Big changes in approach are inevitable in early-stage companies: things become apparent after 1-2 years that weren't initially. Early stage companies involve so much uncertainty that trying to stick to a plan for the sake of sticking to a plan is folly.
His numbers are in a different context; but I get his point. Venture investing is a portfolio business; you fully expect a high failure rate.
He makes the interesting point that 2/3rds of his successful investments made major changes along the way.