ILD 2012 | Jack Goldsmith: National Security Law and Cyberspace

Global Affairs Academy
6 followers
Aug 29, 2026
AI Generated·May contain errors
We're very fortunate to have as a keynote speaker Professor Goldsmith from Harvard University.
Those of us who work in international law are certainly familiar with his remarkable scholarship
and experiences. You have his impressive biography in your biographical materials. I will not
reiterate them here. I will just say, Professor, thank you very much for taking time out of your
busy schedule to be part of our conference. Thank you. Thank you very much, Colonel, for inviting
me. Thanks to you and Michael Schmidt and everybody else who's involved in inviting me
and organizing this conference. I have a disclaimer to make. I'm sitting down because I'm not well.
Since last Wednesday, I've been having very bad dizzy spells. This is more information than you
need, but the doctors have assured me after days of tests that it's nothing serious,
but the room is still spinning a little bit for me. So if anything I say is incoherent
or unpersuasive, I'm not usually this way.
So the organizers gave me a little bit more leeway in the title. The title I think they
gave me was national security and cyber law or something like that. I took this as an invitation
to broaden the scope. A lot of our conversations are going to be down in the weeds in the next
three days. And so I thought that near the beginning of the conference, it may be useful
to give a broader perspective on how I see law and cybersecurity intersecting in very big picture
terms. And so some of what I had to say is familiar, but hopefully the whole picture taken
together will be useful for the conference. Next slide, please. So this is my claim.
The digital revolution creates large structural gaps and effective legal regulation that
significantly disadvantage U.S. national security in particular. That the changes in technology
as they intersect with changes in our legal regime and in legal constraints in our regime
significantly disadvantage the United States national security broadly conceived.
And these gaps are very hard to close through traditional legal means. That's my pessimistic
assessment and that's what I'm going to try to convince you of in very broad terms.
Next slide, please. So I'm going to start in very general terms just by thinking about how we
normally regulate offshore threats. I teach jurisdiction. I've taught jurisdiction for 20
years and I've thought about the problem of offshore threats in many different contexts.
And I think it's useful to start off just by thinking about abstracting away from the cyber
context and just thinking about what are the means we use to regulate threats from abroad.
Let's start with crimes and attacks from abroad. Obviously, the first means of defense is geography.
And one of the things that the digital revolution has made clearer to us
than we used to appreciate is just how significant geography is. Of course,
we've always appreciated this, but I think we appreciate it more in the digital era just how
important geography is to defense from foreign attack. It used to be that you had to get close
to the thing you wanted to attack to in fact attack it or it took extraordinary means,
truly extraordinary technical, technological means to attack it from abroad. And by the way,
in crimes and attacks, I'm speaking of any kind of organized threat from abroad,
ranging from low-level crimes, so conspiracy to engage in the importation of drugs or
committing antitrust violation that has effects on the U.S. market, up to physical attacks and other
types of kinetic attacks on the United States. So geography was a large barrier to, was an
important element of regulation of these attacks from abroad. Domestic and international criminal
law were and remain important elements. Obviously, domestic law to the extent that the perpetrator
could be found inside the United States or to the extent that the perpetrator from abroad had
assets in the United States, the United States could use legal coercion to go after that
external threat and to try to raise the costs of the threat to lower the incidence of the threat.
And of course, that was supplemented by international criminal law, which established
a regime of extradition and mutual legal enforcement, mutual information exchange
that enabled countries to project their international criminal law abroad largely by
getting information from abroad or by getting the person, the defendant, the perpetrator from
abroad. Because one of the things about criminal law is that you need a defendant. You need to have
an entity on the other side that you can, in the court, that you can exercise criminal law against.
And one of the things international criminal law enabled you to do is to pull the perpetrator from
abroad, get them to the United States so that you could apply your criminal laws.
Changing scope a little bit, USADBELM was a means of regulating attacks from abroad,
but to the extent that USADBELM rules were consequential, they placed limits on what states,
the primary, the things we were primarily concerned about for kinetic attacks,
put limitations on states using force against our country. And then, of course,
various deterrence strategies, and I'm thinking here about deterrence through threatened
retaliation, various deterrence strategies were used to regulate the incidence of threats from
abroad. Espionage was never terribly well regulated by international law. It's been
regulated largely by domestic law, as most people in this room understand.
