YC "bugs" are
- startups that are accepted and fail
- startups that are rejected and fail, but would have thrived had they been accepted
(Of course this last kind of bug is extremely difficult to track down.)
But if a startup makes a killing after being rejected by YC it means they didn't need you.
If you consider this a bug, there's some kind of moral problem because it means YC is trying to siphon every promising team in order to maximize YC's success, independently of who YC is likely to help more.
You're not just betting on horses, you're also training them: it seems what you should want to maximize is YC's impact...?
YC is trying to rewrite a lot of the rules of venture investing, and has had quite some success there, but at its very core it's important to remember that they are investors. At the end of the day, the measuring stick of an investor is their returns, not how many (proto-) companies they pulled from the brink of implosion. As such, your second measuring stick rings false.
I suspect that AirBnb, Dropbox and Heroku are simultaneously among the investments that YC is the most proud of and the companies that would have been most likely to succeed had they not been a part of YC.
PG has specifically stated that this is not a bug (it's a feature). The goal of YC is not to take mediocre or bad startup teams and turn them into wins, but to take the winners and turn them into Super-Winners(Dropbox).
As far as maximizing YC's impact, I think Dropbox maximizes it more than small wins from teams that would have lost had they not been in YC.
I think that's how it is, yes, and PG seems to have stated as much; in essence, you either die or don't die; if you don't die, the scale of your success depends on a lot of things that are beyond your control.
(Actual quote from PG: "The formula for the success of a startup includes a random multiplier of roughly zero to a thousand." — http://news.ycombinator.com/item?id=3050280 )
If one subscribes to this point of view then the purpose of YC is to help startups not die, and the purpose of the YC admission process is to
1. select all startups that have a chance of not dying if admitted (eg, refuse startups that don't stand a chance in hell of not dying)
2. if the sample from #1 is too large, reduce it at the low end first (startups that are not very likely to not die) but maybe also at the high end (startups that seem certain to not die no matter what)
Maybe reducing the selection from the high end is just plain stupid / counter productive, but if you don't do that, it's not very clear why one would be in this business in the first place (as opposed to other lucrative businesses).
These guys got 10k users in a month. But maybe they'd get 100k in 3 if they had YC. Even if you understand that either way it would be a success anyway. It's still reasonable to assume that it would be a worthwhile deal for them. If YC missed the opportunity to help them get even better, even faster. It's reasonable for both sides to treat that as a bug.
The mission of a prep school is to help good students become brilliant; if it becomes stuffed by already brilliant candidates (for example because the admission process is more selective than the admission process of the college it preps for), so much so that there's no place left for second-tier students who would make the most of its teaching, then there's a problem.
I'm just wondering if this is something YC worries about or should worry about.
Also, elite prep schools do try to fill their classes with nothing but brilliant students, for a variety of reasons, including money and prestige. There are some negative consequences, to be sure, but it happens. At least Y Combinator increases the size of its classes every year, unlike prep schools.
Prep schools are businesses too, BTW. And yes, businesses have a responsibility to their shareholders to maximize profits. But if that was the only purpose of any business, everyone would be selling luxury watches.
YC is not a pension fund putting money in a VC; YC "makes it happen". They invest in very early stage companies because they believe that they can help them succeed (and they do).
In 2004-2005, PG famously predicted the second coming of Apple; if he had put all of his money in Apple stock then, he would be 17 times richer today. And yet that's not what he did. At the same time he was making this prediction, he founded YC.
- - -
I can't seem to explain myself clearly, but I'll give it another try.
The total return of an investment fund is really the return on a very few top investments, which tend to dwarf all others.
Now, if you had a magic prediction device that let you identify for certain the next Google, it wouldn't make sense from a profit-maximizing point of view to invest money in any other company: just put all your money in the next Google and you're done.
It's safe to assume that YC, who have been in business for 7 years (really 14 since they have 2 rounds per year) are getting better at what they do. They should be able to predict who's likely to succeed and who's not, better and better each year.
Therefore in all logic they should be reducing the number of startups they fund: put more money in sure winners and don't fund anybody else.
And yet they're doing the exact opposite: each batch is larger than the previous one.
