Startups have a really long lifecycle associated with them. I have answers to the question with respect to startups I was not professionally engaged with. I don't have many data points regarding ones which I was, because the jury won't be in for a few years on the numeric supermajority of them. My set of data points where I have reasonable confidence in terminal outcome is very small relative to e.g. a professional investor who was active over the same length of time I've been on HN.

This is not a new insight, but the thing pg says about successful startups often looking like a toy appears to me to be true. One benefit of having a searchable archive of my written thoughts going back 10+ years is that I can see what I thought of e.g. Facebook or Airbnb back in the day. In both cases, I thought they were unlikely to find a business model. I was very clearly wrong about that. I am, accordingly, very careful around the fact pattern "it's clearly something people love but is there really a business there?"

Within my specific interest in B2B SaaS companies, a thing I have historically overindexed on is amazing product teams without demonstrated success in sales in the appropriate model (low-touch, high-touch, etc). I have sometimes been wrong about "Ahh, clearly they will figure out enterprise sales; how hard can it be if they can clearly ship a good product?" It is, in fact, possible to be a great product owner and not successfully build a sales organization strong enough to take a company to the sort of trajectory successful funded startups want to walk down.