The kind of “decisiveness” Altman is talking about doesn’t involve making research or business decisions that matter on the scale of months or weeks, but responding to emails in a few minutes. In other words, the minds Altman is looking for are not just generically decisive, but quickly responsive—not spending long slow cycles doing a new thing and following their own interest, but anxiously attentive to new inputs, jumping through his hoops fast.
I think this proves too much. That is, it also seems like this kind of decisiveness would be helpful for a movie director, where there are thousands of details, and a core task of being a director is when an underling presents you with option A or option B, you pick the option that better suits the project’s gestalt, and then the next underling presents you with the next choice. There are also decisions that happen at the scale of months or weeks, and the director needs to be able to make those decisions too. But inability to make snap judgments on problems that only require snap judgments means they’ll spend way too much of their time on trivialities and their underlings won’t have a rapid feedback loop for iteration, which is poison for a movie and death for a startup.
Our hypothetical movie director doesn’t have to be motivated by responsiveness-to-authority; they don’t have to have a desire to pass tests. They quite possibly are following their own interest. They just have to make a lot of decisions rapidly in response to input.
Now, quite possibly the thing that’s going on is the ‘red-haired founder’ thing, where if you know future investors won’t invest in red-haired founders then you shouldn’t either, unless you don’t think the company needs any additional investment rounds to survive, and so he’s just checking whether or not the founders have the desire-to-pass-tests and skills such that future investors will be interested, and the basic story of the post goes through (where they advertise as being about production, but are also filtering heavily on approval extraction).
But the basic effect that rhetorical move had on me is make me suspect that you’re not considering the alternative hypotheticals that seem relevant to me, or think they aren’t worth including in the post. Like, in the hypothetical world where normal startups and VCs are 20% about production and 80% about whatever other fluff (both waste and predation), and Y Combinator is 40% about production and 60% about whatever other fluff, would you be writing about how YC still has 60% fluff, or about how there’s twice as much production? [Numbers, of course, chosen out of thin air to serve as an example; I’m not sure where the actual fraction is, not paying much attention to YC.]
Both options serve valid roles, but it seems important to advertise which sort of an analysis you’re doing, and I think this connects to complaints about word choice of ‘apex predator’ and ‘scam’ and so on; if ‘scam’ normally means “less focused on production than average” then using it to mean “some focus devoted to anything besides production” will cause predictable misunderstandings. [Noting that, in my hypothetical example YC is still more non-production than production, but the population standard isn’t ’50%‘.] But also there are cases where ‘what the normal population does’ really isn’t good enough, and you have to hit a much higher standard; but it’s not obvious to me yet that this is such a case.
I didn’t use the word scam in the post. Cousin_it’s comment was defending YC by saying, substantively, don’t worry, it’s just a pretend gatekeeper, so I tried to make that more explicit. How, precisely, does that not constitute being a scam?
Are we supposed to be at war with lions now? What’s wrong with apex predators?
Cousin_it’s comment was defending YC by saying, substantively, don’t worry, it’s just a pretend gatekeeper, so I tried to make that more explicit. How, precisely, does that not constitute being a scam?
I interpreted that exchange quite differently. I understood you as saying, “look, the people that want to pass tests and seek approval, those people are going to flock to YC, because it’s at the top of a mountain that people who want to pass tests will seek to climb. The result will be that anyone who fully buys their advertising will be disappointed upon arrival (or too unfamiliar with what actual production looks like to notice the difference).” It reminds me of a friend who went to Harvard, thinking they would meet original thinkers and deep scholars there, and being distressed by how much their fellow students were people who wanted to pass tests and end up in finance or management consulting.
Further, you point to the ways in which YC is explicitly a gatekeeper (for entry into the YC keiretsu / alumni community), which is relevant to the broader question of “is YC flooded by approval-seeking applicants, or is YC seeking out the true producers?”, and also relevant to the question of “are YC companies doing real production?”.
cousin_it replied on the question of “does a startup need to go to YC in order to succeed?”, to which the answer is “no,” because success still exists outside of the YC keiretsu. That is, if you want to practice medicine in the US there is a hard gatekeeper of whether or not you have a valid medical license, but if you want to start a company in the Bay there is no such hard gatekeeper. That is, YC’s gatekeeping is in no way ‘pretend’, you’re just disagreeing about what fence the gate is in.
