Hacker News Reader

> 1. Make this "HN selects startups which get invited for YC interviews", i.e., a feeder into the existing system. Essentially, use HN to supplement the network of YC alumni who help out with the application-filtering.

I'd actually like to see the winners bypass an interview stage. As discussed on the previous results thread I think this is an opportunity for YC to shed any biases in their interviewing process and see what the results are.

Are there biases in the YC community that will come into play? Surely. Are they different biases than the YC interview process? That's why we experiment.

> Make this "HN selects startups which get invited for YC interviews", i.e., a feeder into the existing system. Essentially, use HN to supplement the network of YC alumni who help out with the application-filtering.

If you're going to have a interview step, then by all means list it. But I think it's an open question whether the interview step is a good idea, at least in the context of relatively rare contests for a Fellowship spot.

I also don't think a more elaborate voting system is needed, but requiring voters to have a certain karma and/or account age threshold seems like a good idea.

> Use a more sophisticated voting system.

How about only HN accounts with a verified email address can vote, and then you vote via a "Sign up for their mailing list" button? That way you're forcing voters to have some skin in the game, and also leaving the startups who participate with a tangible asset.

I'm pretty sure most HN readers can bring to bear a nearly infinite number of email addresses with minimal effort.

Also a karma threshold?

Age of account is a better metric in my opinion.

Why?

Well, I've had my account for over 2500 days, and my karma only yesterday broke 1000. But I read HN daily, and feel part of this community.

I just only comment when I feel like I can add something to the conversation, and always pause before commenting (stops me from getting into flame wars)

So age of account should have some say, or at least if they can see how often you read and vote for stories and comments here.

> 4. Use a more sophisticated voting system. I think a "which of these two looks better" combined with a form of Elo rating could work quite well.

I think that's a bad idea given how different the startups are. It doesn't make sense to me to compare for example AirBnB (what it was when they applied) with Tarsnap or AutoMicroFarm with Cruise.

To me, the difficulty of comparing startups makes asking people to compare two at a time even more important: If comparing AirBnB and Tarsnap is challenging, how can people possibly compare AirBnB and Tarsnap and AutoMicroFarm and Cruise and 14 others at the same time?

I think a bigger problem is what criteria are we supposed to be selecting the startups on? My application wasn't even a startup (just an idea). Should people have voted for it if it was a good idea, if they think I have the track record and resources to build it, if it solves an interesting problem, if it could be a unicorn, if it could raise future investor funds, etc, etc?

Yes, that definitely needs to be clearer. Given the context of "wisdom of the crowd vs. wisdom of YC", I interpreted the sole criterion to be "YC will benefit by making this investment"; it seems that many people took it instead as "who do we want YC to give $20k to", which is a very different question...

The mere fact of who won suggests that you and I were in the minority on this. Most of the problems Dan and Kevin have suffered is due to most people not thinking of what is best for YC.

I agree that we were probably in the minority, but statistically speaking it doesn't immediately follow: Votes for "who do you want YC to give money to" may have been far more concentrated than votes for "who will YC benefit by investing in". I don't know the total number of people who voted; if most people only voted for one or two submissions, then the pinboard-voters would be in the minority.

Only YC knows the answer to this. I would not be surprised to learn that the votes were split into a large pool of people that voted for Pinboard only and another large pool that voted for multiple applications. All I know for sure is they were not voting for mine :)

At what point do you start paying the voters for all this work?

Which probably isn't the best phrasing, but I'm serious about the question, for people on HN the motivation to participate is that it is fun or interesting or whatever, so I wonder how many rounds of votes you can squeeze out of that.

That's a very good question, actually. I'm sure there are quite a few of us who take "interesting startups happen" as motivation enough, but if this experiment becomes routine there might well be a reduction in the number of hours people volunteer for this.

I have a lot of suggestions to make, but I am waiting for the postmortem thread that Dan has said we are going to have to discuss the experiment and how to improve the process.