Basically, unless, as someone on the last panel said, unless you catch the perpetrator in your
country or perhaps in your embassy abroad, then in those circumstances, to the extent that the
perpetrator was an individual, in those circumstances you could apply criminal law
domestically and to that extent regulate espionage. International law has never very
robustly regulated espionage. In some senses, it doesn't regulate it at all. It leaves it
to criminal law. But for the most part, espionage from abroad is not regulated by international law.
So this system, and it's been very crude and very 40,000 footish, this system obviously led to gaps
in our national security. Crime still occurred. 9-11 occurred. Attacks happened. Antitrust
violations occurred. But the point is, is that I think it's fair to say that until the rise of a
mature digital regime, that on balance, this regime did a pretty good job of checking attacks from abroad.
Next slide, please. So all of these characteristics of the cyber, the next slide is basically about
how cyber upends the equation of traditional international regulation of, excuse me,
traditional regulation of threats from abroad. And this laundry list of characteristics of
the cyber problem, which is not a comprehensive list, will be familiar to most people in this
room, so I will run through it quickly. First is, obviously, we're more and more dependent,
all of us, every day on digital technologies, and dependence upon digital technologies are
growing in this country and throughout the world in many dimensions, and growing interdependent
as well. And with the dependence and interdependence, obviously, comes vulnerability.
Second, distance can be obliterated by these technologies, as we all know. This doesn't mean
that geography is irrelevant to the regulation of foreign threats. For reasons I just said,
where you find the defendant can determine, if you can find the defendant, can determine
how, whether you can apply your laws. Also, to some extent, nations can take steps against
intermediaries within their country, such as ISPs and the like, to have effects abroad. It's not that
geography doesn't matter, but as far as perpetrators are concerned, to a large degree,
on the network, distance is obliterated. Attribution, as we all know, is hard,
I would say sometimes impossible, and often slow. The government, until three or four years ago,
used to say that the attribution problem was unsolvable. The last few years, they've been
talking about how, basically the way I just described it, that it's hard and slow, and
sometimes slow, but it's uncertain. That's the important point. Attribution is uncertain. It's
uncertain about whether and when you'll have attribution, and the confidence of attribution
is often not as high in the digital world as it is in real space with traditional attacks.
I'll come back to the attribution problem later. The public-private distinction is muddled. This
is very important for legal regulation, obviously, because so much of law assumes distinctions
between private actors and public actors. This is true of all the law relevant in this area.
Just to take two examples, the legal regime differs whether the attack from abroad comes
from a non-state actor, private entity, and whether it comes from a state actor,
and the difficulty of that type of attribution can be legally confounding. At home, it's legally
confounding as well. I'll talk about this more in a moment as well, but there are many ways to
describe how the public-private distinction works, makes the application of law to cyber harder at
home. The easiest way to say it is simply in the well-known fact that, depending on the source you
look at, 90 to 95 percent of our military and intelligence communications travel on the private
network, and that by itself creates all sorts of legal problems. It creates problems for our ability
to defend those channels, because to the extent that they're in private hands, we have a different
legal regime. It has implications for the laws of war and what kind of target that is. In general,
cyber obliterates the public-private distinction that's—obliterates is too strong—muddles the
public-private distinction that's so important to so many of our legal categories. It also muddles
the domestic-international distinction. This is largely because of the attribution problem,
the geographic attribution problem. It's often not clear where the attack or exploitation originated.
Another characteristic relevant to the analysis is that attack agents and exploitation agents can
be hard to distinguish ex ante. This is a well-known phenomenon also, but the idea is that
both attack and exploitation require a vulnerability and the introduction of some
kind of agent. And often, until the attack occurs, it's difficult, assuming that the defender
discovers the operation. It's very difficult to tell whether it's an attack or an exploitation.
That has lots of implications for deterrence and law. As a result of all of these factors,
obviously, significantly more adversaries are empowered to do harm inside the United States.
