23	vendors, then the problem with interoperability becomes
24	much more difficult, but there are ways of solving them.
25	There are recognized ways of solving them. Some involve

46

1	installing a client on Windows that would allow
2	interoperation with the non-Windows server and
3	applications running on it.
4	There are also standard protocols that are
5	available and supported in Windows. This provision
6	requires another way of assuring interoperation, that is
7	requires Microsoft to disclose its proprietary
8	protocols, to license them to software developers so
9	that they can interoperate. The near-term goal would be
10	for them to be able to write programs that will
11	interoperate as well with Windows clients as
12	applications running on Microsoft servers.
13	The long-term goal is to allow -- is to preserve
14	in this network context the so-called middleware threat
15	that was the focus of the government case. The
16	middleware applications running on servers, the concern
17	is, may eventually evolve into platforms that could
18	rival the Windows desktop and thereby erode the
19	application's barrier to entry. Essentially the theory
20	of the government case.
21	In spite of its apparent obscurity, this
22	provision has been given an unusual amount of importance
23	by the District Court enforcing the Microsoft judgment.
24	She's referred to it as the most forward looking
25	provision in the final judgments and as necessary to

47

1	assure that the other provisions don't become
2	prematurely obsolete. It's now being implemented by the
3	two sets of plaintiffs in the Microsoft litigation, the
4	Antitrust Division and the nine settling states, and
5	also by the group of non-settling plaintiffs who were
6	awarded essentially the same relief, but there are
7	different enforcement mechanisms.
8	There's the technical committee that Renata
9	referred to in the Antitrust Divisions's consent decree
10	and there's a technical consultant to the non-settling
11	states under their decree, but they're coordinating
12	their enforcement efforts. Both of these judgments went
13	into effect in 2002.
14	And the plaintiffs in both cases and Microsoft
15	has been filing status reports every two months about
16	the enforcement of both of the judgments, and I have
17	studied these reports with the help of a research
18	assistant, who was also a software developer and a
19	management consultant, and so he has been sort of my
20	technical consultant. He provided all of the technical
21	expertise in this study, because I certainly claim none.
22	The enforcement of this provision, this one
23	provision in these judgments has dominated these
24	reports, particularly in recent years. It by far
25	occupies most of the reports and certainly most of the

48

1	time of the technical committee. And I'll argue that
2	this provision has not accomplished its purpose, and
3	that we can draw some lessons from that experience.
4	So, I want to first describe what I take to be
5	the principles of Section 2 remedies, I'll then suggest
6	that most of the provisions in the Microsoft judgments
7	adhere to these principles, but that this provision, the
8	protocol licensing provision, departs from the
9	principles and that is part of the reason why it has not
10	been successful.
11	I'll describe briefly how it has been
12	implemented and then in the end I'll try to draw some
13	lessons. And incidentally, this is a very brief summary
14	of a much longer article which I hope to post on SSRN
15	shortly.
16	The goals of Section 2 remedies should be to
17	restore competitive conditions that would have existed
18	but for the illegal conduct. They should not be to try
19	to restore or to create some sort of ideal competitive
20	condition or to supervise market outcomes. I take the
21	primary antitrust remedy to be deterrence, through fines
22	and covered damages. If deterrence can be effective, if
23	an optimal penalty can be imposed, that's always going
24	to be preferable to having an administrative structure
25	imposing remedies. It's simply the direct costs of

49

1	imposing those remedies will be -- will impose a greater
2	cost than effective deterrence.
3	Assuming that some sort of injunctive relief is
4	required, I would suggest that injunctions should be
5	limited to preventing reoccurrence of proven
6	anticompetitive behavior. The Sherman Act, unlike
7	sector-specific regulation, I believe reflects the
8	assumption that if specific impediments to competition
9	are removed, then private contracting within the market
10	will lead to the efficient outcome. And if that would
11	not be the case, then that would argue that the market
12	should be regulated.
13	Beyond that, I would suggest that injunctions
14	are problematic. First, divestiture, at least in the
15	case of a unitary company, should be a last resort,
16	primarily appropriate to dissolve recent combinations.
17	Regulatory decrees also, as many have observed, should
18	be avoided. As the Supreme Court said in Trinko, they
19	require antitrust courts to act as central planners,
20	identify improper price policy and other terms of
21	dealing in roles for which they are well suited.
22	Most of the Microsoft final judgment provisions
23	reflect these principles. They do not require any form
24	of divestiture, and most provisions respond more or less
25	directly to the liability holdings in the case that were

