talking about something "at scale" has weird effects to it. if i say something "at scale", you might think i mean largest % of population, but in doing so i sort of gloss over the fact that <10th percentile is the most expensive to address.
eg, "if i can serve 8 out of 10 requests, who cares about the other 2", but i ignore the content of those requests.
idk, my innate capacity to apply stupid linear interpolation and false equivalence to these things seem dangerous. all populations deviate.