This will be the natural spawning ground of Digital Entities (DEs), because it is here that digital power will be applied
with the utmost of force and ferocity. It is here that the fate of civilizations hangs in the balance and so the powers
that be will compete in this realm and will apply the necessary resources to win in this realm.
The importance of cyberspace for nations cannot be overstated as we are now fully dependent upon information systems,
as described in a recent White House report: "By 2003, our economy
and national security became fully dependent upon information technology
and the information infrastructure. A network of networks directly supports the operation of all sectors of our economy—energy
(electric power, oil and gas), transportation (rail, air, merchant marine), finance and banking, information and telecommunications,
public health, emergency services, water, chemical, defense industrial base, food, agriculture, and postal and shipping. The
reach of these computer networks exceeds the bounds of cyberspace. They also control physical objects such as electrical transformers,
trains, pipeline pumps, chemical vats, and radars."
The Chinese are otensibly the most tenacious in the current realm of cyberwarfare (circa 2007), with stories circulating
of their having repeatedly probed US defenses. Of course, every nation of sizable resources and wealth has need to protect
its infrastructure and its military and financial systems and as such will have made some investments in both defensive measures,
and presumably in offensive measures, if for no other reason to than to gameplay the likely challenges to its defenses.
Link to Recent Alleged Chinese Efforts:
Symbiosis: NSA and Neural Architectures
The challenge in cyberwarfare is to search for and exploit vulnerabilities in the target systems, which will include
transit systems, banks, military installations, the power grid and other key national assets. Equally important, but
less conspicuous, targets are the human agents of these organizations, as they provide a ready means across all known firewalls.
In peacetime, the predominant activities will be espionage and infiltration, including the mapping of e.g. US information
systems, planting backdoors and the like. In wartime, disruption of military and economic assests becomes the primary
objective. Because the US and other first world countries are now entirely dependent upon information systems, the pressure
to field the best cyberweapons becomes critical.
The 3 Levels of Digital Agent Advancement
PR = Pattern
NLP = Natural Language Processing
Key areas of importance in this realm are systems penetration (hacking, password finding and guessing), pattern
recognition and natural language processing. Purely digital and meaningful Natural language Processing agents
will have the means to go far beyond the P and PR levels to understand systems architectures and functions and then identify
and control the relevant human and digital resources needed to access and control those systems. In order to best consider
how NLP agents will extend beyond current generation cyberweapons, we need to consider current generation approaches and goals
and identify their key limitations.
P = Penetration. Penetration entails the finding of passwords or routes into systems
that are intended to be off limits to all but select users. Select users is the key operative term here, because all
systems of any importance have two key ingredients:
(1) the fact that they carry out functions deemed important, and
(2) their use by human agents.
Penetration currently depends (I hope) upon active control by human agents that are looking at specific targets
with a specific goal in mind (disrupting, seizing control, planting backdoors). A key to building increasingly powerful
cyberweapons is to increasingly automate these tasks so that DEs can select targets, and select penetration approaches-- and
ultimately select goals. Currently humans need to identify the goals and targets, and presumably the strategies to locate
the targets. Current programming is perhaps limited to guessing passwords, trying combinations, and perhaps looking
for alternative routes into a target. Certainly, DEs are not currently able to, on their own, identify human controllers
and find means to alter their behavior. But human cyber attackers certainly can employ a variety of automated strategies
in their efforts to breach specific defenses. As I am not a hacker, I am not privy to the tools and strategies used
at this level of cyberwarfare.
PR = Pattern Recognition. Patter Recognition entails the derivation of key bits of
information that would be useful in various cyberwarfare activities ranging from the locating and penetrating
of systems to the identification and manipulation of human controllers. Pattern Recognition is a routine task in
AI applications and so is presumably a widely-used mechanism in espionage and law enforcement. Where digital agents
can become more powerful is by means of detecting more subtle patterns, using a wider variety of sources of information.