50

1	affirmed by the D.C. Circuit in 2001, prohibiting
2	retaliation against computer manufacturers for promoting
3	rival software, requiring uniform licensing terms,
4	giving computer manufacturers the flexibility to remove
5	the visible means of access to Microsoft middleware
6	products and so forth.
7	The protocol licensing provision does not
8	respond directly to any illegal conduct. Server-based
9	applications were mentioned in the findings of fact,
10	only to exclude them from the market.
11	Interoperability in networks was not an issue in
12	the case, and in fact developing and refusing to license
13	incompatible proprietary software was not held illegal,
14	in fact, it was specifically held to be legal, if
15	nothing more than that were shown.
16	So, where did this come from? The idea for this
17	provision actually arose, according to Ken Alletta's
18	book on the Microsoft litigation, after the findings of
19	fact had been issued. In other words, after the record
20	was closed in the case. The feeling was that Microsoft
21	essentially was not going to continue the conduct that
22	was actually the subject of the litigation, the browser
23	wars were over, Microsoft had already stopped the
24	discriminatory pricing, it had gotten rid of the
25	exclusive terms in its contracts, so we needed to be

51

1	more forward looking and what was forward was this
2	network environment.
3	The fear was that in this -- you've got to, you
4	know, as the computer market moved toward networks, both
5	local corporate networks and the Internet, it was
6	necessary to assure that Microsoft would not
7	discriminate in allowing rivals to interoperate with the
8	dominant Windows client.
9	And, so, various proposals for various
10	interfaces by Microsoft were made. After the original
11	judgment was reversed, of course the Antitrust Division
12	reached an agreement with Microsoft on the consent
13	decree and it included a version of this. The protocol
14	licensing provision, which essentially we now have, in
15	both that consent decree and in this -- the states'
16	judgment.
17	Judge Kollar-Kotelly approved this provision,
18	even though she recognized that the government was not
19	strictly entitled to it, because it was not responsive
20	to proven illegality, and she also recognized that there
21	were these other ways in networks of achieving
22	interoperability besides requiring Microsoft to license
23	its proprietary protocols.
24	Nevertheless, she found that -- and here's the
25	key language, it's closely connected to the theory of

52

1	liability in this case, and furthers efforts to prevent
2	future monopolization.
3	So, under this program, Microsoft has developed
4	the Microsoft communications protocol program, which is
5	an extension of its Microsoft developers network, and
6	under this program, it offers a license to these
7	protocols, and technical documentation. In the initial
8	response in August 2002, actually before the consent
9	decree was approved, but nine months after it was
10	originally agreed to by the parties, Microsoft produced
11	5,000 pages of technical information, documentation, on
12	the protocols, which it reported with a product of the
13	work of five technical writers working essentially
14	full-time for nine months.
15	By July 2003, however, eight months after the
16	entry of the final judgments, only four developers had
17	licensed these protocols. And Judge Kollar-Kotelly told
18	the parties in a status conference, this is reported in
19	the report, that she was very, very concerned that
20	nobody was taking these licenses. And both Microsoft
21	and the government responded to this by various efforts
22	to promote them. Microsoft took out ads, they
23	evangelized these protocols, but with very little
24	success. And finally the government conducted a survey
25	of developers asking them why aren't you licensing this

