Is startup validation bullshit?

Is validation bullshit? No. Can you "validate" the wrong things (measure the wrong metrics) and make yourself believe that you have real validated learning when you don't? Absolutely.

If you read The Four Steps To The Epiphany you'll note that @sgblank really emphasizes the importance of getting someone to actually give you money. For many (most?) classes of products, you don't have any validated learning whatsoever unless someone has given you money for the product/service you're building (if you're going for an ad supported model, then all bets are off. That space doesn't interest me, and I don't think about it).

Beta signups? If they're free (and you intend to charge for the "real" version), they don't really count as validated learning. They're suggestive maybe, but they don't validate "somebody will pay money for this X". Email list signups? Useful, suggestive, but not validated learning. Twitter followers, G+ followers, Facebook likes, and six tons of email from people who aren't in your target market, or who haven't paid you any money yet? Might be useful, but not validated learning from where I'm sitting.

The problem I always have with the validation stuff is that it often leads me down a road where I'm doing something I don't want to do or unimaginitve. In other words, it often leads to the apocryphal quote of Ford "If I had asked people what they wanted, they would have said faster horses."

As I continue working with validation and such, I'm beginning to feel like the most successful are simply those who know when to slow down and validate and when to hit the gas and follow their vision. Too much of either leads to a crash.