Calibration · removed
It never graded anything, so it is gone.
Calibration scored whether past calls turned out the way they were expected to. It needed recorded outcomes to do that, and after six months all 112 decisions still read “pending”. The number it would have produced never once existed.
It refused to guess, and that was correct.
The tool would not score below five closed decisions. It reported that it had too little to say instead of manufacturing a confident rate out of nothing — the same discipline that makes a metric with no reading say “not measured” rather than zero.
That restraint was right, and it also meant the feature said nothing at all for its entire life. Every visit answered the same way.
Removing it takes away a promise that was never kept, not a capability that was working. Something which has never once produced an answer is not waiting for more time; its premise did not hold.
Nothing, deliberately.
There is no scoring left anywhere in StratOps. A decision that turned out badly is marked outdated and the reason is written beside it, which is what a person actually wants to read a year later — a sentence, not a percentage.