I think if your goal is launching you've got to make sure the work you're doing is enabling that to happen sooner. Once you start prioritizing as little as possible into pre- and as much as possible into post- launch bad ideas start filtering themselves out during the wait.

I think one of the big traps is convincing yourself part of 'make x to do y' always includes 'learn a new stack' that will be 1% technically-better but 800% slower than what you used last time. Pretty crap optimization if you goal is to launch anything.

I think for developers on HN it's also very easy to build something with the expectation we'll want to discuss it 2 - 3 times a week on the front page for them to get users, neatly solving the business side in advance!