RECOMPARE

Change the baseline or comparison class behind a judgment, because every evaluation secretly depends on what it is being compared to.

Your 5K time came in and you feel deflated. Compared to the runners posting on your feed, you're slow. But that comparison snuck in without your permission. Compared to your own time from March, you've improved 90 seconds. Compared to not running at all, which was your actual alternative this year, you're in a different league. Every judgment secretly leans on a baseline, and yours was chosen by an algorithm.

RECOMPARE means deliberately changing the comparison behind a judgment. The cue: an evaluation depends on an invisible or misleading contrast. Generate the standard set: compare against your prior self, against genuinely relevant peers, against the feasible alternative, against doing nothing, against the best achievable, and against the full distribution rather than just the average. Then choose the contrast that answers the decision you actually face, not the one that flatters or punishes you. If the question is "should I keep training this way," your March self is the right baseline; strangers online are not. It works when the ranking or interpretation changes for a reason you can explain. The failure mode to avoid: benchmark shopping, cycling through comparisons until one tells you what you wanted to hear.