Speculations on Armed Conflict

In a Time of Free Silicon


Chapter 6


The Net is the converging global system which puts people and their information in close electronic contact with each other. The growth of the Net, by permitting subnational and transnational communities alters the basis for international conflict. The Net, itself, however, presents certain exploitable vulnerabilities for societies that depend on it.

Patterns of war reflect the relationships of individuals, the communities they form, and the nations they live in. The information revolution has already and will continue to alter this flux, but in unexpected ways. Thirty years ago, the glib consensus was that information technologies would create a global village; to a large extent it has. The further rise of the Net -- a future complex of sensors, processors, and communicators -- will create global villagers. The new villages will be unbound by geography, but bound by their own parochial reflexes. They may find new and exciting ways of getting along.

The impact of the information revolution on the sources of conflict -- the political construction of societies and the expectations of their members -- is both more and less obvious than its impact on purely military operations. Most of the technology necessary to power the civilian side information revolution has already been invented; it only needs lower cost (as happens continually) and wider distribution. With it comes the elaboration of the information revolution to new uses and new users (particularly in the South). Harder to assess is the dynamic of commercial competition in information markets. Military revolutions tend to be driven by well-known forces. Technologies proven useful are likely to be adopted by someone. Once they are demonstrated, complementary and countervailing capabilities follow. Commercial competition is a complex game involving competing vendors and multiple consumers with varying needs. Only some of the possible converts to the probable because the calculus of individual desire does not lead to a closed set of outcomes.

In its commercial adaptation, the Net is, in one form or another, inevitable. The declining cost of acquiring, processing, and transmitting bytes will call forth an infrastructure which puts people (and their machines) in closer, faster, and denser contact with each other. The Net may be likened to our phone system extended first globally, and then to every possible digital device, removed from its land-bound linkages, given the power to transmit multiple video streams, and overlaid with enough filters and translators to find every needle in the global haystack.

The impact of information technology can be discussed in terms of five broad trends: the erasure of distance, fixed and floating networks, universal translatability, the mutability of truth, and, as a consequence of all this, the rise of the global villager. This forms the context of national security. The next section deals with the ghosts in the Net.

From Global Village

By and large, the information revolution has spread knowledge faster than fearful governments can slow it down. Cheap cassette players and tapes help spread the 1979 Iranian revolution. Fax machines helped power the 1989 uprisings in Tiananmen Square. Leaders of the Soviet Union's abortive coup in 1991 failed to appreciate how modern telecommunications (e.g., voice, video, and E-mail) in the hands of those who understood them (e.g., allies of Boris Yeltsin) could become such powerful weapons. Although ham radio operators in Bosnia did not stop atrocities, they have prevented their taking place in secret. AsiaSat is sending television signals that travel past the reach of censors. Traditional regimes cannot easily control the information that their populace receives. However, if a populace like the Serbian does not wish to hear bad news about itself, only modest amounts of media repression will be sufficient to keep society closed.

The ubiquity of broadcast media has CNN-ized perception-- hence, world politics. The instant access to world news available in the United States since roughly Huntley-Brinkley days is now available overseas as well. Many world leaders talk to each other via CNN and other networks. This capability has forced the West to respond to suffering in places such as Somalia otherwise beyond public attention.

Yet, cheaper telecommunications, while obliterating the dominant role of propinquity in creating communities, cuts both ways. It is easier to create communities that traverse geographical boundaries, but it is harder to find a unifying force or a common set of cultural assumptions in communities that are defined only by geographical boundaries. The state is not ready to wither away, but its suzerainty over a world of global villagers (despite some resurgences of nationalism in the second world) will be redefined. Such redefinition could affect national security much more than would the advent of battlefield meshes. The latter come into play only during those rare moments when strife erupts into war.

The Erasure of Distance: The cost of doing business over wide distances (especially overseas) will keep dropping dramatically. The volume of international calls will keep rising briskly, and low-power cellular phones are likely to, in ten years, permit satellite-connected phone calls from anywhere. Most cities will also have the infrastructure for dial- up videotelephony. Emerging technologies of virtual reality could let people sense, in whatever detail required, a physical phenomenon (e.g., a malfunctioning refinery, a wounded person) half the world away.

