Engineer finds quality gate weights irrelevant after hard excludes fix output pool

A software engineer discovered that reweighting scoring signals ten times in their content pipeline produced no change in which posts survived, with the same 208 out of 399 posts passing every time. The finding revealed that hard exclude rules — not weighted quality scores — were solely determining which content made it through the gate. The weights only reordered an already fixed shortlist, meaning a week of tuning runs had no downstream effect on the surviving pool. The engineer concluded that adjusting hard exclude thresholds is the only lever that actually changes which posts qualify. They now recommend others running similar pipelines to sweep weights and diff surviving pools before investing time in weight tuning.