There’s a high failure rate in finance, too—it’s just hidden in the “up or out” culture. It’s a very winner-takes-all kind of place, from what I’ve heard.
If you want to be a portfolio manager who makes, say, macro bets, yes, it’s very much up or out. But if you want to be a quant polishing fixed income risk management models in some bank, it’s a pretty standard corporate job.
Startups are shockingly diverse too. And despite the super-high failure rates I hear about, the group of friends I’ve been tracking the past 5 years or so seem to be doing pretty darn well, despite some of them having failures indeed.
I strongly suspect the degree of failure in startups correlates inversely with rationality skills (as it should) so rationalists should not be placing themselves on the same reference category as everyone else. Execution skills matter a lot too, but doing a startup has worked miracles for my motivation too.
This isn’t “I’m smart and rules don’t apply”. Smartness alone doesn’t help.
But, to put it this way, if rationality training doesn’t help improve your startup’s odds of success, then there’s something wrong with the rationality training.
To be more precise, in my experience, a lot of startup failure is due to downright stupidity, or just ignoring the obvious.
Also, anecdotally, running a startup has been the absolute best on-the-job rationality training I’ve ever had.
Shockingly, successful entrepreneurs I’ve worked with score high on my rationality test, which consists of how often they say things that are uncontested red flags, and how well-reasoned their suggested courses of action are. In particular, one of our investors is the closest approximation to a bayesian superintelligence I’ve ever met. I can feed him data & news from the past week, and almost hear the weighting of various outcomes shift in his predictions and recommendations.
In short,
Rationalists are more likely to think better, avoid obvious errors.
Thinking better improves chances of startup success
Rationalists have better chances of startup success.
I do understand this sounds self-serving, but I also try to avoid the sin of underconfidence. In my experience, quality of thinking between rationalists and the average person tends to be similar to quality of conversation here versus on YouTube. The problem is when rationalists bite off more than they can chew in terms of goals, but that’s a separate problem.
What you say sounds intuitive to me at first, but as of now I would say that rationality training may boost start up success rates up just a little.
Here is some reasons why rationality might not matter that much:
People tend to be a bit more rational when it counts, like making money. So having correct beliefs about many things doesn’t really give you an edge because the other guy is also pretty rational for business stuff.
Well, at this point we’re weighing anecdotes, but..
Yes! They do tend to push their rationality to the limit. Hypothesis: knowing more about rationality can help push up the limit of how rational one can be.
Yes! It’s not about rationality alone. Persistent determination is quite possibly more important than rationality and intelligence put together. But I posit that rationality is a multiplier, and also tends to filter out the most destructive outcomes.
In general, I’d love to see some data on this, but I’m not holding my breath.
Agreed. Interestingly, the latest post in main points to evidence supporting rationality having a significant relation to success in the work place – not the same as entrepreneurship, nonetheless I update slightly more in favor of your position.
I agree that a more rational person has a greater chance, ceteris paribus. Question is, how much greater.
A part of the outcome is luck; I don’t know how big part. Also, the rationality training may improve your skills, but just to some degree.
(Data point: myself. I believe I am acting more rationally after CFAR minicamp than before, and it seems to be reflected by better outcomes in life, but there is still a lot of stupid things I do. So maybe my probability of running a successful startup has increased from 1% to 3%.)
I question the stats that says 1% success rate for startups. I will need to see the reference, but one I had access to basically said “1% matches or exceeds projections shown to investors” or some such. Funnily enough, by that metric Facebook is a failure (they missed the goal they set in the convertible note signed with Peter Thiel). If run decently, I would expect double digit success rates, for a more reasonable measure of success. If a driven, creative rationalist is running a company, I would expect a very high degree of success.
Another thing much more common in rationalists than the common population is the ability to actively solicit feedback, reflect and self-modify. This is surprisingly rare. And incredibly vital in a startup.
Success at startups is not about not doing stupid things. I’ve made many MANY mistakes. It’s about not doing things stupid enough to kill your company. Surprisingly, the business world has a lot of tolerance for error, as long as you avoid the truly bad ones.
It is hard to survey startups. What is usually done is to measure success rates of companies that raised a Series A round of funding. Many companies fail before achieving that, though they necessarily fail faster, producing less opportunity cost.
Here is a chart of returns to a VC, taken from this paper by a different author. 60% of dollars invested are in companies that lost the VCs money (lost them 85%). This is a top VC that managed to triple its money, so this is an overestimate of success of a regular VC-backed company. This is a common bias in these surveys.
Based on the fictitious figure 2, 63% of dollars is actually 69% of companies, because successful companies get more funding. So 31% of companies with a Series A round at a top firm succeed by the metric of a positive return to the VCs. Double digit success would require that at least 1⁄3 of startups get a Series A funding and that companies funded by typical VCs are as successful as companies funded by a top VC.