The reduced cost of coordinating a world-wide enterprise will strengthen the internationalization of corporate business, particularly manufacturing. Whether or not corporations then become truly global, the competition between semi-skilled workers in the West and skilled workers in the South will grow sharper. Joining footloose manufacturing will be footloose backroom services and perhaps even some frontroom services that require face-to-face contact.

This transfer cuts both ways. On the one hand, check processing, for instance, is moving from South Dakota (itself relocated from Manhattan) to Barbados. On the other, thanks to remote virtual sensing that allows a person here to manipulate robotic instruments there, a surgeon in Chicago, could work on a patient in Caracas who would (if awake) perceive the doctor as an apparatus.

Freer communications tend to cut the cost of conducting both routine and knowledge-intensive business in the South. This should work in favor of broadening economic growth (competing with other factors that will narrow it). The easy spread of text and image could spread education everywhere and thus most help bring Southern workers to Western standards.

Other barriers to business that derive from differences in language or currency would also fall. Computers that can recognize anyone's speech will reach the market by the mid- 1990s (replacing systems that must be trained to the nuances of each speaker). Language translation is making comparable progress. Good but slow and domain-specific real-time translation is already possible. Newly invented devices can read signs in foreign languages and flash the translations to video devices such as "heads-up" displays associated with your glasses.

In the meantime, more people will want to learn English to understand the growing warehouse of entertainment and educational material about to become globally accessible on-line (just as they now learn English to conduct business). Most people who have attended high school anywhere in the world should know English well enough to talk without translators.

Similar barriers are falling in currency translation. Electronic banking and the virtues of automatic currency markets will let people keep bank accounts with equal facility in any currency (unless governments stand in the way). Money, after all, is a measuring rod of value just as a yardstick indicates length. Exchanges and contractors can be denominated in them even if neither side owns them. Thus, little prevents considerably more business in the South from being conducted in Western currencies. This is good news for countries plagued by high inflation and unstable currencies.

Globalization, in the 1980s at least, promoted old fashioned liberal values (e.g., free commerce) in both West and South because if freed wealth from state influence. The information revolution can only deepen such trends. When wealth is reified in physical, largely immovable objects like land, resources, factories, and buildings, it is subject to diversion by governments. When markets must be serviced from local sites, the state gains similar leverage. As more wealth is contained in the movable intangibles of information, or when markets can be served from anywhere the influence of the state recedes. Ultimately, major corporations can be run out of a collection of networked briefcases each situated in one or another vacation spot, where the weather is equable and the taxes are low. In the 1970s the South imagined the golden road to wealth led from control over resources; hence governments tried to raise the prices of commodities they commanded. In the 1980s, the surer road was to create a subservient but well-educated workforce that multinational corporations could exploit or trade networks could tap. In the 1990s and beyond, savvy nations will complement human capital with dense robust information infrastructures to jack their growth path upwards.

The Global Net: Traditionally, the South (and rural West) was characterized by a sense that its denizens were simply out of touch with the greater universe. The ubiquity of the Net will connect individuals and give them access to a vast library of knowledge. Its core will, more likely than not, be the global Internet and its fifteen million subscribers (and growing fast). The Internet provides a vast, fast, and reliable electronic mail network, the ability to download information from public files located anywhere, and support for on-line news and bulletin board groups of every shade, variety, and flavor. All three foster the growth of global communities linked by interest and earlier separated by geography.

With time, the Net should allow anyone with a video-input- phone to see the world's accumulation of organized information: scientific and medical articles, papers, books, serious journals, newspapers, photographs, and maps. Navigating through such seas will, at first be daunting, but tomorrow's information pilot fish, so-called "know-bots," would stand by to swim through this enormous data base. With growing sophistication, it could find answers for those questions that data can answer. Personal filters could cull listener-specific news items from the glut of world news broadcasts and other sources of new information. Global access to the Net facilitates education and business from all ends of the globe. Moreover, it overlays the economic potentials of the West atop Southern societies which are structured to cope with far more restricted economic and social potentials.

