I love the comparison to writing an essay here. There’s something to be said for putting pen to paper when you’re struggling to form a strong hypothesis, but even then one is starting with a weak hypothesis. I often advise interview candidates to try forming a strong approach first, but if they fail to find one, fall back to a weaker approach that they’ll at least be able to implement code for. Then try to come back for a stronger approach. This way, you don’t leave empty-handed.