It's a fallacy (though a very commonly held one) that if x% of people succeed at some task, the chances you will succeed are x%. That's only true for tasks where aptitude is not a factor.
While true, so, so much of your life is luck (or more accurately, good fortune through chance, only affected by positioning). While your chances may not exactly be 0.00006% that you'll succeed, I doubt that you can improve them to more than about 0.0001%.
Truly, we don't get where we get because of our capacity or our ability (in most cases). We get lucky, a lot, and we can't really change that.
For many people, that luck already happened based on who their parents are and what sort of upbringing they had. Someone born to middle-class American parents has a whole lot better shot at being a billionaire than someone born in the inner-city ghetto, or, for that matter, in rural China.
Much of the remaining "luck" comes from our choices. Think of it this way: how many people choose to work a steady job for all of their life. They will never be a billionaire, regardless of how lucky they get, and so their contribution to the average is zero. Actually, I suspect this invalidates your math: more than 1 in 2 people have probability zero of being a billionaire simply by virtue of not taking the chance of being one, and therefore of the people who take the chance, the odds are significantly better than 0.0001%.
Though it's still a bit early to say, it looks like the probability of a YC startup ending up worth a billion dollars is at least .5%. So the truth is somewhere on the continuum between (a) that you're off by a factor of 5000, and (b) that we improve people's chances by 5000x. And while it would be a great compliment if people chose extremum (b), I think it would be a stretch to claim we can improve the probability by even 10x.
You've just added information that will alter the probability. The prior can still be 0.00006% and the addition of new evidence (in the bayesian sense) of being admitted into YC could result in a posterior probability of 0.5%. (Sorta like using Series A as a filter.)
Yes, this is extremely simple, and I'm surprised to see intelligent people apparently confused over it. The percentage in the article is obviously referring to all companies, and more information will obviously alter the percentage. Are people genuinely confused by this, or are they just trying to discredit the article by feigning confusion?
This only adds (Bayesian) evidence to an abstract frame of knowledge that doesn't matter to anyone in real decision making processes: a frame in which the observer uniformly samples all startups.
People who actually care about these odds (in the sense of betting on them) are founders and investors. Neither, in their decision making process, gets to (or wants to!) uniformly sample all companies.
PG's numbers are thus much more useful to anyone actually trying to make a decision about a pool of investments: assuming the distribution of YC startups is fixed over time, and you are someone like Start Fund who will bet on the pool (i.e. equivalent to a repeated uniform sampling in expected value), the 0.5% is actionable information and the 0.00006% is not.
I don't disagree with you re:investors (or pg's numbers) who seek to obtain as much evidence as possible so as to maximize their ROI. But priors do matter.
For example, if the prior on "making a successful company" (defined however you want) were a vastly higher 40%, then I'd imagine a lot more laypeople would take the plunge. Reading sites like TechCrunch makes it seem to the layperson that building a successful company is much easier than it really is. So yes, knowing that "mega success" is a massive outlier (to the tune of 1:1,000,000) is indeed actionable information to a layperson thinking about starting a company without any additional evidence.
As the source article notes: The goal of the entrepreneur is to learn as much as they can, thereby increasing their own odds of success (or minimizing their odds of failure). Obviously, getting into YC massively improves your odds and would probably be a good decision! As a YC alum, my advise would jive with this observation. ;)
From another perspective, likelihoods like these depend on information. Every iota of market validation that is compatible with a $1B company improves the likelihood that you will create one.
If one factors in the 2-3% YC acceptance rate, and makes the assumption (that I admittedly haven't checked out) that no startups rejected by YC are now billion dollar companies, that closes the difference down to around 100x
If there are limited choices for resources and YC gets to choose the best options, then how much is actually attributed to being in YC vs. those startups having received mentorship? Out of the companies that will become or are billion dollar companies, how many already had their direction/focus and business model decided before joining YC? Correlation != causation, right?
We get lucky, a lot, and we can't really change that.