The communications revolution will also accelerate the transfer of open-source defense-relevant technology, making it much harder to control. Any computer chip reducible to an algorithmic formula could, one day, be manufactured in one of hundreds of facilities. The world will not lack small rogue fabrication shops willing to evade export controls to make money. Technology control regimes for unclassified software and micro-electronics will be virtually impossible to police.

Inevitably, every device worth talking to or hearing from will come equipped with low-power communications, high-power computations, and a virtual address. Networks will increasingly link equipments -- automobiles and other vehicles, traffic lights and toll booths, factory machinery, remote cameras, various utility meters, medical and scientific instrumentation, and instruments otherwise useless if not network -- rather than people. Mobile equipment with GPS receivers will communicate their location periodically. Coupled to this network will be various sensors used to monitor certain environmental activities such as weather, soil conditions, toxic emissions. Others would monitor the health of the sensitive measuring blood chemistry, brain wave readings, heartbeat, perspiration, pneumatic functioning and so on.

What will they all talk about? Remote operation and monitoring of machinery may be a major topic. Cars and traffic lights will have long and loving conversations about road conditions. Devices will babble, "I'm OK, are you OK?". Note this picture of the future city -- all these networked sensors coupled with intelligent nodes -- increasingly resembles the Mesh (without rockets). How much technology separates the sensor- rich automobile increasingly sensitive to its immediate surroundings from a sensor-rich tank?

On the one hand, building the Net requires all these devices interoperate so that their communications protocols, data formats, and associated algorithms work with each other. The amount of attention paid to standards will undoubtedly rise even though increasing computer power would make the operation of gateways, translators, and virtual device layers less visible to users.

Ironically, the groups that represent the various connected domains may even discourage communications that standards supposedly permit. Should that seem odd? Communities ranging from the professions to street gangs maintain their own jargon. Ostensibly they (or least the professionals) cling to a separate jargon to make precise distinctions unavailable from ordinary language. Yet the more powerful motivation may be unstated: to exclude outsiders (for whom a little knowledge is dangerous), establish status distinctions, or preserve privacy. The Net that unites also divides.

To Global Villager

What turns the world into a global village, with everyone capable of looking over each others' shoulders, may also promote the creation of global villages -- communities of interest and inclination that span the globe but let members isolate themselves from others outside.

CNN, for instance, not only lets you keep tabs on the rest of the world, but also lets you keep track of events at home when you are on the road. Thus can a community of expatriates (e.g., Iranians who live in Southern California), better maintain their isolation from the worlds outside their door and remain united in a common lingua franca of interest. The Net's ability to connect people with the world is what allows them to identify their own communities, archives, and news groups and pretty much stick to them.

The Chinese expression "same bed, different dreams" carries over with fuller force to the evolution of perception. By mid- Century, for instance, the reduced costs of transportation and communications forged a mass American consumer market from a collection of smaller regional ones. Further declines in the cost of communications and computer-driven direct mail promoted subnets which further fractionated it along various demographic, professional, avocational, religious and ethnic lines. Communications at first enables the CNN-ization of perception. Continued evolution results in de-CNN-ized perception and the rise of the same voluntarily isolated communities that pre-date mass consciousness. This time, though, such communities will subset and superset national boundaries and thus the states that govern them. This distinction could matter a great deal to how national security is defined and ensured.

Cellular technologies exacerbate this trend. Text-oriented desk-bound computers are weak devices to maintain communities. Few want to live tethered to a box and bits alone cannot convey the look, sound, and feel that normal human contact requires. With cellular, the network need not be associated with a fixed phone connection and box. Voice commands make keyboards unnecessary. Screens can be built into eyeglasses. The basic box can be shrunk to the size of a hearing aid. You need never be out of effortless touch with your virtual community or it with you. Virtual reality may be so compelling that people need leave only to eat and exercise (perhaps the resemblance of this description to a jail cell carriers deeper meanings).

Thus, although information technology can bring the world together and erase bonds of geography, they also let utterly different communities maintain their identity against assimilation.

The Mutability of Truth: As the amount of information increases, its marginal utility declines, but so does its veracity. Why?

Start with broadcasting. Today's satellite broadcasting systems relay material to terrestrial broadcasting stations which then relay them to television sets. Tomorrow's systems will reach television sets (equipped with 18-inch satellite receivers) directly -- bypassing the investment in both television stations, and the political control that nations can have over transmissions.