Adversaries outside the United States, it's hard to know what the multiple is, but it's many,
many, many, many multiples. Not just because so many states can do naughty things in the network,
but so many individuals can as well, and that includes individuals at the border between
state actors and private actors. There's a rapidly growing weapons market and a rapidly
growing arms market. And there's a secret cyber arms race going on. I call it secret because
it's not like the good old-fashioned bilateral arms races that we had with the Soviets, where
there are often verification difficulties and there are often uncertainties in knowing
exactly what the state of our adversary's technological development was. But I think
nothing like what's going on now. And also, it's very hard to gauge. I think the logic of arms
races and everything we know anecdotally in the public about what other nations are doing and
what the United States is doing suggests that there's a fierce cyber arms race going on in
secret. Okay. So that, I think that this slide basically suggests, boy, there have been a lot
of radical changes that cyber brings to the regulation of threats from abroad. Next slide,
please. The legal response, however, has been, I think at best, tepid. Maybe that's an exaggeration.
Basically, I think it's fair to say, and I'm happy to be corrected on this, and I think it depends
at the level of abstraction which you look at it, but I think it's fair to say that we've had a
pretty tepid response to these dramatic changes in the context for regulation of threats from
abroad. Basically, what we've been in the process of doing for the last 10 or 20 years,
30 years depending how you look at it, is applying our pre-cyber legal tools, the tools I described
on my first slide, to this new world that is described on the last slide. And we've been
trying to work out the analogies and trying, and that's a lot of what this conference is about,
I think, especially in the law of war context, is trying to work through the proper use of the
laws of war from the kinetic context to the digital context. I think it's fair to say that
there's been relatively little change in domestic law. A lot of it has been organizational and
bureaucratic. It seems, if you read the New York Times, that the Department of Defense has organized
itself much more aggressively and successfully, not perfectly, obviously, to defend itself,
and the intelligence communities, perhaps, than the other critical elements of our society,
especially private critical infrastructure. But while there's been a lot of discussion and
there's been incremental change, I think it's fair to say that we've barely begun
to take the legal steps in domestic law that I and many people believe are necessary,
and I'll talk about those later. There's been little change in international criminal law.
The Cybercrime Convention is now a couple of decades old. The Cybercrime Convention,
which is, for those of you who don't know, it was a convention designed to basically harmonize
domestic laws with regard to certain defined cybercrimes and to enhance international
cooperation, including extradition and shared legal information assistance among nations related
to cybercrimes that were defined basically 20 years ago, 25 years ago. The Cybercrime Convention
itself was a non-trivial step, but as I'm going to talk about in a second, it wasn't a giant step
in changing international criminal law, and it hasn't been a very successful treaty either,
as we'll see. Other than that, I think there's been very little change in international criminal
law to address the changes in the digital world. There's been little change in the laws of war.
And as I say, basically, both on the USADBELM and the USINBELO side,
both in public scholarship and it appears that the government behind walls of secrecy,
are engaged in an elaborate process of trying to figure out the proper analogy and how to
analogize kinetic effects, excuse me, digital attacks to attacks with kinetic effects. That's
basically what we've been doing. Michael Schmidt wrote the canonical article on this and several
cents, basically trying to analogize digital attacks to kinetic attacks and try to figure
out how the law should apply by analogy. And I think it's fair to say there's been no law,
no change in the law of espionage. Next slide, please.
So basically, if I've got the picture right, we've got dramatically changed situation on the ground
in terms of technology and capabilities and threats, and very little in terms of legal response.
And it seems obvious that the digital changes create a huge gap and growing gap. This is,
I think, there are people who question the conventional wisdom that this is a
giant threat, and not just a threat, but a reality of a national security problem.
But there are those who question it. And in truth, it's hard to know the extent of the problem,
because the government and private sector keeps so much information secret about the nature of the
problem. But both the logic of the problem that I've just described, and public reports and the
alarmist calls by the government, which we should always question, but which I believe to be largely
credible, suggests that there's a large and growing problem and is a large and growing gap
between the threat itself and the effectiveness of our legal tools. And very dangerous national
security fall in these gaps, so something is missing there. Sorry about that. The basic point
is that large national security problems are created by this gap. And it's not just
the problems that national security lawyers usually typically think about. We typically think about,
well, it's not just military and intelligence threats. It's not just the threat of attack.
It's a different in kind espionage exploitation threat. And there's an economic component to it
that is really the largest part of the problem right now, by all accounts,
that really expands the notion of what a national security threat is. And a lot of our legal tools
don't even really credit, a lot of our legal tools don't even really credit the theft,
the intellectual property theft, the trade secret theft, the commercial theft and exploitation
that's going on. That doesn't really normally even fall inside of our legal tools. It's not
normally classified. It is to some degree, but not normally classified as a national security
threat. Finally, I want to claim that these gaps are very hard to close by law. So the
next few slides are going to be making this argument about the problem with the response.
