Stigmergic Notes

I’ve been doing a lot of reading and thinking about the concepts discussed in my last post. It’s a fascinating, if a little bewildering, topic. I’m not sure I have a great handle on it, but I figured I’d share a few thoughts.

There are many systems that are incredibly flexible, yet they came into existence, grew, and self-organized without any actual planning. Such systems are often referred to as Stigmergic Systems. To a certain extent, free markets have self-organized, guided by such emergent effects as Adam Smith’s “invisible hand”. Many organisms are able to quickly adapt to changing conditions using a technique of continuous reproduction and selection. To an extent, there are forces on the internet that are beginning to self-organize and produce useful emergent properties, blogs among them.

Such systems are difficult to observe, and it’s hard to really get a grasp on what a given system is actually indicating (or what properties are emerging). This is, in part, the way such systems are supposed to work. When many people talk about blogs, they find it hard to believe that a system composed mostly of small, irregularly updated, and downright mediocre (if not worse) blogs can have truly impressive emergent properties (I tend to model the ideal output of the blogosphere as an information resource). Believe it or not, blogging wouldn’t work without all the crap. There are a few reasons for this:

The System Design: The idea isn’t to design a perfect system. The point is that these systems aren’t planned, they’re self-organizing. What we design are systems which allow this self-organization to occur. In nature, this is accomplished through constant reproduction and selection (for example, some biological systems can be represented as a function of genes. There are hundreds of thousands of genes, with a huge and diverse number of combinations. Each combination can be judged based on some criteria, such as survival and reproduction. Nature introduces random mutations so that gene combinations vary. Efficient combinations are “selected” and passed on to the next generation through reproduction, and so on).

The important thing with respect to blogs are the tools we use. To a large extent, blogging is simply an extension of many mechanisms already available on the internet, most especially the link. Other weblog specific mechanisms like blogrolls, permanent-links, comments (with links of course) and trackbacks have added functionality to the link and made it more powerful. For a number of reasons, weblogs tend to be affected by power-law distribution, which spontaneously produces a sort of hierarchical organization. Many believe that such a distribution is inherently unfair, as many excellent blogs don’t get the attention they deserve, but while many of the larger bloggers seek to promote smaller blogs (some even providing mechanisms for promotion), I’m not sure there is any reliable way to systemically “fix” the problem without harming the system’s self-organizational abilities.

In systems where many people are free to choose between many options, a small subset of the whole will get a disproportionate amount of traffic (or attention, or income), even if no members of the system actively work towards such an outcome. This has nothing to do with moral weakness, selling out, or any other psychological explanation. The very act of choosing, spread widely enough and freely enough, creates a power law distribution.

This self-organization is one of the important things about weblogs; any attempt to get around it will end up harming you in the long run as the important thing is to find a state in which weblogs are working most efficiently. How can the weblog community be arranged to self-organize and find its best configuration? That is what the real question is, and that is what we should be trying to accomplish (emphasis mine):

…although the purpose of this example is to build an information resource, the main strategy is concerned with creating an efficient system of collaboration. The information resource emerges as an outcome if this is successful.

Failure is Important: Self-Organizing systems tend to have attractors (a preferred state of the system), such that these systems will always gravitate towards certain positions (or series of positions), no matter where they start. Surprising as it may seem, self-organization only really happens when you expose a system in a steady state to an environment that can destabilize it. By disturbing a steady state, you might cause the system to take up a more efficient position.

It’s tempting to dismiss weblogs as a fad because so many of them are crap. But that crap is actually necessary because it destabilizies the system. Bloggers often add their perspective to the weblog community in the hopes that this new information will change the way others think (i.e. they are hoping to induce change – this is roughly referred to as Stigmergy). That new information will often prompt other individuals to respond in some way or another (even if not directly responding). Essentially, change is introduced in the system and this can cause unpredictable and destabilizing effects. Sometimes this destabilization actually helps the system, sometimes (and probably more often than not) it doesn’t. Irregardless of its direct effects, the process is essential because it is helping the system become increasingly comprehensive. I touched on this in my last post among several others in which I claim that an argument achieves a higher degree of objectivity by embracing and acknowledging its own biases and agenda. It’s not that any one blog or post is particularly reliable in itself, it’s that blogs collectively are more objective and reliable than any one analyst (a journalist, for instance), despite the fact that many blogs are mediocre at best. An individual blog may fail to solve a problem, but that failure is important too when you look at the systemic level. Of course, all of this is also muddying the waters and causing the system to deteriorate to a state where it is less efficient to use. For every success story like Ra^thergate, there are probably 10 bizarre and absurd conspiracy theories to contend with.

This is the dilemma faced by all biological systems. The effects that cause them to become less efficient are also the effects that enable them to evolve into more efficient forms. Nature solves this problem with its evolutionary strategy of selecting for the fittest. This strategy makes sure that progress is always in a positive direction only.

So what weblogs need is a selection process that separates the good blogs from the bad. This ties in with the aforementioned power-law distribution of weblogs. Links, be they blogroll links or links to an individual post, essentially represent a sort of currency of the blogosphere and provide an essential internal feedback loop. There is a rudimentary form of this sort of thing going on, and it has proven to be very successful (as Jeremy Bowers notes, it certainly seems to do so much better than the media whose selection process appears to be simple heuristics). However, the weblog system is still young and I think there is considerable room for improvement in its selection processes. We’ve only hit the tip of the iceberg here. Syndication, aggregation, and filtering need to improve considerably. Note that all of those things are systemic improvements. None of them directly act upon the weblog community or the desired informational output of the community. They are improvements to the strategy of creating an efficient system of collaboration. A better informational output emerges as an outcome if the systemic improvements are successful.

This is truly a massive subject, and I’m only beginning to understand some of the deeper concepts, so I might end up repeating myself a bit in future posts on this subject, as I delve deeper into the underlying concepts and gain a better understanding. The funny thing is that it doesn’t seem like the subject itself is very well defined, so I’m sure lots will be changing in the future. Below are a few links to information that I found helpful in writing this post.

• Stigmergic Systems (This was the primary resource used for this post. Well written and very helpful, especially the Primer)
• Self-Organizing Systems FAQ
• Superhuman Intelligence
• Power Laws, Weblogs, and Inequality