Are we supposed to be at war with lions now? What’s wrong with apex predators?
I read some disapproval into your choice of ‘apex predator’ instead of ‘paragon’ or ‘zenith’ or so on; perhaps that isn’t what you intended to communicate. [Like, I’ve been using “fluff” to refer to “non-production”, but some central elements of “non-production” are predation and conflict and waste and so on, which “fluff” is an overly positive word for, such that I wasn’t willing to use it the first time without calling out those examples.]
---
To move towards what I suspect is the center of the disagreement, my current best guess of what’s going on is that YC is trying to do ‘production in 2019’, which is not pure production and nevertheless has some real production. It’s not robust to Goodharting on growth and the alumni network likely artificially props up early companies whose products get an early in from passing through an interview that’s only somewhat related to quality. It is sometimes accepting the sort of anxious test-passer that you’re talking about here, and Goodharting on growth might be a product of this (where the only way to get an anxious test-passer to think about making real products is to get them to focus on the growth metric).
But also they do have some of the ‘real production’ thing, and they’re not McKinsey and able to make use of any anxious test-passer; they have to point towards the ‘real production’ thing in order to get the more useful sort of founder, and if someone sufficiently manages to pass that test through anxious perfectionism, mission fucking accomplished.
[Interestingly, this take reminds me of the Confucian focus on ritual; “yeah, they’re not going to understand what it’s for, but it’ll help, and it’s by just doing the ritual that they eventually figure out what it’s for, with the side benefit of not causing problems in the meantime.”]
The “false advertising” charge, as I understand it, is that ‘production in 2019’ is not the right goal, and a different thing is, and this distinction is worth making because it’s correct. [I noticed that I really wanted to put an appeal to consequences after the because, as I think it makes the statement more forceful / relevant, but it wasn’t how my inner Benquo would put it; the point isn’t that people will be confused and that’s bad, the point is that it’s not correct.]
Altman’s responsiveness test might be more useful if it measured a founder’s reply time to an underling rather than to the guy holding the purse strings/ contact book.
I think this proves too much. That is, it also seems like this kind of decisiveness would be helpful for a movie director, where there are thousands of details, and a core task of being a director is when an underling presents you with option A or option B, you pick the option that better suits the project’s gestalt, and then the next underling presents you with the next choice. There are also decisions that happen at the scale of months or weeks, and the director needs to be able to make those decisions too. But inability to make snap judgments on problems that only require snap judgments means they’ll spend way too much of their time on trivialities and their underlings won’t have a rapid feedback loop for iteration, which is poison for a movie and death for a startup.
Our hypothetical movie director doesn’t have to be motivated by responsiveness-to-authority; they don’t have to have a desire to pass tests. They quite possibly are following their own interest. They just have to make a lot of decisions rapidly in response to input.
Now, quite possibly the thing that’s going on is the ‘red-haired founder’ thing, where if you know future investors won’t invest in red-haired founders then you shouldn’t either, unless you don’t think the company needs any additional investment rounds to survive, and so he’s just checking whether or not the founders have the desire-to-pass-tests and skills such that future investors will be interested, and the basic story of the post goes through (where they advertise as being about production, but are also filtering heavily on approval extraction).
But the basic effect that rhetorical move had on me is make me suspect that you’re not considering the alternative hypotheticals that seem relevant to me, or think they aren’t worth including in the post. Like, in the hypothetical world where normal startups and VCs are 20% about production and 80% about whatever other fluff (both waste and predation), and Y Combinator is 40% about production and 60% about whatever other fluff, would you be writing about how YC still has 60% fluff, or about how there’s twice as much production? [Numbers, of course, chosen out of thin air to serve as an example; I’m not sure where the actual fraction is, not paying much attention to YC.]