53

1	material, and they gave a list of reasons, some of which
2	focused on the license itself, said it was way too
3	complicated, it was pages of technical terms, and they
4	were too expensive, the technical documentation was
5	insufficient, the royalty was too high, whatever. But
6	some said, we just don't need them for our development
7	efforts.
8	All of these, except that last one, were
9	addressed over the next three years. The license term
10	has been extended, the limitations in it have been
11	relaxed, and simplified, the royalties have been
12	reduced, many of the open standard protocols that
13	Microsoft supports have been made available under the
14	royalty free license. Microsoft has made its source
15	code available to licensees.
16	Now, to become a licensee, you need to show you
17	have a legitimate purpose. So, you can't go and ask to
18	see the source code, but if you are a licensee and you
19	can show that you have need for it, under the license,
20	then they'll show it to you and they'll actually provide
21	support to show you how to use it. It's also provided
22	500 hours of free premier technical support, it's
23	provided a dedicated account manager, it's provided
24	three-day, what they call plug fests, where you can
25	bring your product and test it and Microsoft engineers

54

1	will work with you to try to make sure it interoperates
2	well with Windows. It's created an interoperability
3	lab, and I should mention, when we had the first plug
4	fest, only two licensees signed up for it, no one has so
5	far signed up for the interoperability lab.
6	So, over the years, what's most dramatic about
7	these status reports is the accounts of how Microsoft
8	and the technical committee have tried to improve the
9	technical documentation of the protocols.
10	In July 2004, the technical committee and
11	Microsoft agreed on a 40-page specification that the
12	documentation was supposed to meet. And the technical
13	committee undertook to develop what it calls prototype
14	implementations of each protocol. There are about 100
15	and 120 protocols, and in order to assure that the
16	documentation of them was sufficient, the technical
17	committee has undertaken to try to actually write a
18	little application using the protocol.
19	And, so, if they could do that, then that would
20	show that the documentation, it could actually be put
21	into effect by the developer. Where they run into
22	problems, if they ran into problems, they treated that
23	as an issue, and they reported that to Microsoft as a
24	bug to be addressed, and depending on its importance,
25	they gave them seven days or, you know, longer time

55

1	limits to respond to it.
2	And this was the approach for about a year, but
3	by early 2006, the technical committee had reported to
4	Microsoft about a thousand of these issues, and only
5	about 300 of them -- 300 of them had been resolved, and
6	in May, this is about a year ago, the plaintiffs
7	reported to the judge that the project had reached what
8	it called a watershed, and at that point, someone who I
9	take to be a strong personality, Robert Muglia, who is
10	the senior vice president of Microsoft and formerly was
11	the head of server division, reviewed this program and
12	said that this process of trying to respond to bugs one
13	by one, as they're reported by the technical committee,
14	was just not working, and that we would need to start
15	from scratch and rewrite all of the technical
16	documentation.
17	And, so, last summer, incidentally, it was at
18	this point that the technical committee made contact
19	with the European Commission's monitoring trustee, which
20	is also administering an order to Microsoft to disclose
21	protocols, and in connection with those communications
22	had with Microsoft, agreed on a new overarching
23	specification. This is now the third standard that will
24	be used to judge the documentation.
25	And Microsoft was given a new set of milestones,

56

1	time tables, to complete the project. At this point, it
2	was clear that the decrees were due to expire in the
3	fall, and it was pretty clear that that was not going to
4	be enough time to do all of this, and so that's when the
5	parties agreed to extend the term of the judgment for up
6	to five years.
7	Meanwhile, Microsoft has suspended royalty
8	payments entirely for its licensees, until the
9	documentation is deemed to be sufficient, and the
10	technical committee has continued to develop these
11	protocol implementations, and interestingly, Microsoft
12	has also undertaken to do something similar, developing
13	what they call test suites, which it's one of the
14	practices of software developers when they're working on
15	an application, they come up with suites of testing
16	applications to see if they work, and Microsoft has
17	undertaken sort of a parallel or duplicate testing
18	mechanism.
19	And in this most recent status report, which was
20	issued earlier this month, the plaintiffs reported that
21	although they've had some questions about Microsoft --
22	apparently Microsoft discovered some new protocols that
23	they hadn't identified before, they said that this new
24	documentation is looking better, although significant
25	additional work needed to be done.

