Why'd you have to go and make things so complicated?

"Why'd you have to go and make things so complicated?
I see the way you're actin' like you're somebody else
Gets me frustrated
Life's like this you
You fall and you crawl and you break
And you take what you get, and you turn it into
Honestly, you promised me
I'm never gonna find you fake it
No no no"

~Avril Lavigne, Complicated

Complexity

It is interesting, if you follow the news press, how frequently one finds references to "complex" problems, issues and questions: "Illegal immigration is a complex [...] issue with no easy solution," "Toyota faces complex legal woes as lawsuits mount," "Senate passes complex, controversial energy reform bill," "[T]he subject of radicalization [...] is a complex problem," and so on and so forth. One is left to wonder, what is not complex?

The difference between complex and complicated is that a complex system has new, emergent features that you would not have seen coming from studying its constituents alone. (For the meaning of "emergent", see my earlier post on Emergence and Reductionism.) The complex problem, it can't be decomposed. It can't be reduced. It's global, interrelated, it's on many timescales, and it doesn't respect professional boundaries either. Worse, you don't know were it begins and ends. It's full of "unknown unknowns." It's not only their problem, it's our problem too.

If you need any evidence for the popular appeal to complexity, even the Pope had something to say about it last year:

"The current global economic crisis must also be viewed as a test: are we ready to look at it, in all its complexity, as a challenge for the future and not just as an emergency that needs short-lived responses?"

In a recent article in the New York Times, David Segal wrote

"[C]omplexity has a way of defeating good intentions. As we clean up the messes, there's no point in hoping for a new age of simplicity. The best we can do is hope the solutions are just complicated enough to work."

Calling a problem "complex" seems to mean nowadays to acknowledge one doesn't really know how to handle it. A complicated problem, sure, we'd figure out what to do. After all, evolution has kindly endowed us with big brains. But a complex problem? Our political and social systems can't deal with that. *Shrug shoulders* Now what? Let's clean up the messes and hope that a complicated solution will do.

It is true that the problems we are facing are becoming ever more complex. This is a consequence of our world getting increasingly more and increasingly better connected. This creates new opportunities and fosters progress, but along the way it causes interdependencies and, when left unattended, lowers resilience.

It is however not true that we don't know what to do with a complex problem. We just don't do it. In contrast to our political systems, humans are good at solving complex problems. It's the complicated ones that you better leave to a computer. Look at the quote from Avril Lavigne that is title for this post. She's talking about relationships. Navigating in a human society is a multi-layered task on many time-scales with unexpected emergent features. It's full of unknown unknowns. That's not a complicated problem - it's a complex one. We have the skills to deal with that.

The reason why we can't use our abilities to deal with economic or political problems is simply lack of input and lack of method. These are solvable problems. And they are neither complex nor complicated.

Optimization

I have written previously on what three requirements have to be fulfilled for a system to be able to develop into an optimal state, to find a good solution to a problem. One is free variation. Democracy and a free market economy are good conditions for that. The second one is to detect whether a small variation is an improvement or not. The third is the ability to react to the result of the variation. That's basically a poor man's way to find a maximum: go a step in each direction and take the direction that goes up*.

This procedure however dramatically fails whenever there is either data missing to find out whether a change is an improvement or not, or if there's no way to react to it. Take the recent economic crisis. There have been people all over the place who found something odd is going on; that this money creation out of nothing didn't make sense. They've had the data, but they've had no way to act on it. There was no feedback mechanism for their odd feeling. Way too late one would hear them saying they've sensed all the time something was wrong. From a transcript of a radio broadcast "This American Life" (audio, pdf transcript):

mortgage broker: ...it was unbelievable... my boss was in the business for 25 years. He hated those loans. He hated them and used to rant and say, “It makes me sick to my stomach the kind of loans that we do.”

Wall St. banker: ...No income no asset loans. That's a liar's loan. We are telling you to lie to us. We're hoping you don't lie. Tell us what you make, tell us what you have in the bank, but we won't verify? We’re setting you up to lie. Something about that feels very wrong. It felt wrong way back when and I wish we had never done it. Unfortunately, what happened ... we did it because everyone else was doing it.

Italics added. (We previously discussed this in my post The Future of Rationality.)

