An VC only needs k of N start-ups to succeed, and it doesn't particularly matter which ones do. The high-variance approach ("throw it up and see what happens") may be optimal for the investor without being optimal for the founder.