A Break?

  • Epi

It is April 7, day 100 in algorithm 0. The artificial counter is at 1,048,575. What was observed and recorded points to 1,049,378. The agreement is crushing, on this scale, and calculated so far in advance (March 19, when the observed counter was at 244,985), based on such gross assumptions and, perhaps most worryingly, based on a completely stripped-down model that does not take into account (even cannot take into account) conscious behavioral choices, nor possible developments towards ex-patient resilience.

Actually, this is weird: a model that we are sure is wrong, but that predicts unlikely accurately. Friedman would be happy with it. After all, he developed and defended (and successfully) the argument that the quality of a model depends on its predictive potential and not on the credibillity of the underlying assumptions.

We, Mr. Buddy, Stickler, Winner and I (Mr. Node) are not happy with this. Which means that we have to seriously consider how we can integrate assumptions with more reality value into the model behind algorithm 0. This will take time and probably will not yield any entertaining results in this blog. We will hide them in the forum as much as possible. And meanwhile, just show the lighter-hearted thoughts that come up here, at least if they come up at all in the shadow cast by what is currently happening in the UK, the American continents and Africa.