JOURNAL: Stochastic tinkerers and warfare

I really respect the work of the financier, Nassim Taleb. His ideas on tinkering networks (the same type of networks that brought us the airplane, the personal computer, and much of the software we use today) provides us useful lessons on the utility of open source warfare (for those new to this, here's an article from the New York Times to get your started).

Here's how. Warfare in our current complex environment (as opposed to the last century and earlier) is very similar to the areas of science/finance where stochastic processes dominate. Since stochastic dominance implies a high level of randomness in outcomes, tinkering networks (ie. open source insurgencies) tend to generate substantially higher returns on effort than highly planned activities (ie. nation-building). The reason is that if you can't plan outcomes due to randomness, the optimal approach to success is through parallel development efforts using a wide variety of methods in combination with a network that readily embraces unexpected but very useful innovations. In contrast, highly planned efforts tend to limit the number of methods/paths used and are resistant to errant results due to bureaucratic inertia/bias.

Finally, these tinkering networks can occasionally produce black swans, or radical breakthroughs. In the context of warfare this is either an event or improvement in method that changes the course of the war. Question: are we, or can you ever imagine us being, in the business of producing black swans in warfare?

Comments

[black_swan]

Although I agree with your comments on "tinkering networks" small iterative change is better than stiff plans, I don't quite follow your reference that quantitative finance is like open source warfare. This would imply to me that (like quant finance) only large actors that have very small transaction costs (in relation to the investment size)can efficiently exploit a statistical "edge".

another point is that taleb's main argument about this kind of "tinkering" is that you don't know if the result you get is because of luck or true causality. i.e. "fooled by randomness"

[John Robb]

True in regards to quant finance. If war could only be reduced to metrics in such a fashion...

Meyer puts it in a way that is probably more applicable:

http://www.helsinki.fi/iehc2006/papers1/Meyer.pdf

[John Robb]

This journal entry puts this story into perspective:

New York Times: "Chaos overran Iraq plan in '06"

http://www.nytimes.com/2007/01/02/washington/02war.html?ex=1325394000&en=0c2ab1e055ce6c16&ei=5088&partner=rssnyt&emc=rss

[scalefree]

Warfare may not be reducable to metrics but I believe that innovativeness can be. There've been several studies that demonstrate a direct link between connectivity within a system & capacity for innovation. We've only begun to learn how to exploit the network properties of social systems, but people are already using SNA & ONA to create specific effects like this.

As for being in the business of creating black swans, we hackers have been doing that for quite a long time now. I expect that trend only to rise when we can super-charge our capabilities with SNA-enhanced forums & project management systems. When hackers really turn on to the power of SNA, it's gonna be a whole new ball game.

[Valdis]

> the power of SNA...

Although the analysis is overly simplified, this is basically the primer for the black swans that John and scalefree mention...

http://tinyurl.com/ympu24

[enigma_foundry]

"Since stochastic dominance implies a high level of randomness in outcomes.."

Perhaps this is not true. I would think that there exists functions which, by increasing complexity, can lower the randomness.

"Finally, these tinkering networks can occasionally produce black swans, or radical breakthroughs."

This is possible by changing the rules the small scale parts of a system (cellular automata, or, in human society, individuals) relate to each other. In automata, for example, if you change the rules which determine whether an adjacent location lives or dies, you radically change the outcome from a given starting state. People, as autonomous decision making agents, will alter the rules by which they interact if they are given diffferent information, which will determine how the chose to interact.

"In the context of warfare this is either an event or improvement in method that changes the course of the war. Question: are we, or can you ever imagine us being, in the business of producing black swans in warfare?"

The lever that you need to use in warfare, is obviously, information. People informed differently will act differently. If for example, everyone believes that their neighbor is going to cheat them, will be unlikely to trust them and do things which relie upon that trust relationship. The example would be the basic level of ethics which is expected to exist in any society, and upon which many business decisions are based. (Amartya Sen has written on this, as have others, but he is current and his work subsumes the best of the prior works.)