57

1	So, Microsoft now has been -- remember the first
2	project, it had a few technical writers working for a
3	certain number of months to produce these 5,000 pages.
4	They now have 313 employees working on this project.
5	And the technical committee also has increased its staff
6	to 40 engineers, and they now have offices both in
7	Redmond, Washington and in Silicon Valley.
8	The bottom line, as of this month's status
9	report, of the thousands of developers writing
10	applications for servers, for server operators, to run
11	on server operating systems, only 27 firms have taken
12	the royalty-based license, and all but four of these,
13	but for very specific purposes, like media streaming or
14	data storage or security, the proxy firewall segment.
15	So, and of those 27, only 14 are producing any products.
16	And none of these products seems likely to have any
17	potential as a platform.
18	So, what are the lessons from this experience?
19	The original rationale for this project was to preserve
20	the middleware threat to the Microsoft monopoly in the
21	network environment. If so, at least so far, the
22	project has not succeeded, because it's attracted very
23	few licensees, despite these enormous efforts, and I
24	think quite admirable, and impressive efforts on both
25	sides.

58

1	What this suggests to me is that the primary
2	reason why we're not seeing more licensees is that
3	licensing Microsoft's proprietary protocols is generally
4	not necessary for these firms to develop software
5	applications to run on non-Microsoft servers. They can
6	use the standard protocols that Microsoft supports in
7	Windows, or they can develop their own windows client
8	which then could run on the Windows client and
9	communicate directly through Microsoft's application
10	programming interfaces.
11	So, to boil it down, what I would say is that
12	what this remedy does is to treat the Microsoft
13	protocols as if they were an essential facility, except
14	that they're not essential. There are other ways of
15	accomplishing the same thing.
16	So, what I would take to be the two primary
17	lessons are first, injunctive relief, particularly in
18	high technology markets, should be limited to responding
19	to a proven need, and the most important proven need is
20	to -- is to interdict and remove anticompetitive
21	practices, proven anticompetitive practices.
22	So, if Microsoft is proven to have engaged in
23	practices that violate the antitrust laws, those should
24	be enjoined. But as we've seen, the protocol licensing
25	provision did not respond to a proven violation, and did

59

1	not even address technology -- and it addressed
2	technologies that were not even the focus of the
3	liability phase.
4	During the remedial proceedings, there was a
5	record developed on network computing and there was
6	evidence introduced of various so-called bad acts, as
7	Judge Kollar-Kotelly characterized them, but she treated
8	them as being essentially irrelevant, because they had
9	not been shown to be anticompetitive, or at least if
10	they were anticompetitive, they may have had
11	pro-competitive justifications that had not been
12	considered.
13	The second, under this heading of only
14	responding to a proven need, I don't want to rule out
15	the possibility that forward-looking or fencing in kinds
16	of provisions may be necessary, but if they are, then I
17	think there should be -- there should be a record built
18	to support the need for them. And I think in this case,
19	for example, we know that the government at one point
20	actually surveyed software developers to see what their
21	needs were in this area.
22	I'm not sure what was done during the
23	negotiation of the consent decree, but perhaps more in
24	that direction could have been done to find out
25	precisely what was needed to ensure adequate

