The skew is the point though. Startup valuations are supposed to work such that a small number of successes make up for a large number of failures. The press loves to harp on the individual cases, but all that matters for valuation is expected portfolio performance.
It's not a request for evidence. If it were, there would indeed be no leg to stand on.
It's a request to do work. It is a legal compulsion to build something against their will.