Virtually every Third-World village is likely to have at least one such television (even if not on the national electric network) and probably several. Each can watch hundreds of stations, only a few of which will be state run. Video signals from space will be harder to jam. Barring state confiscation or control of such sets, they should be able to get a video signal from any group with enough money to rent satellite space. Just before Desert Storm, President Bush asked for permission to appear on Iraqi television to explain our actions in the Gulf. Two decades hence, his successor will not have to ask. Dictators will be hard pressed to keep rivals off the air. No coup plotter could keep his prey off the air either. By the same methods, conservative regimes will have difficulty preventing the diffusion of the West's best values over the tube: sex and drugs, rock and roll, and guns.

Technology will also make it impossible to distinguish among real and fake photographs, video, or recordings. Anyone with a good machine (tomorrow's Silicon Graphics box, perhaps) could create for broadcast a video of an opponent counseling acquiescence to his people. This image would look and sound like the real thing, being indistinguishable in both grammar, nuance, and gesture. Western audiences, more used to special effects manipulation will grow skeptical of everything on the tube after a while. Third-World audiences are likely to remain credulous targets a good while longer.

These two trends -- the ability to force information past controls, and the ability to create false information -- work both with and against each other. People tend to believe what they want to believe (or what others they fear or respect want them to believe). Contrary reports can be easily discounted, particularly as people come to understand how easy faking a video can be. The same technologies that let people freely experience the world are those that allow people to deny its reality. The resulting cynicism works in favor of people trusting only the information generated by their own village -- not the globe as a whole. Reality is not universally validated but personally validated based on networks of trust.

At the same time, the privacy and authenticity of personal communication is likely to improve. Current mobile phone communications are even easier to intercept than line-based communications are; cellular is generally considered unacceptable for secure communications. Thanks to the digital telephony, public-key cryptography, and free silicon, secure digital communications will need but one cheap phone chip. Encryption will be so easy as to be norm. Such encrypted messages will be unbreakable by any supercomputer. Eavesdropping would have to take place at the source or the destination but not in between. Intercepting signals intelligence as a way of figuring out the what the other guy is doing will soon be useless.

The development of digital signature technology will also lend authenticity to private communications as well. Digital signatures work by having people post public keys which alone can unscramble messages. Successful unscrambling proves that only the person with the corresponding private key could have written it. Such techniques also keep third parties from altering the message without its being obvious. All this assumes that the posted public key is authentic and actually linked to the poster. Such facts may have to be verified, again, though a trusted network -- again, the global villager at work.

Today's virtual reality is far more virtual than real. Tomorrow's, though, may look, sound, smell, and even taste, and feel as real as reality. Information technology alone will not convince the sane that the virtual reality is reality (prosthetic reception devices are one reason why); yet it can convince them that virtual reality is better.

The New Parochialism: Would all this communication among groups hitherto separated by language and geography make people more or less likely to deal with each other in friendly and civilized ways?

Ubiquitous communications could promote a global superclass transcending national boundaries. If this class can define a sufficiently tight set of class interests -- an issue of more-than-academic consideration every since Marx -- transnational warfare may be muted (but perhaps at the expense of class warfare) in some ways. Other less exalted supra-national communities -- linked by bonds of profession, ethnicity, or avocation -- are possible. Whereas such communities have always existed, technology will let them conduct a much larger share of their daily interactions with each other.

Would the formation of supra-national classes make their members feel more solidarity with each other and less with their local community? Would they be more likely to respond to attacks on outposts of their superclass, or would all this communication only remind people how deeply different national origins impress their marks on otherwise similar people? Will inter-ethnic communications lead to greater understanding and thus more tolerance? Conversely, would a little knowledge delude people into thinking that they understand how others think? Internecine conflict is often far less civilized than similar conflict among those who originate from opposite ends of the globe.

The ascendence of Net over Nation could alter what people would fight over. Historically, wars have involved challenges to territorial control -- and not just because everyone has to live and work somewhere. Before the industrial revolution, rural land was the source of wealth. The industrial revolution made factories, infrastructure, and resources -- all of which could be physically seized -- the source of wealth. Even today's post- industrial service economies are tied to place. Otherwise why would so many put up with Manhattan when Maine or the Ozarks would be much more pleasant? Yet, the true assets of Wall Street -- the knowledge, connections, and legally valid financial claims -- are, themselves, place-independent.