Next slide, please. Everybody in the room knows the claims about cyber espionage,
about how much information we're losing in the military and in the private sector.
And it's easy to see why. It's just because of the reasons I said before. So many more people
are empowered. Offense tends to beat defense here. Geography is no barrier. There's every
reason to think that there's massive underreporting of the theft we've suffered over the years.
The result is that, and I'm going to get to this more in another slide as well,
is that there are real losses here in terms of technological advantage, which is another thing
the digital era helps us to appreciate, is how important our technological advantages are,
because we're seeing in very real terms that they're being stolen in both the public and
the private sector. And as Joel Brenner and many, many other people are fond of saying,
if this theft were being done, if things were being hauled off in cartloads from the Department
of Defense, or from defense contractors, or from the Googles of the world, if the information were
being hauled off, if we could see it in physical form, there's no doubt that we would declare a
national emergency. But we don't see it. We hear about it. We in the public, I mean, we don't see
it. We hear about it. We hear it reported. Some of us suffer disruptions. But largely,
the problem with this problem, the espionage problem, is that it is mostly invisible to the
public. Another problem created by cyber espionage is, for reasons I stated earlier, that espionage
can be and is a precursor to attack. So a legal regime internationally that doesn't regulate
espionage is a legal regime that doesn't regulate preparation of the battlefield. And that's a
change, I think. Maybe not of kind, but a large change in degree in how national security works.
I think it's fair to say that the law is, and again, I'm happy to be corrected on this,
the law is useless against all but local, the law of espionage is useless or nearly useless
against all but local insider threats. Threats inside entities, firms, or the government that
are located within an area that we have jurisdiction over, mostly the United States.
It's not regulated by international law. Many of the things, most of the things,
probably all of the things that the government complains about in terms of theft,
both of the government's networks and private networks, is not an international law violation,
at least as international law is traditionally understood. Domestic law is only effective to
the extent it can get the actors, the individual actors. And to the extent they're state actors,
you don't haul states into criminal court. It's not going to happen. And to the extent that they're
private actors abroad in places where we don't have extradition agreements or we have ineffective
ones, then criminal law is going to be effective. Cyber espionage is not redressable by military
means, not credibly anyway. And in fact, if, again, I only know what I read in the newspapers
these days, but if you just read the newspapers carefully, you would, and the reports that the
government puts out, you would conclude that the U.S. government isn't doing anything about the
problem of cyber espionage. We can barely bring ourselves to even mention that the Chinese,
only recently has the government been patting itself on the back or even mentioning that the
Chinese are one of the main culprits here. And of course, we've been engaged in an international
cooperation initiative for the last few years that is in part a response to this problem.
But other than that, you'd be hard pressed to see any diplomatic or political responses
or military responses or even threatened responses in response to what the government
claims is a crisis of theft and espionage and exploitation. Next slide, please.
On cyber attack and on, so the attribution problem is well known. I just want to take
small issue with what I heard someone say in the last panel. I think the attribution problem,
even to the extent that I'm talking now about the problem of responding and regulating cyber
attacks from abroad. The attribution problem, first of all, there's the problem of it's
uncertain. It's not always clear that we'll be able to attribute and it's not clear how,
what kind of confidence we'll have in the attribution. But let's say we,
one of the speakers, I may have misinterpreted him. He said that oftentimes you'll be able to
tell from context how to attribute. So if there's a rising tensions with Iran and then something bad
happens, this is what I took the example to be. If I'm misinterpreting it, forgive me.
Rising tensions with Iran, some cyber episode happens, we can infer that it was Iran. But,
of course, we can't just by the virtue of the context infer that it was Iran. All sorts of
other mischief may be going on. Other adversaries may want to provoke a war with Iran and may want
it to seem as if it's coming from Iran. So I don't think we can infer much from context.
There's another problem, at least in that context, I took it that was the example.
I don't want to deny, because I know it's not the case, that there can't be attribution
with pretty good certainty through a combination, especially for large scale attacks,
through a combination of technical means and human intelligence means and other means.
Don't want to deny that. But even if you have perfect attribution in secret, there's another
hurdle that cyber presents, some other things do as well, but that cyber presents that differentiates
it from, say, the good old days when we had bilateral nuclear threats from the Soviets.
