The "market for lemons" paper was not about the average quality of the market---it was about the selection effects generated when market participants cannot judge quality. Within a online labor market, there are platform-imposed features that reduce information asymmetries & make relative comparisons possible e.g., tests, feedback, verified records of billed hours. The informational problem is much worse without an intermediary, which is why these markets exist.