Optimising for the wrong objective
Screening protocols are tuned for the highest possible signal on day two. That optimum frequently sits just above the toxicity cliff, which is harmless when the cells are discarded at day two and fatal when the plate is destined for two weeks of selection. If the endpoint is a stable line, the optimisation objective should be viability-weighted efficiency, not raw signal.
The same inversion appears at scale. A 24-well experiment optimises the reagent-to-DNA ratio for expression; a 100 L bioreactor is limited by mixing time, plasmid cost and the plasmid ratio that maximises functional titre. Choosing the bench-winning condition can actively reduce process yield.

