Ask YC: Suggestions for funding a small-scale open-source development project (a la the X Prize)

[kn0thing]

Greetings, YCers. I'm planning to announce a "breadpig prize" (http://breadpig.com/prize.html) but wanted to get some feedback on implementation.

Ideally, I'd like to use about $3,000 I've earned in profit from various breadpig projects (tees/unholidaycards) as a reward for creating a nifty little (open-source) app.

Is there some software you wish existed (that would be worth only 3K of motivation)? I'm thinking along the lines of a nifty FireFox extension -- the sort of thing you've always said you'd build yourself if you weren't so busy.

The other problem is creating a reward system that makes it even worth attempting. What are the criteria by which a winner is chosen and is it even possible to create something on this small scale that makes it worthwhile to developers to even attempt to be that 1 winner (you all are better than most at probability).

Thanks in advance for your thoughts.

-Alexis (the reddit co-founder who didn't know Lisp)

[martythemaniak]

I'm writing a web app that has a PHP/JSON backend and a very javascript-heavy front end. After spending some time in JS, it really sucks going back to PHP, as the language itself lacks so many things Java/ECMAScript has.

So the idea is to be able to use EMCAScript as a backend programming language with the ability to access files, SQL databases, sockets, do system calls etc. Since its too much work writing these from scratch, the project could be a cross "compiler" between ECMAScript and Python (or any other suitable language), so that people can write web software in one language only.

[apgwoz → martythemaniak]

are you looking for Helma? (http://dev.helma.org/) It's basically what you asked for (Javascript backend and front end) but without the Python.

[soundsop]

I like the Netflix competition model where there is an objective measure that is being optimized.

This allows there to be a leaderboard where there is publicity benefit for people who do well, even if they don't win.

[emmett]

I think that's an awesome idea. I've wanted to do much the same thing. I think putting the money out purely as a bounty will probably work fine, and if it doesn't you can always try something else instead.

A few ideas I've had for plugins I'd like:

A browser plugin that makes zip files transparent, so you can browse them and open pieces of them.

A browser plugin for native multiple file upload (credit to Paul B. for this one)

A p2p file transfer app built into the browser.

I have more specifics on each if you're interested.

[shwouchk → emmett]

Opera has a torrent transfer client built in since version 9 (if I'm not mistaken) I'm too lazy to check at the moment but I'm willing to bet there's a firefox extension for downloading torrents as well...

[kf → emmett]

multiple file upload would be great! I vote for that.

[makanikai]

I'd love an extension that garbage collects my browser tabs.

I open a lot of tabs. Closing them is painful, and something I usually put off until my browser crashes, or I have to do a restart.

Then there's this process of manually serializing each unread URL to disk (cutting and pasting each unread URL and emailing it to myself).

I'd like something that does two things.

- Make it so that I never have to close a tab again, when I get to 15 or 20 active tabs, this firefox garbage collector starts writing them to bookmarks/saved pages, and closing the oldest ones.

- Provide some mechanism for winding back the wheel - and either searching the contents of - or just scrolling through my history.

I'd love to write it today, but feel my time is better spent hacking on mobile devices. If I don't find an implementation of this within a couple of weeks - I'll probably reconsider.

Cheers to anyone who sends me a link, or writes the code that makes my dreams possible.

[zach]

For a large prize, you can get a small amount of experts to work for a long time on something straightforward but technically demanding.

For a smaller scale, flip everything. Get a large amount of dabblers to work for a shorter time on something that might be hard to quantify but is not really hard to do.

So maybe you should reward inspiration that has just enough execution. That is, enough that people get it and it will sustain the project, but not too little that someone would have done it without the prize.

Now, I'd say get some more specific goals than "nifty" because otherwise we're all going to say "libraries for Arc" which is nifty to us but maybe not yourself.

Plus if you pick some criteria that inspire people, are clever and/or people really get behind, you can get a lot more exposure, which will increase the expected quality of the winner. And then maybe you can hit up some people at Community Next or something to add to the prize pool.

But if you really do want merely nifty, you could just set up a restricted subreddit for judging... although I'm sure I didn't have to tell you that!

[m0nty]

I'd suggest that $3,000 wouldn't pay for a great deal of development of any kind (except you could theoretically get three-and-a-bit Digg clones I guess ;) so maybe you should look at hacker-friendly prizes instead. XKCD goodies, 100,000 reddit karma points, something like that.

I used to run a fantasy football league in a school and there was massive interest just for the fun of playing and the kudos of winning. The only prizes on offer were a team shirt, a pair of boots and a wooden spoon (for the loser, natch). If you could find similar things which would appeal to hackers, you'd spend less, get more interest and get to keep some cash for next year.

[maximilian → m0nty]

$3,000 would pay for my services for a summer. It would for a lot of students. I'd love to have a cool flexible summer job where I get to implement something interesting. Shamelessly i'm putting my name into the hat. I'll build whatever cool thing you want.

You're basically suggesting google summer of code on a smaller scale. I'm applying for that too, but its a long shot.

XKCD goodies and karma points sure don't help pay for bread, rent or tuition. Or all those other things I would buy if I had a reasonable income.

[tim2]

1. Divide the money into several prizes.

2. Make some sort of contest. Perhaps for a slightly academic challenge that has immediate applications? Eg. recommendation algorithms, AI problems, compression, encryption, p2p problems.

That should give you maximum return on this money.

[SwellJoe]

I've done this many times in the past, if on a more private scale (e.g. on the developer mailing list of a particular project). In fact, a few years back, I started work on a website to facilitate this kind of project, though I never launched it (but I do still hold the ransomware.com domain name just in case I decide to come back to it). All of those projects ranged in cost between $500 and $20000, and most eventually turned out good (though I sometimes had to double down and hand the code over to someone else, or work on it myself, to get it polished and ready for production).

In the Open Source world, this kind of thing is not at all rare or unproven. It can work extremely well, and if it's something that works within an existing community of users and developers you get free quality control and vetting of the results...if it's a standalone project, I suspect you'll be doing your own QC. Though I'm sure you wrangle some of your fellow YC'ers into helping bang on it if it's generally useful.

Firefox plugins are a good choice for this kind of model (and I had a Firefox plugin made for about $4k about three years ago, in a very similar process), as they are small, easily testable by a wide audience, and can leverage existing infrastructure for the hard stuff.

The tricky thing, however, which you've noted, is that since only one prize is being offered, you'll have a hard time getting real buy-in from a large number of developers. $3k guaranteed is a great motivator for a student to spend a month working on something. A 1-in-50 chance at $3k is far less of a motivator. So, I've almost always broken it up into a "prove you can and really want to do it" phase and a "OK, now do it" (and the occasions where I didn't I already knew the work of the people offering to work on it from their Open Source contributions--this is tricky if you aren't following their projects reasonably closely). I didn't always phrase it as a "prove you can do it" project...instead I would pick the highest value low effort part of the project and offer a pittance for the first person to solve it--$500 is plenty, if the task is small enough. Then when the best version is delivered, I'd hire them for the rest of the project.

All of this begins to sound more like traditional contract work, rather than a contest, of course. But it does work, and making it a public process does tend to get more community involvement than a "I hired this guy to work on this project and now it's done, let me show you it" situation--if you need to get the project merged into the core project codebase, this separatist practice can make the merger very difficult...personalities get bruised if they think you're absconding with good ideas and dropping in a bunch of code they've never seen or talked about.

[tim2]

Well there is one OSS app that I'd really like to see:

- A successor to (no longer maintained) centericq