Large Entities Viewed as Emergent Behavior Systems

Why do we hate "the media"? Or "the government"? Or "the medical system"? Or other large entities?

When we meet the people in these organizations we find that they're generally good. Certainly there are some bad apples, but in general we find that people's intentions are acceptably good.

And yet somewhere between the individual and the group of individuals we begin to see things that we don't find acceptably good.

Why is this?

My theory for today requires a little background in artificial intelligence. In AI one of the hot concepts is distributed intelligent agents. Think of an agent as a tiny digital bug... a tiny computer program. Individually an agent doesn't have much intelligence. It's filled with very simple rules. But when you throw a bunch of agents together they suddenly do amazing things. Things that weren't expected. Things that weren't programmed.

This is called emergent behavior. The large scale behavior emerges from the small scale interaction of many agents.

The natural analogy is the ant colony. Each ant is an agent... relatively stupid alone. It knows to carry food from point A to point B. But the entire colony carries out amazing behavior like building, defending, attacking, collecting, etc. No individual ant controls the behavior... it simply emerges from the collection of individual agents (ants).

I was asking myself today "why the media is so horrible?" This question was triggered by cnn.com's coverage of the mining disaster that killed 12 people. They wrote it as an anger piece about the company owner who mistakenly announced at 12AM that most miners were alive, only to sadly announce at 3AM that only one was alive. I know that if the company owner hadn't announced anything at 12AM we would have seen headlines like "owner witholds information from families unfairly." The media is becoming increasingly blameful in the worst of crises and it truly pisses me off. There was no ill-intent, yet cnn.com was running what amounted to a hate story.

Why?

I know that the people in the media are acceptably good. (If you like, attack me for believing that people are good... whether or not they are doesn't affect the theory I present in this entry.)

So unlike the media itself I can't just blame the person who wrote the story. That's too easy... and unfair.

I turned my thinking to processes, where it generally goes on these sorts of issues. I've had a belief that the behavior of these big entities like "the media" or "the government" was programmed unintentionally into the process. In other words, while everybody was doing what they were supposed to do, somehow badness came out. Loopholes in the process. Logic problems. In these mental exercises I generally attempt to understand the motivations of those involved.

For example: What does the author want: a story to get onto the home page of cnn.com. What does the ceo want: profits. What does the editor want: proper grammar and a particular political slant. What does the consumer want: news.

I imagined pathways of influence where one would affect the other. And here's where I went wrong: I tried to find a person or a process to blame!

But today I looked at it differently. I looked at each of the members of the media as an agent. Individually they're all doing what they should be doing. They're all good people. But the collective outcome of their organization is emergent behavior, unpredictable by its very nature.

In this light the analysis I did before is still valuable. The processes are the rules of the intelligent agents. But the lessons of artificial intelligence programming show us that given any set of agents we can experience rather unexpected behavior.

Seeing "the media" or "the government" as an emergent behavior helps me reconcile my belief that people are good but the actions of those groups of people are bad. And it frees me from having to find blame.

The people of the media need to see the world in this light too... maybe they'll stop creating blame stories of hate for every situation they report on. "Child dropped off at school... parent's fault? We investigate tonight at 10."

Some of you will likely disagree with me on grounds that these organizations differ from intelligent agent systems because they have leaders... CEOs, presidents, etc.

True.

A couple of answers to that. First, something like "the media" doesn't actually have one leader. Second, the one leader of, say, "the government" (today George Bush) actually has little personal effect on each person (agent) in the organization. The leader's role (AI-wise) is to program the agents with new goals, priorities, etc. But the agents (people) still do the actual work and interact with each other. You still see emergent behavior.

Viewing the effects of an organization as emergent behavior doesn't reduce the responsibility of any one member of the organization to do good. (Although I don't doubt that some day a CEO will use an emergent behavior defense against a Sarbanes-Oxley charge.) And it doesn't mean that we can't search for blame in situations. People (even good ones) still do bad things from time to time and need to be punished.

Viewing the effects of an organization as emergent behavior frees us from the need to find a person to blame. In many situations there simply isn't a person to blame.

Likely some holes here... let me know.