Monday, 17 August 2009

Semi-evolved or poorly designed?

Many people believe that man is the peak of evolution, or creation, depending on their point of view. But it seems to me that actually the species is only partly developed, neither quite one thing nor another, but somewhere between the two. Take walking upright, for instance. We can’t really walk on all fours any more, but we haven’t completely mastered walking about on two legs, which is why there’s so much lower back pain around. We seem to be in transition, basically a work in progress.

Now that’s fine if you accept evolution. A creationist, however, would have to believe that this halfway house is the result of design. Don’t get me wrong: I fully accept that you can argue that there is evidence for design out there: the eye, for instance, seems to have been designed for seeing. It’s just that, as I’ve suggested before*, I’m not sure you can call the design ‘intelligent’.

All this was brought home to me again recently when I learned about the concept of a ‘cytokine storm’.

Cytokines are molecules in the body which alert our immune cells – T-cells and macrophages – to the presence of infection and guide them to its source. They also stimulate those cells to produce more cytokines. It’s obviously a powerful mechanism, but a dangerous one because it’s based on a positive feedback loop: the more cytokines there are, the more get produced.

We’ve all experienced positive feedback. It can happen in electronic music when the mike picks up sound from the speakers, feeds it back through the amplifier, and back out of the speakers, louder still, setting the cycle off again. Quickly we find ourselves deafened by a high-pitched wail till someone has the presence of mind to break the loop or cut off the power.

The body generally controls the cytokine feedback loop with specific inhibitors. But from time to time the process gets out of control leading to the explosion known as a cytokine storm. It’s particularly likely to happen when the immune system is dealing with an unusually virulent infection. In those circumstances, a paradoxical situation arises: those with the strongest immune systems, the fittest people, have the worst reaction. A strong immune system out of control is more dangerous than a weak one.

A notable example was the Spanish flu pandemic of 1918-1920 which led to between 50 million and 100 million deaths worldwide. The infection killed not just the young and old, as you’d expect in any flu epidemic, but also a high proportion of 20 to 34 year olds. Because it triggered a cytokine storm, the strongest were often the most vulnerable: young women in particular, and many soldiers thinking they’d got safely home from the killing fields of the Great War, were struck down as their powerful immune systems filled their lungs with macrophages and their lives were literally choked out of them.

The immune system on which they depended and which generally works so well, turned against them, killing those it should have protected. An ugly paradox. And if you’re looking for design, this one certainly seems flawed, wouldn’t you say?

Strangely enough it’s a flaw I know well from my own work, in software development. I work with astonishingly bright people, who come up with complex designs to make sure software calculates its results exactly right. Sometimes this involves combining many different sources of data to build new material for further cycles of processing. Often we only have small sample sets for testing, and that may hide problems that will arise when volumes become big. A full-scale environment may lead to the equivalent of a cytokine storm, with data generating data until memory overflows, disks fill and the system crashes.

Inadequate data is only one explanation of problems with testing. Another may be incomplete documentation, as brilliant designers aren’t always the best documenters. The specification may not be good enough, and that makes QA’s task all the more difficult: testers have to try to work out why a system isn’t functioning properly without full and unambiguous information about what it should be doing in the first place.

All these are problems of design.

Now what are things like in human development?

The lower back pain problem I mentioned before does sound like a bit of a bug, doesn’t it?

And then look at the documentation. The great religious works of the world, not least the Bible, are shot through with ambiguities, gaps, sometimes downright contradictions. How can you build a proper test programme against them?

And that leads to the cytokine storm. Full scale testing would have revealed the problem. But it never happened, did it, and with incomplete documentation, how could QA have done better?

By all means believe in design rather than evolution. But it needs to be a lot more rigorous, and contain a lot fewer flaws, before you can call it intelligent.

P.S. Me and that flu epidemic. My grandmother was engaged to a soldier who made it through the trenches of the First World War, only to be struck down by the flu when he got home. She was distraught and within eighteen months had married another man on the rebound. My grandfather.

Without that flu death I would never have been born. To what curious chances we owe our existences.

* On intelligent and unintelligent design, see:

No comments: