Garry has always been one of my favorite YC founders, and not only because he bought me a large quantity of gin shortly after Posterous got funded. Aaron made a huge impact on my life through his gutsy blog post on RSI [1]. Congratulations, guys, and thanks for being awesome.
What software does YC write for itself again? I've heard a bit about the software for collecting and analyzing applications to YC but I'd like to hear more. Is the HN software mostly written by pg and volunteers or do other hackers at YC do a significant amount of it? Is there anything else?
HN is mostly written by me and Rtm, but Garry and Aaron have already started building new things. Garry made a sort of private Facebook for founders in the current batch, and Aaron has started in on the problem of doing statistical analysis of YC applications and has already made some interesting discoveries.
What an awesome idea. You must have over 25,000 applications to-date, correct? That's enough to tease out some interesting, statistically significant correlations. Looking forward to hearing more about those results.
PG or Aaron: Could you de-identify YC applications data and make it publicly available for others to analyze?
[EDIT:] Or maybe you could make a prediction challenge using the YC application data? You could host the challenge on http://www.kaggle.com/ or http://tunedit.org
It would probably take them a lot of thought to fully de-identify the data and once they did it might be disappointing. A lot of so-called "de-identified" data out there actually contains things that could be used to identify people.
I would assume that it would be just about impossible to de-identify YC applications, as they are all fairly unique and there is generally a lot of information online about each startup. So if an app had a story about a previous product and then that story shows up again in their blog or when they are doing publicity it makes it pretty easy.
They could release it like this: pick a small number of fields like number of founders and age, and release each of them as separate arrays. Ex: {"founders": [3, 2, 2, ...], "ages": [31, 22, 27, ...]} (the size of the founders array and the ages array would be different because the first is per-startup and the second is per-founder) Just to get an idea of how vague de-identified data might be. And I'm not even sure that would be de-identified enough.
As an investment fund, your trading algorithms are essential IP. I realize YC does a lot more than simply put money in companies but the statistical analysis used to determine what early stage startups to bet on is probably something they'll keep close to their chest.
Congrats! I knew Aaron at Google, and he was one of the smartest engineers around.
He also built one of the coolest pieces of software I've ever seen (Robocraft). It was a Starcraft clone where users had to code up AIs for each of the units (and then those AIs battle each other). MIT uses it as the foundation for a programming competition that they run every January. I think it was done by him and one other person.
Most of that software has been rewritten several times over now, but the competition lives on (it's called Battlecode now)! We actually open sourced a bunch of the code base last year if you're interested in checking it out [1].
Additionally, the competition has since been opened to outsiders so you can participate even if you're not at MIT. We never publicized that fact very well, but I imagine it would appeal to much of the HN crowd. I also recommend checking it out if you've got some time to kill [2].
Trying to make a competitive AI for Battlecode is uncannily similar to trying to create a successful startup (and much, much less intimidating.) Aaron deserves a lot of credit for starting it (the other founder is David Greenspan, who also did Etherpad with Aaron.)
What's the role of a venture partner at YC? For instance, are all partners present at interviews? Do different people do different stuff? Or does everyone kind of do a bit of everything?
They're not investing any money. They'll advise us how to invest existing funds in the sense that like all of us they read applications and participate in interviews. But the main thing they (and we) do is advise people we've funded.
Congrats, guys! I had an opportunity to meet Garry and Aaron on a few occasions, and I was blown away with how smart, driven, genuine, and kind they are. Of course real experience building successful products helps too.
Being able to grab office hours with these guys is a massive boost to YC companies. Keep on rockin'!
[1] http://aaroniba.net/articles/tmp/how-i-cured-my-rsi-pain.htm...