Most conventionally, this would entail the examination of communication patterns, travel patterns or other patterns that concerned
human activities, e.g. identifying potential terrorists and terrorist plots. Such work employs computers mostly
(I'd guess) for lower-level kinds of pattern recognition, e.g. phone calls or emails passing between two locations, usage
of words and word combinations, etc. Low level pattern recognition is perhaps easily implemented.
But patterns can come in many forms, including forms that are not obvious to people involved in the operation of key
target systems. A classic example of pattern recognition (by humans) occurred during WWII in which german signal man
were identified based upon "fists", that is their subconscious patterns they used when transmitting morse code signals.
It was extremely valuable to allied forces because it enable the tracking of units of different sizes and provided other information
of both tactical and strategic importance. We cannot at this juncture estimate the kinds of "fists" available in cyberspace,
but the creation of agents that could look for myriad kinds of subtle patterns could generate unique kinds of information
that would be totally unsuspected by the targeted organizations and individuals. Especially valuable would be digital
agents that could combine many diverse modalities of information: phone calls, emails, credit card usage, web browsing activities,
travel data, etc, and find subtle relationships that no human could detect, especially since humans cannot scan billions of
pages of information per second. The raw brute power already exists to make DEs of frightful power -- where
we are lacking is in teaching machines to think.
NLP = Natural Language Processing. NLP entails the use of language understanding
to advance diverse goals including the abstraction of information from diverse sources about target locations,
functions and human controllers. NLP goes beyond such rote functions as highlighting communiques with words like
Jihad and Infidel. It would instead process language sources (email, phone, web, etc.) to create hypotheses about the
nature of the actors, their likely activities, their personal and business networks and their intentions. It
would then test these hypotheses by researching further into the information available on such entities, and by means
of comparing past cases of (in this instance) terrorist behaviors and events. Importantly, NLP builds upon advances
in pattern recognition by using the best recognition/learning tools to probe for patterns in a linguistic meaning realm. More
importantly, NLP does this on a scale that dwarfs the mind of any human operator, allowing it to pull out extremely subtle,
yet critical relationships, between humans, between organizations, between times and places, between topics. Once digital
entities attain skills and understanding of NLP at levels comparable to humans, their vast speed and reach will put all human
agents at such a disadvantage that we may seem like ants or toys to them. If they have e.g. identified all the major
players of an organization and their likely roles, probed their behaviors and home computers, it is likely that by one or
more means of human manipulation they will find new and unpredictable means to engage a target.
tbd: Ways of Manipulating Humans:
References on Cyberwarfare
Here are just a few sources collected from a brief navigation of this topic.
From a US War College Page:
Entire Bibliography is at:
White House. The National Strategy to Secure Cyberspace
. Washington, D.C.: February 2003. 76pp. (QA76.9 .A25U57
2003) Also available from <http://www.whitehouse.gov/pcipb/
>. Internet. Accessed 29 May 2003.
Cordesman, Anthony H., and Justin G. Cordesman. Cyber-threats, Information Warfare, and Critical Infrastructure Protection:
Defending the U.S. Homeland. Westport: Praeger, 2002. 189pp. (UA929.95 .E43C67 2002)
Rattray, Greg. Strategic Warfare in Cyberspace. Cambridge, MA: MIT Press, 2001. 517pp. (U163 .R28 2001)
Williamson, Jennie M. Information Operations: Computer Network Attack in the 21st Century
Research Project. Carlisle Barracks: U.S. Army War College, April 2002. 22pp. (AD-A402-018) Also available from <http://handle.dtic.mil/100.2/ADA402018
>. Internet. Accessed 28 March 2003.
These links more concern tactical operations:
Ackerman, Robert K. "Army Cyberwarriors Prepare for Broader Future." Signal 56 (March 2002): 23-26.
Ackerman, Robert K. "Information Technology Drives Tip of Spear [in Iraq War of 2003]: It Takes a Network to
Connect Far-Reaching Attack Assets." Signal 57 (June 2003): 17-22.
Ackerman, Robert K. "Intelligence Technology Development Accelerates." Signal 56 (June 2002): 27-30.