As networks expand to enable better remote communications, the validity of holding on to any one place becomes increasingly questionable. A future Hong Kong could as easily be relocated to Vancouver or even to a hundred Chinatowns scattered about but networked together. Singapore has a core competency in bulk materials handling not only because of its port, but for other reasons such as knowing how to conduct intermodal transportation efficiently. Such knowledge could be transferred to any other similarly wired port. Today's multinational heavily networked knowledge-intensive corporation is an increasingly movable feast. That being so, push less often comes to shove, and more often to slide. Data does not even have to be sent ahead at the last moment; it is already distributed to begin with. People need but change their real network addresses; their virtual addresses (the ones people write to) stay the same. The less wealth can be captured by physical possession, the less motivated others will be to use physical means to capture wealth.

The shift to Net from Nation lets communities be knit by constant communications regardless of place. Communities without political self-governance can maintain their cultural mores by establishing their own subnetworks as self-contained universes. Eastern Europe is seeing its fiercest fights over ethnic and linguistic cultural clans used to contesting over limited media space. A single medium suggests a single culture broadcasting its own values in its own language to everyone else -- to wrest oneself free is to band together to form a competing state. Multiple media, however, suggest the support of multiple cultures. As information technology spreads, any group can turn inward in a broader variety of ways -- fostering its networks apart from state or majority interests. True, a believer can be reinforced in his fanaticism by picking and choosing among competing media so that no contrary view intervenes. Such choice could exacerbate the energy of those who sought to impose their culture over others. Yet the ability to carve out separate media spaces also lessens the angst of those who wish only to keep their culture from being trampled upon. It will become increasingly easier to tune out the rest of the world, for better or worse. The expansion of communications (and easy syntactic if not semantic translation) and ability to accommodate separate domains gives competing cultures room to roam without collision. Hence, more sprechenraum, thus less strife.

Otherwise, traditional cultural mores will be harder to maintain in a high bandwidth society. Traditional cultures maintain themselves through the coercion of geography (village life is all they know), custom, and language. Failing that, maintaining group cohesion by coercing less affiliated members (e.g., the restless young and worldly intelligentsia) leaves community-imposed censorship. All these are harder to maintain when anyone can get access to any information. Thus, as Iran's experience presaged, traditional cultures in an urban environment have to become more aggressive about such coercion. Minority subcultures, kept to themselves, posed little threat to the transmission of traditional ways. In the Net, they can create temptations for young of the majority communities. Hence more strife, even though traditional cultures are fighting a losing battle, regardless of how vociferously waged in the coming decades.

The ability of information technology to promote trans- national communities does not mean that every or even most people will become avid members of them. In the West, most people are part of several communities simultaneously: professional, avocational, ethnic, neighborhood (or some are members of no community). Even where they arise, dispersed network communities are unlikely to be so tight as those which live together (cults, for example) and rarely so large as to threaten the state to any serious extent.

Ghosts in the Net

The dependence of cities on networks, both internal and external, creates -- as all dependence does -- a major vulnerability. This vulnerability is likely to take different expression in the West and the South. For the foreseeable future the Net will be more important to Western economies because the West will realize a higher percentage of its value added from Net flows then the South will. Thus, its vulnerability will be greater, and the payoffs from the Net's subversion to private ends will be greater as well.

Yet, network warfare is likely to be most salient in the South, and politicized from the start. Binding Southern cities into the world economic network draws them into a game whose rules are written by the West. The more important the Net is to a city's life, the more a city depends on an external order of things and the more independent assets are from the state apparatus (and thus also from social claims). Networks are also avenues of cultural infiltration. The ease by which information can pass back and forth challenges the social controls exercised by closed systems (a problem that even efficient states such as Singapore will have to contend with soon).

Societies that depend on the Net can be attacked by harming the Net just as industrial societies can be attacked by shutting down electricity. Losing faith in the Net is akin to losing faith in the State. Overt threats against the Net may yield useful concessions. Picking up the right information on the Net can be used to pressure individuals. Subverting the Net may yield illicitly gained resources.

