It doesn't follow that the theory is wrong, but there is a large gap between results attributable to it and the grandiose claims being made. Wouldn't the true scientific attitude be to hold off claiming to have a science of startups until significant results have first been produced, then reproduced?

A lot of people badly want this theory to be true. It's easy to see why; it says it is a method to success — not only that, but a quasi-scientific method that has tremendous appeal to the technically minded. But the fact that the desire to believe the theory is so strong is not evidence in its favor; it's reason for doubt.

I'm skeptical of the theory for two reasons. The first is that I keep feeling like I've seen it before, in another risk-ridden field where people badly want a reliable way to get results: the field of corporate software projects. Failure and unpredictability are so common there that people are eager to buy the idea of a repeatable process. Lean Startup Theory seems very much like software process transplanted to startups – so much so, in fact, that a more accurate name for it might be Startup Process. I could be wrong about this, but the sense of deja vu is striking.

If that's a correct analysis, it's possible to describe what it would look like. One would expect to see social clubs of believers who share techniques and inspiring stories about how their process is working. One would expect not to see any decisive outcomes, though people will expend great energy interpreting all outcomes as evidence for their process. One would expect to see gurus and consultants selling the answers that people are so eager for, whether they work or not. (In some cases it's better when they don't work, since one can then sell them repeatedly.) One would expect the gurus and consultants not to have any significant track record of success in the areas that they're teaching, since that's not actually what they're good at. And one would expect many people's desire to believe the theory to be strong enough for them to overlook that anomaly.

The other reason I'm skeptical is that, if the theory is true, it would amount to a science of history. There's nothing intrinsically different between startups and historical undertakings; that's all the word "enterprise" means, after all. Most don't lead anywhere, while a few lead somewhere important. An empirical technique to create the important ones would amount to an empirical technique to create history. That seems unlikely.