> but we shouldn't forget survivor bias here
Hi HN, entrepreneurial scientist here. You have just given me an idea. Maybe I shouldn't be posting this buried in the comments but can always be reposted elsewhere.
"survivor bias"
This is a constant talking point in HN. It allows anyone to plausibly dismiss absolutely any advice from a successful entrepreneur (no offence to the poster here intended, they may be right - it's just that it can be used anywhere).
It's so powerful, in fact, that the only plausible retort is to look to serial successful entrepreneurs - Jobs, Musk etc. - as evidence that it's at least not entirely luck that leads to success. And even that counterargument is on shaky ground.
So what we need to do is to get more scientific. We need to know how many other startups were formed in that area, or with that idea, or just in general. I think this is totally doable. What we need is a bad ass business scientist in some university to take on this challenge. Actually, hell - this is a startup idea in itself. So it probably exists somewhere.
I would begin by monitoring all new company registrations in my home country (UK) and keeping a look out for tech startups. You can check their websites at regular intervals, look at Companies House metadata, and scrape news to classify them correctly. Monitor accounts returns to see when/if they trade, and fold. Although it'd take time to build the database, after a year I think you'd already have interesting observations.
You could also do phone interviews with (a sample of?) founders. The goal would be to get them to almost "register" themselves in the same way as a medical trial might be registered. You'll definitely have a problem with confidentiality, so you may need to delay the release of data by 10 years, or to release only aggregate data, or else to rely only on aggregating public domain information to infer activity and talk in general terms to founders.
Other sources of data: incubator batches, demo days, product hunt, advertisements placed by the startups, job adverts. I'd also do "exit interviews" with founders that are no longer registered as directors at Companies House, or when their companies cease trading.
If we can do this, we can start to get absolutely solid numbers on whether survivor bias is really a thing. Like, are their explanatory factors as to why Company A did well and not B, or was it just A was lucky? Because if B won because they took VC or because they launched earlier or whatever it is, then we can start to say - survivor bias isn't really a thing, because the best funded or quickest company wins. And here's the best strategy.
This data would seem so valuable to VCs and founders that you'd think you could make money from it.
And then we could put an end to "survivor bias" as an argument - which presumably would be the outcome (ah confirmation bias! i can't be that scientist) because otherwise it's disproportionately luck and we can all go home now.