把失败变便宜，把学习变快

Paul Graham has a point about venture capital that I find interesting. In normal business we assume a higher success rate is always better. In VC it isn't that simple: the big money usually comes from one or two companies out of ten or twenty returning tens or hundreds of times, while most fail. A VC who works hard to lower his failure rate becomes more and more conservative, choosing safe, easy-to-understand deals, and those rarely produce huge returns because everyone can see them and competition arrives. The opportunities that produce outsized returns often look strange, uncertain, even like bad ideas at the start. Founders aren't VCs: we can't fail our own company fifteen times. So the goal isn't to chase failure, but to give ourselves more failures we can afford. If testing an idea takes twenty people, RM2 million and eighteen months, you get one or two tries. If two people can test the same hypothesis in two weeks with little money, you can try ten or twenty times a year, and one may produce a non-linear result. Not random tries: each needs a hypothesis, a cost ceiling and a decision rule. Stop when wrong, continue on a weak signal, concentrate resources on a strong one. The real danger isn't failure; it's only daring to do things that can't go wrong. Make failure cheap, learn fast, stay at the table, and keep playing until you find an opportunity worth raising your stake on.