And that is, even if we have perfect attribution in private, how do we prove it in public?
And this is hugely important to the credibility of our public response.
And unless we can, with confidence, prove the attribution in public, and there are all sorts
of hurdles to that, means and methods hurdles, and also just the interpretation of even if we're
willing to open up the whole vault, which we never are, the difficulty of publicly proving
the attribution of a cyber attack is a serious problem for being able to respond to that attack.
Which leads me to deterrence. It's very hard. So those are the difficulties of attributing
a cyber attack. It's hard, as many people in this room have thought about, for traditional
ideas of deterrence to work as applied to cyber attacks. I'll just check off some reasons.
The key, I'm talking now about deterrence by threatened response. That's my phrase for it.
I know that there are other phrases for it. There's deterrence by defense. I'm talking about
the deterrence that if you do something bad to us, you can expect to be punished,
and that you will lose more in the punishment than you gain in the attack or exploitation.
So there are many reasons why we're having difficulty establishing a deterrent strategy
in the cyber world. I'll just mention some of them. Thresholds are very difficult to establish.
You need to be able to credibly commit to responding. Your adversary needs to know.
You need to be able to be pretty clear where the line is. And the line against which you'll
credibly respond has to be in the right place. The Department of Defense, I can't remember if
it was late last year or early this year, finally came out with a new declaratory policy. And the
basic threat was that we would respond with overwhelming force in the face of a, quote,
crippling cyber attack. Crippling cyber attack. That's the red line. A crippling cyber attack
to which we claim that we'll respond. Well, I'm not sure what counts as a crippling cyber attack.
That threshold is very, very, very high, suggesting that if we're trying to have a
deterrent strategy, anything below crippling, are we implying that we won't respond to that?
And what counts as crippling is obviously not crystal clear.
There's the attribution publicity point that I made that's a hurdle to a credible response.
The reason, if we, to the extent that we cannot publicly prove the source of a cyber attack from
abroad, to some degree, that makes it harder for us to respond, at least in public, through publicly
noticeable means. And to the extent that's true, that makes the credibility of the response
less. To the extent that, as we all know, the effects of a cyber attack are hard to control,
that makes the response less likely, and that makes a credible response less likely,
because the costs to the responder are higher. Many people have mentioned that because offensive
weapons are secret, it's very hard to have secret weapons be a deterrent. I think it's
more complicated than that, but there's something to the point. Finally, this is a point I'll come
to in the next slide. Our well-known legal constraints on our U.S. military are, it's
publicly known that our military is very tightly legally constrained. I'm not saying too tightly
legally constrained, I'll just get to that in a second, but that we have very elaborate legal
processes and restrictions, and to the extent that is true, that makes a credible threat of attack,
given the other uncertainties I mentioned, less likely. Those are all reasons why,
this is the wrong slide, I'm so sorry, I'm not paying attention, you need to go back a previous
slide, I apologize, I've been working off the wrong slide, I've just been doing the first bullet
point, forgive me, I need to watch what's going on here, I can see it. Okay, so one hurdle to,
given the digital changes, to laws response to cyber attacks is attribution and deterrence
is problematized. Another is that legal gaps and uncertainties, there are gaps and uncertainties
in the face of very serious threats. Cyber attack on my simple slides. So I think,
and this is going to be a discussion for this afternoon and tomorrow, it seems to me that the
kinetic analogy, this I think is obvious, underplays or doesn't get at what might be
very serious cyber attacks. So economic targets, economic disruption with no kinetic effects
is, I think, a hard problem within the Schmitt model, within the kinetic effects model.
Stuxnet, it's not obvious to me that Stuxnet was a, let's assume away,
assume away the self-defense rationale for Stuxnet. And let's just ask yourself,
is it an armed attack? Is it a use of force, I'm sorry, under the UN Charter?
It's not obvious to me that it is under the Schmitt analogy, the Schmitt rationale,
I think at least it's an open question and a debatable question, whether tiny changes to
digital, to the computer's operations that lead to minor but deeply consequential changes in
physical operations counts as a use of force under that model. I'm not saying it does or doesn't,
I think it's a hard question. The point is, there's uncertainty about that. Also, as I said
more than a few times, international law doesn't regulate espionage. Espionage has just unfathomably
more consequential consequences now than it did before the digital regime. It's completely under
the legal radar screen. The inability to distinguish exploitation from attack, ex-ante I've already
mentioned, that further emboldens offense and is a further hurdle to defending against it.
