Exploding brains at warpspeed

Have you ever read something that was so…amazing…that it broke something inside of you? Afterward, your brain seems to explode and implode at the same time and before you go into to shock…before the adrenaline dries up…you become acutely aware of the fact that eventually you’re going to have to clean up the gray matter mess you’ve created, but perhaps there won’t be any time.

This Wired article on the potential of sheer information to obsolete the need for models, theories, and therefore, the scientific method itself, seems to kinda do that to me.

I, for one, like theories, models, hypotheses, and the like. It goes to my philosophy…I think that the most important thing to know is why something happens. This why, however, incorporates a whole lot of other useful details…when you know a why relationship, then you also know what will happen, when it will happen, and how it will happen.

In fact, I was talking with someone about this just the other day…I asserted that something deep down inside, some logical part of us, wants to believe that we can make a bite-sized theory for everything. We want a Theory of Everything that will encompass all possible situations.

Unfortunately, we haven’t seen too many of those. Instead, as George Box points out, we find that “all models are wrong, but some are useful.”

So even if Newtonian physics doesn’t *always* work (and quantum physics requires dimension gerrymandering), at least they are useful for us to sort out data.

But what happens when we don’t need to sort out data? We don’t need to pick and choose what data is relevant and irrelevant (or…which correlates or causates and what is noise) because now, we can just chuck it all in a computer and cast some math on it. That’s what Google did to become king of search engines…and they’ve been incredibly successful, obviously.

The new theory…so to speak…is that we don’t really need a theory or a model. As long as we have the capacity and capability, we can just take all the data we can handle and then we use that. It’s like sampling in statistics…the only reason we really sample is because we can’t handle the full population in many times. Sampling gives us close models, but obviously, the entire population has no outliers, no skew, etc.,

I don’t know. It seems kinda grim.

But I guess I, for one, welcome our new robot overlords.