Covolution is evolution written as a composition of five operators whose composition is itself retained and rebuilt; whether that composition is real is what convalidity tests.
Convalidity is the test of whether the parts of an explanation add up the way the explanation says they do.
Suppose you explain how a cell adapts by naming four separate processes: it produces options, some options survive, it stores a record of what survived, and it uses the record to rebuild how it produces options next time. You measure each process on its own, by changing one and holding the others still. Now you change two at once, and before you look at the result, you predict what will happen using only your four separate measurements and your stated rule for how they connect. If the prediction comes true, your explanation has convalidity. If it fails, or if you have to adjust the rule after seeing the result, it does not.
Three things make the test fair. The prediction must be written down before the experiment, not fitted afterward. The prediction must be different from the lazy guess that "two changes just add up," because that guess needs no explanation at all. And a pass counts only for the combinations and the kind of system you tested; passing in yeast says nothing about bacteria, or about software, until you run the test there too.
Why it matters: a good vocabulary can describe any result after the fact. A causal explanation has to say what a new combination will do before anyone has seen it. Convalidity is the difference between the two.
댓글 0