Five years of experience is a lazy metric

Update on entry 23-2’s twenty-million offer: registered, and probably filtered out already. The questionnaire asks for “at least 5 years of experience with artificial intelligence” ▸ 17:14, and answering honestly means answering wrong.

The critique deserves its own record. Time-of-experience is a proxy so lazy it measures almost nothing: a year of unemployed, all-day, paper-reading immersion can equal three years of someone studying exhausted weekends ▸ 17:33. The friend studying Chinese only on Saturdays has, after a year, not reached HSK1; two intensive hours daily got there in months. And the AWS requirement: becoming functionally expert is a month of casual study, a week and a half at full dedication ▸ 19:33. Duration is what recruiters can verify; intensity is what actually compounds. The metric survives because verification beats truth.

duración se verifica, intensidad se nota →

Then the offer itself dissolved under inspection: the message came from Torre AI’s bot, not a recruiter, likely clickbait to reactivate dormant accounts from two years ago ▸ 20:39. Their survey promised five minutes, burned one minute on 10%, and got abandoned half-answered: why put effort in if they didn’t? Symmetry as a filter for opportunities.

The closing confession is the best data point against the whole edifice: the exam for the last job was passed with ChatGPT open beside the screen. They never knew. And the job got done well ▸ 23:05. Using the tool didn’t predict failure; it predicted how the work would actually be done from now on…