That depends on how, exactly, you define "luck". I posit that a lot of what people call "luck" can be manufactured, or at least cultivated through directed action.
Remember the article that showed up here a while back about "How to date a supermodel"? The premise was that if you want to date a supermodel, you have to move to a city where there are lots of supermodels, and hang out at the places where supermodels shop, work out, dine, etc., and you have to start conversations with supermodels, blah, blah..
So if one of you buddies shows up next year dating a supermodel, everybody is probably going to go "Dude, that's amazing, you are SO lucky!" And this will completely ignore the fact that he did a lot of things to create the opportunity.
It's SUCH a cliche, but I guess cliches exist for a reason, so I'll just come back to:
"The premise was that if you want to date a supermodel, you have to move to a city where there are lots of supermodels, and hang out at the places where supermodels shop, work out, dine, etc., and you have to start conversations with supermodels, blah, blah.."
Likewise, if you don't want to date anyone at all, you move to silicon valley and work on a startup!
You don't even need to move to Silicon Valley to do that! You can "date no one" in Raleigh/Durham, North Carolina, while working on a startup, quite effectively.
Supermodels are a special elite cohort of models. They are household names, people with instant recognition that can generate news articles simply by announcing that they are doing something involving your brand.
To date one you would need to be compatible with her lifestyle. That means highly successful or at least part of a similar industry like fashion or music. It means that you have to be somebody that can be announced in gossip columns as dating her. It has to be a good career move for her.
Naomi Cambel once walked past me while I was hanging out backstage at New York fashion week. Even if I had said hi there are still significant reasons why I am not now dating her.
And so it is with startups. Pedigree matters, the pedigree of your investors matters. Press matters and is heavily influenced by your position in the network. People in the game are deciding who the winners and losers are. Public perception is influenced by that.
Yes, this is like telling a newlywed that s/he has a 50% chance to eventually get divorced. True for the population, irrelevant for the individual. When one has a more direct knowledge of the relevant factors, it's crazy to make a decision based solely on into what demographics one falls.
> Yes, this is like telling a newlywed that s/he has a 50% chance to eventually get divorced.
FWIW, I really dislike it when people perpetuate the myth that 50% of marriages end in divorce. We actually have no idea what the true percentage is, and I think they myth got started because people compare the annual marriage rate with the annual divorce rage.
Also true! There's a small literature in philosophy that worries about paradoxes in this area. But regardless of how that turns out, it seems safe to say that you shouldn't avoid getting married because of divorce demography, and you further shouldn't think of your marriage's chance of success in purely demographic terms. You instead need to pay attention to features on the ground like how you handle disagreements, your sex life, etc.
Yes and no. The way the rate is calculated is relative, so the people getting married in a given year and divorced in a given year aren't necessarily the same people.
> True for the population, irrelevant for the individual.
Not necessarily, because people are notoriously bad at evaluating themselves, and often tend to assume that statistics (especially troubling ones) don't apply to them for whatever reason. For the divorce statistic (which I understand is not necessarily accurate, but let's pretend it really is 50%), how many of the divorced couples previously thought the statistic was irrelevant for them, because they're "truly in love" or some other reason?
You're right but you're speaking to the title of the article only which is unfortunate because the article is really good but the title itself isn't representative of it. Normally this wouldn't be an issue but with you being you it's the number one comment and now the entire conversation on hacker news is revolving around the title.
The article makes some really great points about solving meaningful problems vs starting a startup for it's own sake. It further makes great points about eliminating luck in the process and offers actionable advice on how to do that. I think that would be the subject of a much better conversation.
Agree and same as a statistic of chance of "getting hit by lighting" or any activities that doesn't take into account where someone is located or even whether an indoor or outdoor location.
For example are there any statistics on the chance of getting hit by lighting while playing golf outdoors with a storm within 50 miles? Vs. in general?
And what if you calculated the chance of starting a billion dollar company in Bentonville Ark (Walmart) which has a population of 38k (which no doubt was way less when Sam Walton decided to locate there). Not bad "odds".