The Net may also be targeted for no other purpose than to return society to pre-Net days. If differential access the Net has too much influence over the distribution of a wealth, losers may wish to change the rules of the game, or failing that, end it. Those who do well may nonetheless resent the power of a non- human system, particularly one, which, unlike the phone system, makes judgements on people's needs. The very notion of a Net that can permit any idea to be exchanged is antithetical to cultures that prefer hierarchical control over ideas.

Networks are thus vulnerable, and totems themselves for attack by forces of the extreme left and right. Future unconventional warfare will target such vulnerability; insofar as the United States supports legitimate regimes, it must find ways of countering this threat. Conflict in the Net would be represented by systematic and organized attempts either to corrupt the operations of the Net or subvert them. The former would strike at the growing heart of tomorrow's urban economy; if people cannot trust commerce over the Net, they would, with no small dislocation, have to revert to earlier systems of commerce whose paths would have become rusty with disuse. To the extent that governance depended on the Net, attacks on the Net, would strike at the legitimacy and effective control of the state.

Attacks on the Net can be categorized at three levels: physical, syntactic, and semantic -- ranked in descending order of risk as casually observed. In practice, the reverse may be true.

Physical attacks on the electronics and wires of the Net (switches, trunk wires, major databases and other key nodes) is certainly possible, but, in and of itself, not a new kind of warfare. Industrial-era targets of the electricity, water, natural gas, transportation, or broadcasting systems will remain equally juicy targets. Moreover, most targets of the Net are both harder to find (because they lack distinguishing physical characteristics), easier to protect (because they tend to be relatively small compared to other key targets), and cheaper to make redundant (particularly the few nodes that hold really critical data). Physical attacks will nonetheless ensue, but society's vulnerability to them can be substantially lessened by appropriate and not expensive measures.

The possibility of syntactic attack -- one which disables the operating logic of the Net and cause it to crash -- is considered very scary. The wars between security forces and hackers will be relatively continuous and they will escalate on both sides (security systems will get better, but new opportunities for mischief will arise, and hackers will get wilier). By and large, however, such attacks will be of minor consequence.

To understand why, start with the celebrated computer virus. Infecting a stand-alone PC requires the user attempt to run an infected program (or what is very similar, try to start with an infected diskette) most of which are bootleg copies of something which, in legitimate form, is mostly safe. Merely uploading a piece of bad data is relatively harmless (for the time being). A computer over the net is a potentially larger worry (because the carelessness of one can infect many), but network operating systems are generally better protected than the operating systems of individual PCs. Indeed, every successive generation of operating systems has security systems increasingly immune to attacks from both remote (e.g., a passed-along virus) or connected attackers. Because most viruses require the complicity of the victim to function, they are unsuitable for anything other than random terrorism. Networks with conscientious users and well-engineered security systems that do not pull programs from the outside are relatively safe. Isolated computer systems are even safer. Thus, the notion of broadcasting viruses to weapons systems, for instance, is specious.

What limits today's viruses is the fact that, although systems accept data from random external sources, they rarely accept programs and only the latter are the venue for viruses. Programs act (and can thus mutate), but data is only acted upon. No data in a well buffered computer can cause the latter to crash either.

Tomorrow's networks will be different, and more vulnerable thanks to four interrelated shifts in how computers are used. Remote procedure calls and object-oriented programming mean that the hitherto safe practice of passing data around will be replaced by the not-so-safe practice of passing data-specific programs around with the data themselves (ironically, object-oriented practices were designed to make computing safer). As office networks expand to campus, corporate and finally to global networks, global direct addressing will allow every byte on anyone's computer to be addressed directly. Tomorrow's 64-bit chips can point to a thousand times more bytes than the world's existing stock of computer-archived data. Finally, tomorrow's networks are likely contain floating filters that roam the silicon prairie looking for game. Know-bots, mentioned above, will be launched by those seeking the ephemeral needle in the infosphere haystack. Auto-filters, in turn, sift through information that others are sending you, inserting some into active programs, bringing others to a user's attention in priority order, and trashing the rest.