I think, and again, I've made some of these points, kinetic responses to all but the most
extreme public attacks are going to be hard to justify because of the attribution challenge.
And then, of course, the deterrence problem, to the extent we have poor deterrence,
that emboldens our adversaries. Okay, now the next slide, please.
So to some extent, everything I've just described is a set of threats that to some level of
distraction every nation faces. And it's one way of describing why offense beats defense.
It also, though, has something to do with our unusual legal law and legal culture.
So there are three reasons why I think the U.S. is obviously asymmetrically vulnerable. And I'll
assume here, I'll assume here that we have the most, because I think it's true, the most powerful
cyber attack capabilities in the world and the most powerful cyber exploitation capabilities
in the world. Let's just assume that. I still think the U.S. is asymmetrically vulnerable.
I'm happy to be, I hope to be, disagreed with convincingly about all this stuff.
The U.S. is more dependent than most nations. And with dependence comes vulnerability. The U.S. is
more restrained than most. And I say this even after the revelation of Stuxnet, Olympic Games,
planes and the like. As a general matter, it's our government's policy that we will do little
government-sponsored economic espionage. Not to say none, but it tends to have to
have a national security nexus. We certainly aren't engaged in the scope of economic espionage
that our adversaries are, in part because there's less to steal from China. But that's not true of
our adversaries necessarily, friendly adversaries who engage in economic espionage against us.
So we tie our hands in that respect. We treat the private network like a communication and
data storage medium rather than a channel of attack. So the main channel of cyber attacks
and exploitations we treat as a civilian network with the presumptive protections of the Fourth
Amendment, the First Amendment, and the elaborate and arcane set of statutes that governs the
government's operation in the private network. And then we have a lot of law-induced caution.
And again, without saying whether we have too much caution or not enough caution,
I think it's fair to say we have more caution than most countries, if not all countries.
And I just list a few of the many sources of that caution, AUMF, Title 10 versus Title 50.
By caution, I mean hurdles to action, the War Powers Resolution, the Fourth Amendment, and the
like. Finally, I think the U.S. has farthest to fall, which is obviously an element of asymmetric
vulnerability. Next slide, please. So why haven't we been doing more about this? And again, this is
the big-picture overview of national security law. I think there are a couple of reasons,
two major reasons why we haven't been doing more to raise our domestic defenses. One is what I've
talked to about the public-private divide in the network. It just seems obvious to me,
and it seems obvious, I think, to those, many in the government, and especially those in the
military and the DOD and the NSA, that the National Security Agency is the agency with the expertise
and the knowledge and the experience of both, of every element of this problem,
of the technical elements of this problem, with identifying threats and responding to threats.
And you can listen to government speech after government speech, especially those emanating
from NSA and DOD, and therefore they should be discounted because of their self-interested
speeches to some extent, but I believe them to be credible, urging the government to allow
essentially the NSA to be more involved in the network, in the private network,
both identifying vulnerabilities and responding to them and working hand-in-hand with the private
sector in a real-time, concrete, active way to find and address vulnerabilities.
And one lens to see this is about the debate about Einstein and Einstein 2 and whether that
should be extended to the private network. There's a debate about the effectiveness of Einstein
and I don't know the answer to that question. I don't have competence to figure out whether a
signature-based system is, to the extent it's cost-effective, but let's assume that it is.
The NSA and the DOD are urging the extension of that, of an Einstein-like system into the
private network, and the country's just not ready to go there. And the simple reason is,
and it's a lot more complicated than this, that the country yet doesn't trust the NSA,
for perhaps good reasons, perhaps bad reasons, but we just don't have the trust yet to allow
the people with the most expertise to address that problem. That's one hurdle
to raising our domestic defenses. Another is what I call the paradox of digital progress.
So much of our defensive vulnerability at home, in the private sector especially,
is a result of misplaced economic incentives. The actors that are either causing the harm
or who could, excuse me, the actors who are contributing to the harm, or who could take
steps to ameliorate the harm, are not being asked to do so. So users, end users who don't use
adequate computer hygiene, they get away with it, even though they contribute to a public bad.