There can be only two explanations to this: either they're getting worse and worse, and are spreading their bets; or, they're getting better and better and can afford to offer their services to more and more companies.
It seems likely that the second option is the correct one.
>> put more money in sure winners and don't fund anybody else
That is not their business model. Their model is to take a small fraction of the company for a small amount of money, and then help them succeed. >> There can be only two explanations to this: either they're getting worse and worse, and are spreading their bets; or, they're getting better and better and can afford to offer their services to more and more companies.
So far YC has a fantastic return on investment. I imagine that their returns are the envy of many conventional VC firms.They invest so little per company that they can afford to invest in every company where the expected return on investment exceeds the amount invested.
As they become better known, they get more and more qualified applicants, and thus the number of bets that they make should be expected to increase over time.
Even if a company succeeds without YC that doesn't mean that YC's potential involvement is a no-op. On the one hand that represents a missed opportunity for revenue. It also means a missed opportunity to perhaps accelerate the growth of that company beyond what it has managed to achieve on its own.
Bugs in the application process can only really be rooted out by startups that apply and are successful.
Startups that apply and fail could point to a bug in a number of different areas - not just the application process.
If the purpose of YC is to make money and improve startups, than they are failing at both of those things if they pass on startups that eventually succeed.
A few thoughts:
- It would be interesting to see what made them 181 instead of 179. Would applicants this close to the cutoff get feedback of any kind about how they fared?
- Whoever ranks 271 might be missed out too in the future.
- Has any thought been given to sharing "you're this close" details with the 50 or 100 applicants that were within the proverbial whisker, or giving them a second review to see if any would move up or down relative to their peers in the same range?
0.02
That said, interviews and applications are inherently subjective and based on a snapshot in time; they will always have "bugs". It is a tough thing to do, and I fully appreciate the effort going into it.
Best of Luck!
Said another way, the delta between where Codiqa stands today vs. where they would be had they been interviewed does not seem as large an opportunity as probably the opportunities you made for the companies you did select. Therefore, I'd say there's no bug.
Like someone said "Micheal Jordan would NOT want to be selected by YC".
This means people like Mark Zuckerberg, Larry Page, Sergey brin, Bill Gates, Steve Jobs are not likely to come out of YC. Because the very fact that they have to go through layers of interview sessions, and then spend time performing to you, to get your attention like some performance artist performing to the king to get his attention means that is not going to happen.
Because people screw up in interviews. And a interview is generally no measure what the true story actually is. The guide to cracking interviews and performance is learning how to game the system. If your process is known to people outside,and they plan enough to cover all the corners and deal their cards well, they can game your interview process. That's how job hoppers who can do nothing hop around in circles of big successful companies. The situation from evaluating a 'good product' changes to evaluating 'how good the interview went'.
Nerds, Geeks of the best quality hate all this interviewing, playing-to-the-galleries-business. That is why they wish to build their own firm. Now if you subject them though the same bureaucratic process which they had otherwise go through in a traditional large corporate(Applying, interviewing etc) they are just going to go away, take money from somebody else and build some stuff in the same time.
If I would be building a business and need funding from somebody. I would go to them and explain them 'whats in it for them' and take their money.
But if have to go through the same large corporate styles process for this, I will be spending my time for something else. Because the very purpose my starting a business is to avoid that corporate styles processes.
You have to seriously ask what value you bring, or how different you are from anybody else. If you are just evaluating profit prospects(which I agree, is important), funding, taking profits back. Then really you are just another VC. There is nothing wrong with that. But that is what it boils down to.
I went back to look at the application. It fell immediately below the threshold for invitation to interviews. I.e. if even one of us had given them a grade one step better, they would have made it to interviews.
When we make mistakes, it's usually in this phase. Most of the time when we screw up it's by not inviting a group to interviews, rather than by interviewing them and turning them down. So this startup is exactly where a mistake would be most likely to occur: in the applications that fell just short of the cutoff for interviews.
The good news is, we'd already decided to fix this problem. We're going to do 3 interview tracks for s2012, which will let us interview 270 groups instead of the 180 we did last time.
If we'd done that last time, we would have interviewed these guys. But last batch was the first time we did parallel interview tracks at all, and we needed to test whether it worked for 2 before we went to 3.