Both options serve valid roles, but it seems important to advertise which sort of an analysis you’re doing, and I think this connects to complaints about word choice of ‘apex predator’ and ‘scam’ and so on; if ‘scam’ normally means “less focused on production than average” then using it to mean “some focus devoted to anything besides production” will cause predictable misunderstandings. [Noting that, in my hypothetical example YC is still more non-production than production, but the population standard isn’t ’50%‘.] But also there are cases where ‘what the normal population does’ really isn’t good enough, and you have to hit a much higher standard; but it’s not obvious to me yet that this is such a case.
I didn’t use the word scam in the post. Cousin_it’s comment was defending YC by saying, substantively, don’t worry, it’s just a pretend gatekeeper, so I tried to make that more explicit. How, precisely, does that not constitute being a scam?
Are we supposed to be at war with lions now? What’s wrong with apex predators?
Yeah, I was referring to your reply to cousin_it.
I interpreted that exchange quite differently. I understood you as saying, “look, the people that want to pass tests and seek approval, those people are going to flock to YC, because it’s at the top of a mountain that people who want to pass tests will seek to climb. The result will be that anyone who fully buys their advertising will be disappointed upon arrival (or too unfamiliar with what actual production looks like to notice the difference).” It reminds me of a friend who went to Harvard, thinking they would meet original thinkers and deep scholars there, and being distressed by how much their fellow students were people who wanted to pass tests and end up in finance or management consulting.
Further, you point to the ways in which YC is explicitly a gatekeeper (for entry into the YC keiretsu / alumni community), which is relevant to the broader question of “is YC flooded by approval-seeking applicants, or is YC seeking out the true producers?”, and also relevant to the question of “are YC companies doing real production?”.
cousin_it replied on the question of “does a startup need to go to YC in order to succeed?”, to which the answer is “no,” because success still exists outside of the YC keiretsu. That is, if you want to practice medicine in the US there is a hard gatekeeper of whether or not you have a valid medical license, but if you want to start a company in the Bay there is no such hard gatekeeper. That is, YC’s gatekeeping is in no way ‘pretend’, you’re just disagreeing about what fence the gate is in.
I read some disapproval into your choice of ‘apex predator’ instead of ‘paragon’ or ‘zenith’ or so on; perhaps that isn’t what you intended to communicate. [Like, I’ve been using “fluff” to refer to “non-production”, but some central elements of “non-production” are predation and conflict and waste and so on, which “fluff” is an overly positive word for, such that I wasn’t willing to use it the first time without calling out those examples.]
---
To move towards what I suspect is the center of the disagreement, my current best guess of what’s going on is that YC is trying to do ‘production in 2019’, which is not pure production and nevertheless has some real production. It’s not robust to Goodharting on growth and the alumni network likely artificially props up early companies whose products get an early in from passing through an interview that’s only somewhat related to quality. It is sometimes accepting the sort of anxious test-passer that you’re talking about here, and Goodharting on growth might be a product of this (where the only way to get an anxious test-passer to think about making real products is to get them to focus on the growth metric).
But also they do have some of the ‘real production’ thing, and they’re not McKinsey and able to make use of any anxious test-passer; they have to point towards the ‘real production’ thing in order to get the more useful sort of founder, and if someone sufficiently manages to pass that test through anxious perfectionism, mission fucking accomplished.
[Interestingly, this take reminds me of the Confucian focus on ritual; “yeah, they’re not going to understand what it’s for, but it’ll help, and it’s by just doing the ritual that they eventually figure out what it’s for, with the side benefit of not causing problems in the meantime.”]
The “false advertising” charge, as I understand it, is that ‘production in 2019’ is not the right goal, and a different thing is, and this distinction is worth making because it’s correct. [I noticed that I really wanted to put an appeal to consequences after the because, as I think it makes the statement more forceful / relevant, but it wasn’t how my inner Benquo would put it; the point isn’t that people will be confused and that’s bad, the point is that it’s not correct.]
Altman’s responsiveness test might be more useful if it measured a founder’s reply time to an underling rather than to the guy holding the purse strings/ contact book.