And ISPs who could, through proper regulation, take much more aggressive steps to clean up botnets
and spam, which are the engines of much crime and sometimes worse. We're just beginning to think
about regulating them and urging them to do that. We can't even, there's a debate going on now in
Washington that many people are aware of, about whether we should take the first, and in my
opinion, lightest step in coming up with some regulatory regime of uniform standards for
the protection of critical infrastructure, CI. And the combination of the
anti-regulation forces and the privacy forces in Washington have made the lightest regulation of
government regulation of critical infrastructure, I think, is not going to pass. It looks like it's
not going to pass. And this is really, as Stuart Baker has written, a pretty modest first step
in dealing with a problem of critical infrastructure. The basic economic problem is,
it's simple to state, is that critical infrastructure in private hands, the persons
in charge of those facilities, if they're successfully attacked and taken down, the
whole country suffers in a very bad way. And the question is, do those entities have the
proper incentives to properly, to optimally invest in cybersecurity defense to protect the nation
accordingly? Economic logic would suggest that they don't, and that's a perfect and indeed classic
context for government regulation. But the question is, is the government competent to do
so? Many people think the government would make it worse. In any event, why have, one reason that
we've, I think, failed to insist on internalizing all the various negative externalities produced
by software manufacturers, ISPs, end users. And part of it, it is the privacy and anti-regulation
ideology. And part of it, though, is simply that this is the paradox of digital progress.
We see the benefits of increased digital dependency. They're immediate, they're obvious.
We always want more from our digital world. We want better, faster, cheaper electricity.
We want better, faster delivery of healthcare records. And we can see those benefits when we
take the next step of dependency. And so we grab them and we get economic and social and personal
benefits from them. We don't see the vulnerabilities. And so we keep pushing them down the road and
pushing them down the road. And they're very expensive and we don't want to pay for them.
And that cycle has been going on for a very, very long time. The U.S. has pursued short-term
economic and military gains from digital progress and ignored longer-term costs
of not securing the systems. Next slide, please.
So the latest move, given that we've not been very good at raising our digital defenses,
and I don't think that's a complete answer, is to seek more international cooperation.
I'm skeptical, very skeptical about the move to international cooperation. I want to leave time
for questions, so I'm going to cut this short. I'll be brief. The Cybercrime Convention is much,
much more modest in its ambition than anything that's being talked about for a kind of cyber
arms agreement or kind of even modest cyber weapons agreement. And I think the cybercrime
agreement is a cybercrime treaty is a failure. It's a failure for many reasons. It got 31
ratifications on goals that were much more modest than offensive weapons control would need.
Its mutual assistance program has worked a little bit. We've got better criminal
cooperation now than before. It's got a large sovereignty exception that swallows everything
when it wants to. And if we can't overcome a sovereignty exception when it comes to cybercrime,
maybe we can in the context of higher stakes gains, but I think probably not.
I think, and I'm going to have to be brief, I'm happy to answer this boring question and answer,
that there are many fewer alignments of interest between, say, China and the United States
on a deal for a cyber agreement to tamp down on offensive use of either attack agents or
exploitation agents. I think there's very little overlap in our interest. We just conceived the
problem completely differently than the Chinese do. There's the problem of mutual concession.
We talk a lot about what others want we should stop doing. We talk a lot about we should have
an agreement to stop attacks on critical infrastructure. And we should have an
agreement to stop exploitation of the private sector. That's pretty convenient. That's basically
what the United States wants to achieve. But what are we going to give up? Are we going to give up
our exploitation capabilities? Are we going to give up our attack capabilities? I haven't heard
anyone seriously contend what, if anything, we're going to give up. And without mutual concession,
there's no agreement. This is a delicate point, but I think it's an important one. The United
States is, and is perceived as, a large part of the problem. This is not something we talk about
much in the United States. But to the rest of the world, when you talk about the cyber bully,
it's not the Chinese and the Russians. It's the United States as much as those, if not more so.
And the polls show this, and the polls suggest this. This is because a lot of reasons. We have
giant unregulated botnets and spam generators. We have a pretty aggressive copyright regime,
which many people view as a threat to cybersecurity. This whole debate in the
United States now about giving up control of the internet to the ITU, which I am basically with
the United States. I think it's much better regulated by ICANN, but that's viewed as a US
institution. And that's the reason people want to take it out of that institution,
is because the United States is essentially lesser than it used to be, but it still
has some control over ICANN. The United States supports activism aggressively, which many people
see as a giant cybersecurity threat, the Chinese especially. And then there's, of course, what's
been in the news in the last month, which is the United States has been engaged in some very,
according to press reports, very aggressive and novel uses of cyber attack and exploitation agents
in Iran and elsewhere. You know, President, Secretary of State Clinton gave a speech,
many speeches on internet freedom in which she's talked about how we have to really punish
nations that engage in cyber attacks. And nations read that, and then they read about Olympic Games,
and the dissonance there is something we're not attuned to, but the rest of the world is.
