"<i>at least, not right away.</i>"<p>I.e., as detailed, the "common advice" that they "<i>need to cycle through different ideas with as little time and money as possible to discover which of their assumptions are wrong.</i>"<p>Which at this point in the game can be done rather cheaply. As pointed out in a recent discussion, "lean" meant something different in Web 1.0, where you had to spend a lot of money just to get basic infrastructure of all types.<p>Today we are <i>incredibly</i> leveraged, not just in obvious thinks like renting cloud time instead of buying servers et. al., using high (enough) quality FOSS software, but also in just finding out stuff with Google, Wikipedia, etc. etc. etc. What works? What is this bug? What did someone do that worked when they faced this situation?<p>In the '70s Jerry Pournelle predicted we would fairly soon achieve a state where the answer to any question could be rapidly found. He packaged this neatly (e.g. ask your implant which talks to an AI...), but I think we've already achieved a crude and significant level of this.