There is an interesting Schumpeter column in The Economist

Source: "Fail often, fail well: Companies have a great deal to learn from failure—provided they manage it successfully"

I quote:

(S)imply “embracing” failure would be as silly as ignoring it. Companies need to learn how to manage it. Amy Edmondson of Harvard Business School argues that the first thing they must do is distinguish between productive and unproductive failures. There is nothing to be gained from tolerating defects on the production line or mistakes in the operating theatre......

Companies must also recognise the virtues of failing small and failing fast. Peter Sims likens this to placing “Little Bets”, in a new book of that title......

Placing small bets is one of several ways that companies can limit the downside of failure. Mr Sims emphasises the importance of testing ideas on consumers using rough-and-ready prototypes: they will be more willing to give honest opinions on something that is clearly an early-stage mock-up than on something that looks like the finished product. Chris Zook, of Bain & Company, a consultancy, urges companies to keep potential failures close to their core business—perhaps by introducing existing products into new markets or new products into familiar markets. Rita Gunther McGrath of Columbia Business School suggests that companies should guard against “confirmation bias” by giving one team member the job of looking for flaws.

But there is no point in failing fast if you fail to learn from your mistakes.

You have to kiss a lot of frogs to find a princess!

Essentially, I think the argument is that companies (and others) should take risks that they can afford if and only if the payoff if the bet turns out to win will be large enough to justify the risk. Part of the payoff may be in knowledge gained from the experience. I think I might be more likely to put money in a start up by people who have failed once or twice in previous startups and learned from the experience.

This is also part of the scientific method. Create a hypothesis that you can afford to test. Test it, and if it fails, do it again.