60

1	interoperation.
2	And also I would just add that the Court of
3	Appeals in the 2001 decision cautioned that remedies
4	should be proportional to the strength of the proof that
5	Microsoft's illegal actions actually reduced
6	competition, and that was why the Court of Appeals said
7	that divestiture was probably not going to be an
8	appropriate remedy, because as they put it, the harm to
9	competition for Microsoft's actions, in other words,
10	whether they had actually prevented Netscape's browser
11	or Java from evolving into a rival platform, that was
12	established by only -- as they put it -- by inference,
13	in other words, there was no evidence that that actually
14	would have happened. And where you have that relatively
15	weak evidence of likely anticompetitive effect, then you
16	need more evidence to support more Draconian remedies.
17	And divestiture is certainly that, but I also
18	think regulatory relief is also a Draconian remedy, and
19	that brings me to my second lesson, and that is to avoid
20	regulatory decrees, especially in high technology
21	markets. And this was recognized, Judge Kolar-Kotelly
22	rejected one principle during the remedial proceedings,
23	on the grounds that it would result in too regulatory of
24	a decree.
25	Well, the protocol licensing has become highly

61

1	regulatory and direct government supervision of price
2	and other terms of dealing and especially quality.
3	Direct government supervision of quality that's being
4	produced. And the device of the technical committee
5	certainly has provided a high level of expertise, but in
6	effect, what its created is a regulatory body, and I'm
7	not sure that the structure of the technical committee
8	and its relationship to the plaintiffs and the court
9	establishes an effective regulatory agency.
10	So, just to conclude, if in the future cases
11	have these characteristics, those should be treated as
12	warning signs, and addressed in the -- in the relief.
13	And with that I'll sit down.
14	(Applause.)
15	MR. HILLEBOE: Thank you, Bill. Marina Lao is a
16	professor of law at Seton Hall Law School. She
17	currently serves on the executive board of the section
18	on antitrust law of the American Association of Law
19	Schools, and she's an alumna of the Antitrust Division,
20	where she was a trial attorney. She has published
21	numerous articles on antitrust law and trade regulation,
22	and somewhat surprisingly on this high-tech panel, she
23	is the only speaker with slides.
24	Marina?
25	MS. LAO: I guess it's even more surprising

62

1	given that I am usually the least high-tech person on
2	the panel. Thank you very much for inviting me and I'm
3	happy to have the opportunity to participate in this
4	hearing.
5	I agree with a number of the speakers who have
6	gone before me who have said that remedies are often
7	treated as an after thought. Unfortunately, that's not
8	a very good idea, because success in proving liability
9	often does not translate into success in remedying the
10	anticompetitive situation, and so it's often best to
11	work your vision of remedy into the case development
12	much earlier on.
13	What I'm going to do, since I'm bringing up the
14	rear, is to try not to overlap too much with what has
15	been said; I'm going to focus on three main points in my
16	comments and I will be skipping over some of the slides.
17	First, where network effects are substantial in
18	the industry that's affected by Section 2 violation, I
19	probably differ from Bill, in that I think that there's
20	a need for broader rather than narrower remedies for
21	some of the reasons that I'll talk about later.
22	Second, again, I guess on this issue I differ a
23	bit from Bill as well. I'm going to talk about the
24	importance of forward-looking remedies. I would call
25	them affirmative remedies that reduce rivals' costs and

63

1	some of the problems in crafting them. I do agree that
2	tayloring these remedies to the problem is a bit
3	difficult.
4	And lastly, I'm going to discuss whether there's
5	any value in bringing Section 2 enforcement action if
6	there is no effective judicial remedy. My conclusion is
7	that there is deterrent value to bringing an enforcement
8	action, even if it is irremediable, so to speak.
9	Let me start with a few words about the ongoing
10	debate among antitrust commentators on the application
11	of antitrust in the dynamic high technology markets.
12	The question that is often raised is: Do we need more
13	rigorous antitrust enforcement or do we need a more
14	hands-off approach? Those who say that less
15	intervention is necessary generally argue that because
16	there is rapid innovation, product cycles are short, and
17	so dominance is fleeting. And there are continuous
18	opportunities for fringe firms to overtake the
19	incumbent. The Microsofts of the world will have to
20	constantly innovate or they're going to be left in the
21	dust.
22	And so for that reason, there's really not that
23	much of a need for antitrust intervention in order for
24	markets to remain robust. In fact, too much antitrust
25	intervention could stifle innovation and competition.