I think that analysis is too one-dimensional - it's not meaningful enough for me to simply say "hardware startups are probably bad and software startups are pro

I think that analysis is too one-dimensional - it's not meaningful enough for me to simply say "hardware startups are probably bad and software startups are probably good." That's not a complete picture, even if it's true that successful startups are not evenly distributed between hardware and software, because all the other factors of startup success are not evenly co-distributed between hardware and software. It is merely one dimension of risk analysis. In the United States, ~32,000 people die annually in car accidents. In New York, ~1200 people die annually in car accidents. I live in New York. Which statistic do I pay attention to? Does it influence my analysis of total vehicular accidents that New York has a much higher population density than most of the country? Am I more likely to die in New York or somewhere else? What is the startup building? What is the existing market the product? What experience and domain knowledge do the founders have? How far along are they? Who are their actual and potential customers? That's the first five minutes of a due diligence conversation. If it were so easy to divide potential winners from virtually certain losers, people would hedge so well they would lose almost nothing and most angel investors would become rich. The reality is very different.