The appropriate definition of success is comparing to opportunity cost. In particular, the above analysis fails to take into account duration. Here is a paper that makes a reasonable comparison and concludes that running a company with a Series A round was a good decision for people with $700k in assets. Again, skipping to the Series A round is not a real action, thus overestimating the value of the real action of a startup. There is an additional difficulty that startups may have non-monetary costs and benefits, such as stress and learning. Edit: found the paper. According to Figure 2, that 75% of VC-backed firms exit at 0, not much worse than at the top VC considered above.
Well Paul Graham has built quite a successful incubator apparently largely based on his ability to predict success of start-ups based on a half-hour interview.
I’m not sure how much the interviews add compared to the Y Combinator model of investing in a lot of startups very early on at unusually favorable terms, integrating with Hacker News, and building a YC community with alumni & new angels. (As far as the latter goes, you can ask AngryParsley why he went into YC for Floobits: it wasn’t because he needed their cash.)
But if you want to be a quant polishing fixed income risk management models in some bank,
What kind of social skills does that require? My impression is that this is the modern equivalent of court astrologer and requires some similar skills, i.e., cold reading.
Not much—the usual ones for holding a corporate job (wear business casual, look neat, don’t smell, don’t be a weirdo). Quants are expected to be nerdy/geeky.
My impression is that this is the modern equivalent of court astrologer
Not at all. Finance has the advantage of providing rapid and unambiguous feedback for your actions.
Finance has the advantage of providing rapid and unambiguous feedback for your actions.
If you’re trading yes, although the feedback is extremely noisy. If you’re designing models not so much. Incidentally a lot of the quants I know are also good at doing Tarot readings, whether they believe the cards have power or not.
That very much depends on what kind of strategy you’re trading. For example, HFT doesn’t have problems with noise.
If you’re designing models not so much.
Yes, so much. Your model has to work well on historical data and if it makes it to production, it will have performance metrics that it will have to meet.
The other thing to keep in mind about failure rates is where you end up if you fail—what other careers you can go into with the same education. (In the case of startups, you can keep trying more startups, and you’re more likely to succeed on the second or third than you were on the first. I don’t know how it is in finance.)
Is finance higher E(money) than, say, a startup?
I would guess yes given the high startup failure rate.
There’s a high failure rate in finance, too—it’s just hidden in the “up or out” culture. It’s a very winner-takes-all kind of place, from what I’ve heard.
Finance is diverse.
If you want to be a portfolio manager who makes, say, macro bets, yes, it’s very much up or out. But if you want to be a quant polishing fixed income risk management models in some bank, it’s a pretty standard corporate job.
Startups are shockingly diverse too. And despite the super-high failure rates I hear about, the group of friends I’ve been tracking the past 5 years or so seem to be doing pretty darn well, despite some of them having failures indeed.
I strongly suspect the degree of failure in startups correlates inversely with rationality skills (as it should) so rationalists should not be placing themselves on the same reference category as everyone else. Execution skills matter a lot too, but doing a startup has worked miracles for my motivation too.
Not from the expected-income point of view (we’re not considering car dealerships and franchise eateries startups, right?).
Oh, dear. “I’m so smart that normal rules don’t apply to me”. What could possibly go wrong..?
This isn’t “I’m smart and rules don’t apply”. Smartness alone doesn’t help.
But, to put it this way, if rationality training doesn’t help improve your startup’s odds of success, then there’s something wrong with the rationality training.
To be more precise, in my experience, a lot of startup failure is due to downright stupidity, or just ignoring the obvious.
Also, anecdotally, running a startup has been the absolute best on-the-job rationality training I’ve ever had.
Shockingly, successful entrepreneurs I’ve worked with score high on my rationality test, which consists of how often they say things that are uncontested red flags, and how well-reasoned their suggested courses of action are. In particular, one of our investors is the closest approximation to a bayesian superintelligence I’ve ever met. I can feed him data & news from the past week, and almost hear the weighting of various outcomes shift in his predictions and recommendations.
In short,
Rationalists are more likely to think better, avoid obvious errors.
Thinking better improves chances of startup success
Rationalists have better chances of startup success.
I do understand this sounds self-serving, but I also try to avoid the sin of underconfidence. In my experience, quality of thinking between rationalists and the average person tends to be similar to quality of conversation here versus on YouTube. The problem is when rationalists bite off more than they can chew in terms of goals, but that’s a separate problem.
What you say sounds intuitive to me at first, but as of now I would say that rationality training may boost start up success rates up just a little.
Here is some reasons why rationality might not matter that much:
People tend to be a bit more rational when it counts, like making money. So having correct beliefs about many things doesn’t really give you an edge because the other guy is also pretty rational for business stuff.
self-delusion, psychopathy, irrationality, corruption, arrogance, and raw driven determination, have good if not better anecdotal evidence of boosting success than rationality training I think.
Well, at this point we’re weighing anecdotes, but..