In short, tomorrow's active networks are likely to be shuttling, not just data, but code back and forth. The success of syntactic attack -- that on the core operating functions of a computer or a network -- depends on what protections are wired into tomorrow's computers. Tomorrow's computers are likely to be better protected than today's microcomputers (where virtually anything can be altered by an operating program), but networks might allow errant code to travel through the network until it finds an open door, buries itself in code too complex to manually find and waits for an external event to awaken. Here, hackers may actually do some good. Uncoordinated hacker attacks that reveal system deficiencies will be responded to with security fixes that leave basic operations intact. Another and greater piece of cleverness would then be required to conduct a follow- up attack. Coordinated attacks which leave many errant programs lying latent may do considerably more damage.

In general, the more critical a system, the more protected its architecture will be from successful attack. Military command- and-control systems are likely, for instance, to be built around nodes that do not accept code except from trusted sites. Money transferring institutions are also likely to have tough security systems. Digital signatures will be required to transfer money (and only a few people will be able to move really large amounts). It would be foolish to predict that such systems cannot be subverted, but most such subversions will be inside jobs, and, as such, one-shot deals.

However, these four elements can be also be recombined in new and wonderful ways that increase the risk of semantic attack. Tomorrow's networks are likely to see the silicon equivalent of conversations between intelligent agents. Consider remote medical diagnosis between a sensor suite that monitors your health and a collection of doctor modules. The latter assess the data, consult with each other, perhaps awaken a specialist, and, in concert, negotiate a series of actions consistent with your values, life-style, and means. The task of obtaining a loan might otherwise require sifting through a thousand banks each with its own rates, restrictions and criteria. Most people would pick a handful based on sketchy or irrelevant criteria (familiarity, propinquity) and start to negotiate with them. The Net permits launching a thousand requests into the system -- each of which is trained to understand your requirements. Each request, in turn, interacts with a similar software from a bank, with, in turn, its sophisticated set of questions and conditions. These conversations result in one or a few choices, which may be dispatched with the usual character assessment via eye contact, but most of the work will have taken place beforehand.

Note, here, how many soft points can be found in the system. A thousand banks now have access to at least some information about both you and your plans. You, in turn, have information on at least some of the lending criteria of a thousand banks. The Net would feature traveling images of our wants, needs, and resources running around interacting with compatible images of the wants, needs, and resources of others.

The challenge of semantic subversion is that false statements will be inserted into the network as real ones. Systems will be vulnerable until well after the mismatches between various inputs and sensors becomes obvious. Such attacks will affect even those chunks of military or financial systems that collect, analyze, and distribute information: those that which negotiate the transfer of information, create profitable patterns of artificial intelligence, and make assessments about the outside world, where the greatest danger for subversion is possible.

Errant code can attempt fraud at levels that a human would find untenable. Consider the bank loan example. Code can survey more banks in less time than a human. Unless a bank's program has random elements, its logic can be figured out by hitting it with a thousand different cases and looking for patterns, biases, and even flaws. Code can keep a poker face and is undeterred by punishment; thus it can be a much more efficient and determined prober than humans are. Once criminal computer code can be reliably connected to persons, the cost of subversion rises -- thus anonymity is key. Security may, in turn, come to demand a digital signature before a know-bot is accepted into a system. A master list of digital signatures would be correlated with some physical manifestation of a user (e.g., a snapshot, fingerprint or DNA print). The tolerance of Western societies for what is essential a national identification card, however, is untested. Systems that hold signature password owners accountable for damage done in their name must account for users whose passwords are compromised (especially if holding a signature password may not be an entirely voluntary act in a wired society).

Humans have a high bandwidth for input and a low one for analysis; computers are the opposite, they cannot forage for data. If launching event probes into the Net teaches attackers how systems react, they can prepare false events that trigger a false system-wide reaction. For instance, if a nuclear reactor turns itself off should it detect several precursors to an earthquake, false precursors could be fed into various sensors and from affiliated computers to effect a power crisis. However, precisely because such events are predictable and fixed, they can be tested for and requisite sensors can be programmed to weed out such false inputs.

Another source of vulnerability might be created as computers learn how to learn. Today, how computers handle the same data varies little from day to day. Tomorrow's computers, however, are likely to change with experience and adjust to the human vicissitudes of taste, fashion, and circumstance (after all, they exist to serve us). Yet, such adaptability makes them vulnerable to false learning based on a flood of false data.

