Unbuilt

The reels are not random: the first one only holds mechanics with a real track record, the second one only settings that ship far less often with that core than chance alone would predict, and the third keeps the presentation ordinary on purpose. That shape is a direction, not a recipe. On tags set before release, the only tags a concept can have, a typical core with one improbable pairing reached 10 reviews in its first week 15.7 % of the time against 10.8 % for an unusual core with no twist — measured on 2,042 unknown indies, and the gap is not yet distinguishable from zero. On tags refined after release the same gap reads 8.1 % against 4.0 %, which is why we no longer quote it.

What it will never tell you is that being first is an advantage — that was measured too, and it is worth nothing. It also will not sell you an empty triple as a discovery: 51.6 % of all tag triples have never shipped, so that fact is arithmetic, not insight. So every concept comes with the two numbers that do carry: how the core has done, and what happened to the games that came closest.