So those are all the problems with cyber agreement before you even get to the problem of
verification, which is the most obvious problem, which I won't go into. And some softer and narrower
forms of cooperation is possible, and I'm not a complete norm skeptic here. The fact is we have
some norms operating now. We don't do everything we could, obviously, and other nations don't do
everything they could against us. And through self-interest of that sort, we do that for
self-interested reasons, because the costs would outweigh the benefits, norms can develop. But I
think harder forms of cooperation are going to be very hard. Last slide, please. So my basic view,
and I think a lot of people in this room share this view, is that we're not, the basic message is,
I think the United States is not being serious on the legal front in addressing this problem,
and for all the reasons I just stated. And I don't think it's going to do anything serious,
and a lot of people do, until a major cyber event happens. It may not be the lights going out,
it may be something happening in the economic realm, or the health realm,
but until we have a large cyber event that causes significant harm that members of the public can
see, only then, I think, will members of Congress take the steps needed to begin to address this
problem. Some of those steps, I'll get to in a second, well, let me just jump over the second
bullet, and I'll come back to that. So what's going to happen? It seems inevitable to me that
the United States and the NSA is going to be much more deeply involved in the private network,
as I described a little bit ago. The challenge there is finding ways to have the NSA credibly
involved, doing what it thinks it needs to do, while at the same time being credibly
restrained. That is the challenge. How can we create enough public trust and public scrutiny
of the NSA to give it, to allow the nation to trust it to do the things it needs to do?
That is the central challenge. It seems we're going to have much more regulation when the
to correct many of the disincentives that now burden the legal regulation landscape.
Let me go back to the second bullet point. So this is a point I make delicately in a room
largely of lawyers, but I think there are largely lawyers. So I think it's very important
that this process not be too law-driven. The process of figuring out what we need to do to
keep the country safe from the cyber threat should not be too law-driven. That the best
strategy should be, the best thing to think, what the government should be doing is figuring out
the best strategy largely without reference to legal restraints. Let policy guide the law.
That doesn't mean you ignore the law. I'm obviously not suggesting you ignore the law.
But there is enough flexibility and uncertainty and novelty in everything I've discussed
to allow for legal innovation, as has always been the case. And just reading the newspapers,
one hears two kinds of stories about excessive lawyer involvement. And I think it can be a good
or bad thing. I've written two books about both the good and the bad aspects of lawyers being
heavily involved in national security. One problem is that lawyers can argue too rigidly by analogy
and say no to an operation that is sufficiently novel on the law that there should be room for
flexible interpretation if it's otherwise the right thing to do, both tactically and strategically.
The other problem, if you read David Sanger's account of President Obama basically agreeing
to ramping up Olympic Games in Iran, you'll see, and Sanger's follow-up account, you'll see him
saying, you know, we didn't really think much. No one really wanted to talk about the long-term
strategic consequences of what we were doing. This is the other side of lawyers being too heavily
involved. Because when lawyers are very heavily involved, sometimes when the lawyers say yes,
that is a green light to do something that is, I saw this a lot when I was in the government,
a green light to do something that might not be strategically a good idea.
So this is not a rant against lawyers. I'm a lawyer. I love lawyers. And I've just written
a long book defending the role of lawyers in the national security world. But we have to be
careful not to let the process of analogy and updating be too lawyer-driven.
Finally, the last two points, and I'll stop. It seems to me obvious that ultimately we're going
to, this is not new to me, obviously, that we're going to have to invest more both on the preemption
side, aggressively being in the network, living in the adversaries network, stopping attacks and
exploitation before they occur, and also on the resilience side. And finally, I'm working on an
essay. I think that we're going to have to rethink the international law of espionage.
I think that the consequences of a regime of non-regulation of espionage used to be relatively
low. They're now hugely consequential. It's hugely consequential for us. And I think that regime
won't survive.
