<i>What Paul ignores is that there is a non-linear reduction in risk that takes place with adoption of a service.</i><p>This was my exact response to pg's article. Or at least that one sentence that seemed to stick out.<p><i>If you're investing at a tenth the valuation, you only have to be a tenth as sure.</i><p>But of course pg cleared that up. The key is not that you can be 1/10th as sure that said investment will be successful, but that you can be 1/10th as sure that said investment will be worth $1B. Fixing final valuation linearizes things.<p><i>Suppose your threshold for investing in a startup is an n% confidence that they'll one day have a market cap of a billion dollars.<p>Suppose instead you split that investment between 10 companies at a tenth the valuation. How confident do you have to be that any given one will become a billion dollar company?</i>