Hacker News Reader

Comment by pg | Hacker News Reader

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.

View on HN

>> doing statistical analysis of YC applications

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

Note to self: There is actually a very sparse body of literature that talks about data driven approach to VC, e.g., http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=4621..., this one seems interesting (predicting exits): http://www.springerlink.com/content/w024ku34221u3258/

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.

It would be awesome if you could share some of those discoveries.

In their position I wouldn't: it'd lead to indicators being gamed and losing their usefulness.

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.

Reply on news.ycombinator.com