Yes! They do tend to push their rationality to the limit. Hypothesis: knowing more about rationality can help push up the limit of how rational one can be.
Yes! It’s not about rationality alone. Persistent determination is quite possibly more important than rationality and intelligence put together. But I posit that rationality is a multiplier, and also tends to filter out the most destructive outcomes.
In general, I’d love to see some data on this, but I’m not holding my breath.
Agreed. Interestingly, the latest post in main points to evidence supporting rationality having a significant relation to success in the work place – not the same as entrepreneurship, nonetheless I update slightly more in favor of your position.
I agree that a more rational person has a greater chance, ceteris paribus. Question is, how much greater.
A part of the outcome is luck; I don’t know how big part. Also, the rationality training may improve your skills, but just to some degree.
(Data point: myself. I believe I am acting more rationally after CFAR minicamp than before, and it seems to be reflected by better outcomes in life, but there is still a lot of stupid things I do. So maybe my probability of running a successful startup has increased from 1% to 3%.)
I question the stats that says 1% success rate for startups. I will need to see the reference, but one I had access to basically said “1% matches or exceeds projections shown to investors” or some such. Funnily enough, by that metric Facebook is a failure (they missed the goal they set in the convertible note signed with Peter Thiel). If run decently, I would expect double digit success rates, for a more reasonable measure of success. If a driven, creative rationalist is running a company, I would expect a very high degree of success.
Another thing much more common in rationalists than the common population is the ability to actively solicit feedback, reflect and self-modify. This is surprisingly rare. And incredibly vital in a startup.
Success at startups is not about not doing stupid things. I’ve made many MANY mistakes. It’s about not doing things stupid enough to kill your company. Surprisingly, the business world has a lot of tolerance for error, as long as you avoid the truly bad ones.
It is hard to survey startups. What is usually done is to measure success rates of companies that raised a Series A round of funding. Many companies fail before achieving that, though they necessarily fail faster, producing less opportunity cost.
Here is a chart of returns to a VC, taken from this paper by a different author. 60% of dollars invested are in companies that lost the VCs money (lost them 85%). This is a top VC that managed to triple its money, so this is an overestimate of success of a regular VC-backed company. This is a common bias in these surveys.
Based on the fictitious figure 2, 63% of dollars is actually 69% of companies, because successful companies get more funding. So 31% of companies with a Series A round at a top firm succeed by the metric of a positive return to the VCs. Double digit success would require that at least 1⁄3 of startups get a Series A funding and that companies funded by typical VCs are as successful as companies funded by a top VC.
The appropriate definition of success is comparing to opportunity cost. In particular, the above analysis fails to take into account duration. Here is a paper that makes a reasonable comparison and concludes that running a company with a Series A round was a good decision for people with $700k in assets. Again, skipping to the Series A round is not a real action, thus overestimating the value of the real action of a startup. There is an additional difficulty that startups may have non-monetary costs and benefits, such as stress and learning. Edit: found the paper. According to Figure 2, that 75% of VC-backed firms exit at 0, not much worse than at the top VC considered above.
Well Paul Graham has built quite a successful incubator apparently largely based on his ability to predict success of start-ups based on a half-hour interview.
Besides what gwern said, Paul Graham is a successful VC. The expected income of VCs is very different from the expected income of startup founders.
My point is that this is evidence that start-up success depends on ability more than luck.
I think both ability and luck are necessary but not sufficient (well, reasonable amounts of luck :-D).
I’m not sure how much the interviews add compared to the Y Combinator model of investing in a lot of startups very early on at unusually favorable terms, integrating with Hacker News, and building a YC community with alumni & new angels. (As far as the latter goes, you can ask AngryParsley why he went into YC for Floobits: it wasn’t because he needed their cash.)
What kind of social skills does that require? My impression is that this is the modern equivalent of court astrologer and requires some similar skills, i.e., cold reading.
Not much—the usual ones for holding a corporate job (wear business casual, look neat, don’t smell, don’t be a weirdo). Quants are expected to be nerdy/geeky.
Not at all. Finance has the advantage of providing rapid and unambiguous feedback for your actions.
If you’re trading yes, although the feedback is extremely noisy. If you’re designing models not so much. Incidentally a lot of the quants I know are also good at doing Tarot readings, whether they believe the cards have power or not.
That very much depends on what kind of strategy you’re trading. For example, HFT doesn’t have problems with noise.
Yes, so much. Your model has to work well on historical data and if it makes it to production, it will have performance metrics that it will have to meet.
The other thing to keep in mind about failure rates is where you end up if you fail—what other careers you can go into with the same education. (In the case of startups, you can keep trying more startups, and you’re more likely to succeed on the second or third than you were on the first. I don’t know how it is in finance.)
I think a higher startup failure rate implies E(startup) > E(finance) since most people want risk-adjusted return
Not necessarily because of different barriers to entry.