learning from success is better than failure

The set of good approaches is often hidden in an exponentially large space. Learning that an approach is not good doesn't help narrow that space significantly.

Suppose I'm trying to learn French, but I don't know anything about French. Instead I just make some sounds, and you tell me, "that's not proper French". Well duh, almost all combinations of sounds are not proper French. Trying and failing alone is not very helpful. It would be much better to hear examples of sentences that are proper French (most learning is by demonstration), and even more helpful to hear a structural explanation.

This is why it's incredibly valuable to work with good people, on projects that succeed. When a project doesn't succeed, it might be for any of a million reasons. You probably don't even know what you don't know---maybe the strategy that would have led to success isn't even part of your current conceptual framework. But when a project succeeds, most things had to have gone right. This gives you an exponential amount of information, and therefore confidence.