This Wired article got me riled up today. Of course I encourage you to read it, but I'd probably summarize it even if I did think you would.
In essence what it's saying is, using Google's statistical analysis approach to marketing as an example, we are getting to the point where we have so much information available to us that we no longer need to generalize when we talk about "the way things work". Scientific theories and models, schemas and mental archetypes, all approximate reality in an inaccurate and ultimately incorrect way. What we ought to do, says the article, is work to describe things as they actually are, now that that's becoming more possible. Throw off the mentality that we need to theorize about causation, and focus exclusively on correlation. Use data and statistical summaries to measure and essentially report the measurements, not our interpretation of those measurements. With enough data, our mathematical summaries could be descriptions of reality in a much more truthful sense than our verbal ones could be.
The article suggests that science and the scientific method are becoming outdated; they're tools of a technologically inferior age, where estimation was necessary.
I find this notion distinctly unsettling, partly because it seems largely true. Why should you talk about the "circulatory system" if you can talk about where a cell is, where it's (statistically likely to be) going (based on where it's been) and where it's been. Why generalize when you can be specific? Part of the response could be that we simply can't handle such specificity as humans. We need schemas.
But computers don't. Computers can be accurate in gigantic and minute detail. Do we need to talk about why things happen at all? Can't we just talk about what happens, and what is likely to? Isn't that the point of explaining things in the first place? Prediction? Why do we need to understand the "grand scheme of things" when we can accurately and quickly describe exactly how things are?
Partly, I would seriously lament the absence of curiosity. If our understanding of the world was as simple as asking where a thing is and will be without really caring about where it's been (since that's the computer's job, not ours), there's a huge portion of appreciation missing. Without caring about the "big picture" I have a real fear that things will spiral into a world of self interest and economic morality. That does sound a little funny coming from a Utilitarianist.
Curiosity and respect are important parts of humanity, necessary for moral and responsible lives, as well as personally fulfilling ones. The anti-scientific pro-statistics outlook sound to me like a marketing firm's dream, where perspective doesn't matter, only numbers and profit do. Now, I'm the first to say that emotion and beauty in all their glory can indeed by reduced to statistics and ratios. But I don't think we function that way cognitively. Part of reductionism's appeal is it's predictive potential, but part is it's ability to open vast chasms of respect and wonder for the intricacies of our world. If we see those intricacies but don't marvel at them, I feel like we've lost something. In order to marvel and understand, numbers have to be simplified into some sort of recognizable system. One that can be compared to others. Awe is a product of exploration, not description.
There's another problem I have with this approach. The world does work in systems and rules. I have a very hard time letting go of this idea. It's all around. Things are predictable. There is some sort of rule to the way things are. At some core, there is a model to be described. Without one the universe would function as a chaotic, entropic body of effect with no cause. We observe that this is not the case. At the very least the human world, the existence between our ears, is not one of fleeting, momentary description. We thrive of making connections between one experience and another, and require to some extent cognitive schemas to stay sane and function day to day.
So although it is true that there really aren't "recessive" genes, there isn't really a weather system that moves accross the globe, and most of our existing models for human behavior or ridiculously oversimplified, that doesn't mean that there aren't rules and models and structures to the world. Rather than thinking that computers will do away with generalizations, maybe we should embrace computer data as extremely complicated and deep generalizations.