Friday, September 27, 2002

Blinded by science

I'm getting some feedback that suggests that some of us have a fundamental misunderstanding about science. IMO I'm picking up a heavy load here, but let's see what happens.

Plato famously made a point with his "cave". In this, people were chained facing a wall, unable to turn around. Things were going on behind them, but all they could see were the shadows of them on the wall before them. Whatever Plato's intent was, mine is to get across the idea that our perception of nature is similar - what we can know of the world is distorted because we have distorted information, filtered through our senses and our instruments.

Another example would be the blind men and the elephant. One found its trunk, another its leg, another its side, etc, and each drew an incomplete conclusion about what an elephant was. Here the idea is that our perceptions can be all too narrowly focused on aspects of a phenomenon which are inadequate or even misleading wrt describing its whole.

Are scientists thus "blind"? Yes, like all the rest of us, in certain respects.

Well, there's no denying that science has had smashing success. So let's suppose we ignore those problems and gather data anyway - now what? We have to decide what data to collect. That requires a hypothesis, which introduces yet another source of distortion. What doesn't get hypothesized doesn't get investigated. So if scientists are politically limited, internally or otherwise, they can remain blind to discoveries that are obvious to less constrained investigators.

OK, now you've collected data. It alone isn't so useful. It is given value by being organized into models designed to behave like the phenomena being studied. Then these models can be used to interpolate values where no direct data exists (in 1995 the population of Chicago was X), or to extrapolate beyond the range were data was available (in 2100 the population of Chicago will be Y). I'm showing my bias toward the "hard sciences" here, but corresponding processes exist where the data is less quantitative.

Engineers like myself can testify to the great usefulness of these models in innumerable situations. But these models impose yet another form of blindness or distortion. They might apply to only very limited circumstances. They take into account only the phenomena that appear in the equations. Other yet to be discovered phenomena might be emerge, or be significant under other circumstances. We may improve our mensuration, thus exposing flaws in the model that were not evident or significant previously. And mathematical models necessarily have the same limitations of mathematics itself (you can't divide by zero, for instance).

If you were successful to this point, terrific, but it ain't science yet. Now you must suffer peer reviewers. Peer reviewers are human beings, complete with biases, blind spots, reputations, and political agendas. Woe to you if your discovery is unwelcome. It can take years before the fogeys give way for a new theory, and entire books have been written about this.

You made it this far? Now we get to the interpretation, where entirely different kinds of errors arise. This is especially true in the hands of amateurs like journalists.

One of these errors arises in application. Models ordinarily are limited in applicability to a range of circumstances. For instance, you can use the "ideal gas law" for gases, but not for liquids (you'd think that's intuitive enough, but I've corrected alleged engineers who thought otherwise). Even that law isn't good for all gases under all circumstances. And beware of extrapolations.

Another type of application error which is depressingly common is overgeneralizing. I don't know how many times people have told me seriously that Einstein's General Theory of Relativity shows that "everything's relative", morality and all. Or they start with microevolutionary changes in, say, the distribution of colors of some moths as proof that all life on earth came from some primordial protein molecules coming together by chance - maybe they did, but you'll need more and better evidence than that.

One of the worst interpretation errors is to confuse the model for the phenomenon. Nature doesn't obey our scientific "laws" - it does what it does . We write our "laws" to accomodate what we observe, subject to the limitations of our observations, our mensuration, and our capabilities for mathematical reasoning. Any conflict between what nature does and what scientists predict is always the scientist's fault. And we can't ever be sure but that tomorrow something new might come along and force us to revisit it all.

Despite it all, science produces results which we engineers use to make life better for all of us.

But science does not give us Truth. To that it will remain forever blind.

No comments: