An interesting infographic would be to plot the number of applicants and class size for YC over the years. I bet the applications are increasing at an increasing rate, last time they had 50% increase, soon it will be 100%, with all the publicity around YC.
I don't quite buy the scalability argument, in the end there's a natural upper limit on the YC class size and I think they're fast approaching it. There are a very small number of partners and there's only one pg. After that point, say a class size of 100 (~200 people) things will get impersonal and YC will inevitably start to converge to other incubators.
Large class sizes also inflate the alumni network, one of the best aspects (the best?) of YC. Up to a certain size, network utility increases with size; after that there are just too many people with too few hubs to connect them.
I think at the end, they just will have to fix the class size to a certain number, which will then make the acceptance ratios insanely small (currently 2%)
"I bet the applications are increasing at an increasing rate, last time they had 50% increase, soon it will be 100%, with all the publicity around YC."
I'll take that bet. You're assuming that there is an infinite pool of quality applicants.
Well, yeah, infinite in an engineering sense. Since YC applications are international, you can have tens of thousands of team applying. Based on the acceptance rate this time they had ~4250 teams, which should already strain the screening process. This could easily go up to 10K.
Maybe there's some limit on the size we can grow to, but so far we haven't hit a bottleneck we couldn't overcome.
Advising startups turns out to be pretty parallelizable. We can easily handle 20 startups per partner, so with 6 partners working full time advising startups, we currently have 50% more capacity than we need.
If I had to guess where we'll hit a limit that's hard to beat, it might be when the number of partners gets too large to communicate easily. E.g. with 20 partners things might get unwieldy.
Education, mentoring, and team building all happen better in smaller environments. Things like class size, team size, and time spent matter. Maybe what you mean is the actual time spent with each participant is still the same per investment as it was with smaller classes (i.e. you're all putting in more time than you used to) but somehow I really doubt this. I'm also sure that certain partners have certain strengths weaknesses that benefit in different areas and not having as many partners advising different startups would be less value.
That said, I'm sure Ycombinator's value is still great compared to other incubators, I just don't really feel like the statement "we're scaling just fine" on a squishy, hard to scale item like "advising" to pass the sniff test.
As an anecdote, I went through YC twice: once in 2009 with 30 companies and again in the last batch with 65 companies. I was skeptical at first as well, but if anything, the value my company got out of advising in the last batch was greater than the first time around. We could still meet with PG whenever we really needed him specifically, and the partners are really just as useful as PG in most cases. In many, they're better - being able to rap with Justin Kan about PR or Garry Tan about design is gold.
The one noticeable weakness is the decreasing familiarity between founders in the batch. My first time through, all the founders knew each other by name. This second time, there were other founders I still had trouble recognizing in the bathroom line at demo day. That's not necessarily as bad as it might sound - niches formed that grew close, and the camaraderie of this last batch was much better than in 2009. I don't really know why that is - just a more gregarious bunch in general, I think.
I suspect that the quality of the advice which YC can bring to bear on a given problem has improved more quickly through experience and contacts than the effects of larger class size has degraded the availability of any particular partner.
It would make sense if YC has added partners with strengths which fill the gaps in the existing partners expertise. To put it another way, PG is the public face of YC, but it has been shaped by JL from the beginning, too.
Advising is anything but squishy. Our performance is measured almost immediately by how well the companies do on Demo Day, and again over the next year or so in how fast they can grow.
If we started to get worse at what we do, we'd know within a couple months.
Even if that were the limit, and assuming 20 startups per partner, the idea of launching 400 startups every six months of ever-increasing quality and with ever-increasing network effects feels like a phase change.
They can't even remember all their names. Not sure how they can guide all these companies properly. Time will tell. It'll be interesting to watch none the less.
Ultimately, the real test of YC is in the quality of the startups that emerge from the program. I've been investing in YC companies since the second batch (W06), and I can easily say that the most recent batch was not only the largest ever, but also the best. Perhaps some of the social elements are lost with the size increases (not all of the founder know each other anymore), but the actual value of the program seems to be increasing due to our greater experience and the increased reach of our network. Quite often during office hours, founders will come to me with a seemingly difficult question and I can refer them to a YC company from an earlier batch that can help them out tremendously, sometimes just with advice, but often by providing the exact solution they need, or by becoming their first customer.
Obviously some portion of the gain in quality is due to YC attracting more high quality startups, but that is not the whole of it. I've watched these startups from Day 1 to Demo Day and the progress and improvements made are incredible.
When you say "the most recent batch was not only the largest ever, but also the best", I have to ask that the "best" is based on what metrics?
On average, in aggregate, everyone, the problem they're solving, age, education, experience, willingness to put in effort, IQ, etc.
Best from an investment point of view. I don't care about their IQ, education, etc -- I care about them starting successful businesses. Obviously we won't have the final answer for many years (big successes take about 5-10 years), but myself and many other investors found this group to be very investment worthy (I put more of my personal money into this group than any previous).
No, we don't administer IQ tests or anything. The point is that investors main interest is in finding good companies that will possibly be the next Google/Facebook/Dropbox.
But wouldn't IQ, education, etc. be factors in whether someone would even make it into the YC class?
If investors are not concerned with these factors, should they be relevant in the YC (or any incubator) application process?
(I realize asking this may be stirring a hornet's nest, and to anyone reading who is offended I apologize, but I am sincerely interested in the mind of the investor and their evaluation process.)
It doesn't really matter if participants have a low(er) IQ, or went to a non-elite school. As long as they are capable of creating great companies. YC does all the screening to make sure they are capable of doing so.
At some point there will be another layer to YC from successful exits in previous batches.
Already the number of partners is growing, and as more YC companies exit successfully you will probably see more of that, remembering your roots and who helped you up would be a powerful motivator into completing the cycle.
If YC can somehow manage to maintain communications through for instance multiple tiers with bypasses if needed it could get very big.
26 comments
[ 5.1 ms ] story [ 60.5 ms ] threadI don't quite buy the scalability argument, in the end there's a natural upper limit on the YC class size and I think they're fast approaching it. There are a very small number of partners and there's only one pg. After that point, say a class size of 100 (~200 people) things will get impersonal and YC will inevitably start to converge to other incubators.
Large class sizes also inflate the alumni network, one of the best aspects (the best?) of YC. Up to a certain size, network utility increases with size; after that there are just too many people with too few hubs to connect them.
I think at the end, they just will have to fix the class size to a certain number, which will then make the acceptance ratios insanely small (currently 2%)
I'll take that bet. You're assuming that there is an infinite pool of quality applicants.
Advising startups turns out to be pretty parallelizable. We can easily handle 20 startups per partner, so with 6 partners working full time advising startups, we currently have 50% more capacity than we need.
If I had to guess where we'll hit a limit that's hard to beat, it might be when the number of partners gets too large to communicate easily. E.g. with 20 partners things might get unwieldy.
Education, mentoring, and team building all happen better in smaller environments. Things like class size, team size, and time spent matter. Maybe what you mean is the actual time spent with each participant is still the same per investment as it was with smaller classes (i.e. you're all putting in more time than you used to) but somehow I really doubt this. I'm also sure that certain partners have certain strengths weaknesses that benefit in different areas and not having as many partners advising different startups would be less value.
That said, I'm sure Ycombinator's value is still great compared to other incubators, I just don't really feel like the statement "we're scaling just fine" on a squishy, hard to scale item like "advising" to pass the sniff test.
The one noticeable weakness is the decreasing familiarity between founders in the batch. My first time through, all the founders knew each other by name. This second time, there were other founders I still had trouble recognizing in the bathroom line at demo day. That's not necessarily as bad as it might sound - niches formed that grew close, and the camaraderie of this last batch was much better than in 2009. I don't really know why that is - just a more gregarious bunch in general, I think.
It would make sense if YC has added partners with strengths which fill the gaps in the existing partners expertise. To put it another way, PG is the public face of YC, but it has been shaped by JL from the beginning, too.
http://ycombinator.com/start.html
If we started to get worse at what we do, we'd know within a couple months.
Bubble?
Obviously some portion of the gain in quality is due to YC attracting more high quality startups, but that is not the whole of it. I've watched these startups from Day 1 to Demo Day and the progress and improvements made are incredible.
Would you say that is true of the other investors as well?
Are investors informed of the IQ, education, etc. of the each member in the group?
If investors are not concerned with these factors, should they be relevant in the YC (or any incubator) application process?
(I realize asking this may be stirring a hornet's nest, and to anyone reading who is offended I apologize, but I am sincerely interested in the mind of the investor and their evaluation process.)
http://news.ycombinator.com/s2012form
By adding more people to the community, you are not substracting resources, but adding to it.
I get your point, and agree.
Off topic -
Could you provide me an email address to contact you?
After reading your blog, I'd like to share something with you.
Regards,
RM
Perhaps something they seem to have or do?
Already the number of partners is growing, and as more YC companies exit successfully you will probably see more of that, remembering your roots and who helped you up would be a powerful motivator into completing the cycle.
If YC can somehow manage to maintain communications through for instance multiple tiers with bypasses if needed it could get very big.
Every brushfire starts with a spark!