No, it's true if you don't have more specific data regarding other factors like aptitude. It's no different than saying "if you live in the United States you have a life expectancy of 79 years." Obviously, with more information (are you obese? are you wealthy? do you have certain genetic predispositions to disease? etc.) each person can get a more specific life expectancy, but that doesn't change the life expectancy of the country.
Does 0.00006% factor in how many companies are started with goals that have no billion dollar potential? While any company may have an outlying chance of accidentally coming up with a billion dollar idea, most have business plans that would never scale to a billion dollar valuation no matter how well executed or lucky they are. I'm curious whether this statistic is only based on the success of those companies that are _trying_ to get there.
if i have 10 coins, 9 of which have 44.4% chance of heads and 1 has 100% chance, choosing a coin at random and flipping it has 50% chance of heads overall. sure, there is one outstanding coin.
the "fallacy" is true when you look at the whole rather than the individual factors, which include aptitude in some cases so i guess you're correct on the last point.
Color me dense, but I don't understand this point. The author clearly had aptitude to be successful, but if we were to randomly put a percentage on his aptitude that weighted him above the average, it would still probably be less than 0.00%
It means that statistics are useful until you try to apply them to an individual. While a statistic may hold true for an entire population, the converse does not logically follow for an individual. I think the fallacy may be "over generalization".
Correct. The value of statistics like these come in making generalizations, which require massive data sets to be applicable (hence generalization rather than specialization). It's akin to in baseball, an average of x% of pitches are hit for home runs. This will likely be fairly accurate over the course of the season across all players, but would not be a useful stat to look at in predicting what a pitcher would do at the plate (or an MVP level hitter for that matter).
Thank you very much for the breakdown! The connection to the article would be that we shouldn't overgeneralize a percentage of success for an individual?
One easy counter-example is "50% [well, technically a bit less, but ignore that for now] of humans are capable of getting pregnant. You are human. Therefore, there is a 50% chance that you are capable of getting pregnant." This ignores the fact that there is another very obvious discriminant - namely, sex - between people who are capable of getting pregnant and people who are not. If you are male, your chance is 0%, if you are female your chance is [close to] 100%, and it is fairly easy to tell without actually trying to get pregnant which one you are.
This also comes up all the time in data-science and A/B testing of products. Frequently you'll end up with results like "This change decreased revenue by 3%." Sounds bad, right? Except then you drill into your results and find out that there was a bug in the implementation of the change in IE6, which decreased revenue by 100% on that one browser which happens to be 5% of traffic, and find that your change actually increased revenue by a percent or two, but that one slice cost you all the benefit and more. You fix the bug and happily start making more money.
Always slice your population. Many times blanket probabilities tell you nothing unless you know what's causing them.
Right. An analogy would be saying "people who eat over 6000 calories a day are, as a rule, overweight". It's a logical fallacy to turn around and say "Michael eats over 6000 calories a day, therefore he is overweight". It's statistically very likely, but in this case Michael is a olympic swimmer with 10 gold medals. It's been a long time since my logic class in college, but the rule of 'statistics are great until you try to apply them to an individual' has always stuck in my mind. Probably because statistics are often one step removed from 'stereotypes'.
Your observation is only true if aptitude is an independently measurable quality. If the only way to measure aptitude is to say that those same x% of people must have it because they succeeded, then no, it is not a fallacy.
As someone in the business of measuring aptitude, I can tell you with some confidence that it is. (If it weren't, we could save ourselves weeks of effort each batch reading applications and doing interviews, and instead just choose randomly among the applicants.)
I certainly understand and respect your belief. But as far as I know, there is no objective proof that your filtering process is actually any better than choosing randomly. Consider a gardener who believes they can predict which seeds will bloom best. They select a few seed from a package, water and nurture them for weeks, then point triumphantly to the blooms and say, "See! They've done so much better than the ones I left in the package."
The competition factor is also very important: if the world population (and entrepreneurs) growth by 400% it is possible that the "same" few companies will reach the billion (adjusted by inflation)