Another example would be: if I belive a dreaded and communicable disease is out there and threatening me, I may be reluctant to do certain things I would otherwise do. By the same token, if, in the course of a disease outbreak, I belive that a certain network is providing me with useful, verifiable, and timely information, I will adhere to that network.

Thus, through complete transparency and rapid dispersion of accurate information, networks can be built to which individuals, acting in their own best interests, can be expected to maintain and adhere to.

Networks that provide inaccurate, self-serving, un-timely information will be abandoned.

[John Robb]

Thanks Scalefree and Valdis. That seems like a book in the making: "manufacturing black swans"

Enigma, the equations become less random with more investigation and ownership (where added complexity pays off) of the variables involved. You can make it relatively predictable. However, that takes a lot ot time. It also assumes that you are not working against those that would endeavor to add unknowns to the equation and the environment stays releatively stable. Of course, if those assumptions are invalid, you will begin to see negative returns for your investment in complexity (ala Tainter, which in his model, if repeated across a wide number of problem areas, means that you have reached the limits and eventual collapse of civilization). Whew, this is esoteric!

[enigma_foundry]

"The lever that you need to use in warfare, is obviously, information. People informed differently will act differently.."

I should have phrased that differently. There is more than information, but there is also beliefs, and those should be kept separate for the purposes of this analysis. So, information is processed according to a rule-based system. Rule-based systems can evolve complexity spontaneously. (Just look at some good sites which explain cellular automata, aka von Neumann machines)

"Whew, this is esoteric!"

But the rubber meets the road rather quickly, when you are studying it with a purpose.

[scalefree]

Valdis: although the paper you link to does describe one specific black swan, systemic disruption through SNA-enhanced targeting, what I actually meant was the idea of using SNA to tune the information-sharing capacity of a social network as a means of encouraging the discovery & exploitation of disruptive technologies (aka black swans), which is something we hackers have historically shown a great facility for.

John: I'm still trying to wrap my head around distinguishing between "good" & "bad" complexity ala Tainter. I have a gut feeling it's connected to what David Brin calls "centripetal vs centrifugal social phases". Which means that what we need to concentrate on is centripetal social technology; reputation systems, innovative voting systems & other information aggregation systems.

(See http://davidbrin.blogspot.com/2006/12/todays-centrifugal-net-is-not-arena-or.html & http://www.davidbrin.com/disputationarticle1.html for more).

[french swede the rootless vegetable]

I can definitely imagine one: Iran.

What with the "surge", everybody seems to think war with Iran is inconceivable.
The purpose of this reinforcement could be not to escalate combat in Iraq, but a feint that allows a troop build-up, and eventually a casus belli through 'Iranian/Shia interference with the "surge"'. If the mullahs go for the bait, they become the aggressors, which is generally a much greater geostrategic handicap than any benefits from strategic, operational, or tactical initiative, as shown magisterially in Mesopotamia.
(now, if the Bush admin. could also learn about the *ethical* reprehensibility of starting wars of aggression...)

In other times and places, it would have been of "peaceful defensive manoeuvres near the border" instead of a "surge".
Then, we'd see what you wrote about collapsing Iran through effect-based operations.
That should tear up all those irksome social Terrorists networks tangling the Gulliverian US, maybe getting them to fight with each other instead, like when you were a kid and started ant wars by shoveling an anthill on top of another. Stigmergic warfare!

Organizational Network Analysis this, Al-Qaeda!
And Social Network Analysis that, Hezbollah!

Then, forget about those crummy loser wars of Afghanistan and Iraq.
Ladies and gentlemen: a brand new War in a brand new theater, stretching from the Khyber pass to the sands of the Arabian deserts, where we can start a brand new Struggle against brand new Terrorists!
Les stochastiques sont faites, rien ne va plus!

[scalefree]

Just wanted to add a couple data points I didn't get around to finding earlier. "Innovation flow through social networks: Productivity distribution" at http://www.citeulike.org/user/nettraq/article/716 & "Modelling Selforganization and Innovation Processes in Networks" at http://www.citeulike.org/user/nettraq/article/593 .

I also just ran across this tidbit, a less formal analysis that I think speaks to the same issue but from the downside: http://www.guardian.co.uk/g2/story/0,,1984956,00.html .