It's not that nobody noticed what was going on. There was a variation taking place, but part of the change it was creating wasn't monitored. And there was no way to feed notice about the change back into the system. Computer programs made a risk assessment. They might not have made sense, but you wouldn't question them because everybody played the same game. In a recent NewScientist article, economist Ernst Fair is quoted saying

"Almost everyone in business, finance or government studies some economics along the way and this is what they think is the norm. It's a biased way of perceiving the world."

"Biased" is another way to say there's input missing.

We notice similar failures with other examples. Our economic systems are slow if not incapable of dealing with ecological problems because the problems don't automatically feed back into the system (at least not on useful timescales). There is a variation, but the optimization process can't work properly.

Backreaction

The reason why this close monitoring of the system (our global political, social, ecological systems) has become necessary and why no return to simplicity is possible is that even small groups of humans can cause a significant change to their environment. That may be a natural environment, social, or an organizational environment, which could be summarized as "background". In the earlier days, we were trying to achieve an optimization in a fixed background. Now, we can no longer neglect that we are changing the background by our own actions. In physics, this is commonly known as "backreaction."

If you take for example the deflection of light at the sun, then to compute the deviation you treat the photon as propagating in the fixed background field of the sun. That is an excellent approximation. Yet to be precise, the photon does actually change the background field too. If you'd take heavier objects passing by the sun, you'd eventually come to notice that they do contribute to the gravitational field too. The approximation of a fixed background is often made. For example, for the Hawking radiation of black holes, one commonly neglects the backreaction of the emitted radiation. This, again, is an excellent approximation, but one that breaks down at some point. (In this case when the energy of the emitted particles comes close to the mass of the black hole itself.)

If you are in a regime however where you can no longer neglect backreaction, as we are now with humans living on planet Earth, then you have to find a common solution for both the system and the background. Or you could say, they form a common system. This necessity to find a solution for both the background and the objects in it is one of the great insights of Einstein's theory of General Relativity, where the background is space-time, formerly thought to be an unchanging, fixed entity. You cannot have a time evolution for any system and just look at what the background will do or the other way round. You have to find a solution for both together. It is somewhat of a stretch to the notion of a "background" but I think that we are facing exactly this problem today when we are trying to find a sustainable solution for mankind living on this planet. We can either return to an era where backreaction was negligible and the background was eternally static and unchanging at our disposal. Or we learn how to find a stable solution to the full problem: us and our environment.

This issue is far more complex than you might think. That's because we are now in a situation were the change we cause to our environment does influence our own evolution and adaption to the environment. Human Culture has demonstrably been an evolutionary force since thousands of years already. And we are now only short of actively shaping our own evolution, not to mention that of other species. Whether that's a good idea or not depends on whether we are able to learn fast enough, ie whether assesment and reaction to a change is fast enough so the system doesn't just run down the hill before we can say bullshit.

Bottomline

And that's why I keep saying we need to finish the scientific revolution. Trial and error may have worked well to organize our living together for thousands of years, but this method has its limits. In an increasingly interconnected world, errors are too costly. We need to use a smarter method, a scientific method.

To be able to find a stable, sustainable, and good integration of the ongoing human development into the environment we need first of all to know what's going on. It is not too far fetched to think that Google will play a role in that with creating "real-time natural crisis tracking system," "real-world issue reporting system" or "collecting and organize the world's urban data" (see: Project 10 to the 100). The next step is to find a good way to extract meaning from all this data to be able to react in a timely manner to changes. People often seem to think that with that I mean the systems' dynamics has to be predicted. And let us be clear again that the system we are talking about is the global political, economical and ecological system. Having a model that makes good prediction would be nice, but it is questionable whether this is possible or even desirable. But that is in fact not necessary.

You don't need to predict the dynamics of the system. You just need to know what parameter space it will smoothly operate in so optimization works. You want to stay away from threshold effects, abrupt changes with potentially disastrous consequences. Think again about how we deal with human relationships. You don't predict what your friends, relatives or your partner will be doing. This would be pretty much impossible. But after you have got to know them you'll have an idea what to expect from them, and you'll be able to maintain a sustainable relationship on a balance of taking and giving. The same holds for the systems that govern our lives. You don't need to predict their evolution. You just need to know the limits. Life's like this...

---

* This does not find you a global maximum, but that's a more complicated problem that we'll discuss some other time.