That was a very neat article, extremely well thought out. I was particularly surpised by his statement about "triple blind" experiments becoming a standard way of performing experiments. We already do this for some types of experiments, particularly longitudinal studies, but the problem is that they can't really determine causality. Kelly's right - with additional computer power, that may not be a problem in the future: there would just be so much data being analyzed so quickly that we might be able to zero in on causal factors. I also liked his idea of mutliple theory matrices - very clever, and, very likely to be used in the future of social sciences, or as preliminary studies for actual experimentation.
But I do have some beefs with his ideas. First of all, he predicts increasing complexity in science, which is the way a layperson might view science, but it's not what actually happens. Science is trying to make things simpler, not more difficult. I don't see why scientists would want to deal with theory matrices. It just overcomplicates their job, and their job is to de-complicate things. Sure, these kinds of fuzzy methods will probably be used as a first-pass analysis before focussed research. But... actual research? I don't see it.
I also don't see people surrending the pursuit of science to computers, as Kelly implies. He predicts we will let AI expert systems do analysis and deep simulation for us, which is probably true to an extent. But, he doesn't seem to grok how technology is actually used. For example, I now use a computer to do structural and mechanical analysis on the machines I design. But, I do not trust the computer results. I am ultimately responsible if that machine fails and someone gets hurt or killed. So I use the computer as a tool to assist me, and check everything. Yes, it speeds up my work, but it doesn't replace it. I still have to be able to replicate everything myself. Usually, I do a back-of-the-envelope-calculation to get the rough answer, then use the computer for the real accurate answer, and try to understand why the two answer differ (if they do). Usually the computer gets the same answer as me (except far more accurate), but sometimes the computer misses something important - and sometimes I just haven't programmed it correctly.
So someone is still going to have to manually check everything the computer does, so the computer can't get too complex, or we won't be able to. (Of course, we may start finding ways to use different types of computers to test each other.)
The other thing I think Kelly completely missed is... nanotechnology. Nanotech will be a huge step in the fusion of the technological and the biological, blurring the line between what is biology and what is machine. It will - in time - eliminate disease, hunger, and may extend our lifespans indefinitely. Poverty will also be elimitated because of cheap "manufacturing" - everyone will have a fabricator in their home that can make anything they need. That probably won't happen in 50 years, but by the time 50 years is up, you can bet nanotech will be not too far away.
I thing it was Socrates who once said that the only valuable knowledge was that which make us better. so as long as this new imporvement in science bring us to be better off than before I will certainly welcome them. But if they involutionate us I thing they are not a good idea.
Very nice article, thank you for posting!