Return to the medical example. Perhaps incoming case data on a new drug indicates a higher cure rate for its target disease with fewer side effects -- hitherto, say, increased susceptibility to caffeine addiction. The word goes out to prescribe the drug more frequently even though the new cases are false and the new version is very similar to the old one. The result is far more caffeine addiction in the real population, and a sharp loss in the credibility of the medical network.

Return to the bank example. A silicon loan officer finds many potential clients want to have their loans paid into their accounts held by the People's Bank of the Third World, which happens to be absent a list of registered banks. The presence of so much business from that quarter (and the ostensible popularity of the bank) suggests that a subtle change in rules is necessary to win loan business. The program deems such an arrangement acceptable. It later turns out these requests have been manufactured. The bank exists but it is a front for illicit arms transfers. Absent the instant credibility from the misinformed loan officer -- more likely, thousands of equally misinformed ones catching the same traffic on the network -- it would never have gotten off the ground. Within seconds, the People's Bank has real assets to play with.

Fraud, of course, was not invented for computers (resemblance between this example and a combination of Penn Square and BCCI is not accidental). But computers and networks let far more and far graver mistakes to be made far faster. Error, gossip, and fads can propagate faster than wisdom. Computers also lack the ability to read subtle clues in personal interaction that have guided human decision making for so long. Computers, while immune to certain human faults, are heir, particularly when overconfidently introduced in place of humans, to their own psychoses. Such psychoses can be targeted for exploitation. Systems that learn from and react to each other may exhibit extremely chaotic behavior if ticked in precisely the wrong way.

How dangerous would net warfare be? In some ways, more because bad karma will duplicate itself much faster and farther than in human systems. In other ways, less. Corrective lessons can also propagate faster. Simple safety rules can save lives (e.g., never have traffic lights show green in both directions) even if complex systems are biased towards gridlock and waste under ambiguous conditions (e.g., if in doubt, shut power stations down but keep life-support equipment on). Security systems can be isolated from external inputs even at the cost of their being harder to work with (e.g., certain computers can be reprogrammed only on site).

So What?

The Twentieth century has seen large wars result from the alignment of national communities with the state violence. Millions died when Germans fighting for the Fatherland fought Russians protecting Mother Russia. The Net works against such correlation by making it easier for people to be different, letting them pick and choose among communications flows and thus messages.

The increasing importance of spanning communities over local or national ones may be a harbinger of less war -- but not necessarily less violence. True, very dispersed communities, for that reason, cannot easily assemble enough critical mass to take on state power, but, even dispersed, they can do considerable damage. A group turned inward becomes deaf to the message of common discouragement and can potentially more hostile to tenets of civilization. The Net promotes, not insurrection, but greater anomie -- in some cases, group anomie -- but not necessarily at levels conducive to unconventional conflict.

Wherever this information revolution takes the world, the United States will get to first. It is the only large nation in this century where random internal violence has killed more people than wars. If nothing else, the United States may have worked through the problems of national meaning while they tear at others whose nationhood is based on thin cultural or genetic ice.

As national cultures compete for global influence in this new era, the United States stands to gain most. Our language is most likely to become universal, our currency is in greatest circulation, our social culture remains popular (if poorly understood), and our political culture is likely to continue its ascendancy in world affairs. The United States generates more good information, in terms of science, technology, business, entertainment, and thought than any other single country. The American culture absorbs information (just as we have absorbed people) more readily. Those who mine the world's data basins are more likely to hit our nuggets than hit anyone else's. People everywhere believe that American life is attractive. This is no small factor in favor of our national security, and one that information technology cannot help but widen our lead in.

The tension between vulnerability and service will characterize the prospects for conflict in the Net. If the Net becomes part of our expectations of a good life during benign years before being targeted, dependence will grow and security will be an afterthought. Attack would lead to great harm. Too much Net security (perhaps resulting from earlier attacks) may keep it from winning acceptance, at the cost of valuable efficiency. Somewhere in between lies a future in which wise security choices and healthy skepticism yield a Net that can ward off most blows, absorb the rest, and maintain its viability.