3 ms·

I know this is all in good spirit and you just want to show some appreciation for the help. But rewarding programmers by number of commits is a bit perverse. No

I know this is all in good spirit and you just want to show some appreciation for the help. But rewarding programmers by number of commits is a bit perverse. Nowadays everyone generally agrees that paying programmers per line of code is a bad idea. I think rewarding based on number of commits creates very similar unwanted incentives and should probably be avoided completely.

collyw 13y ago
Agreed. I was trying to package some Python files yesterday, so I could easily install them via git / pip. That took a lot of commits before it actually worked.
ColdHawaiian 13y ago
I guess a better measure of contributions would be number of distinct features implemented and number of bugs fixed.
konklone 13y ago
We could also do PRs accepted, or issues closed, or other things a little less squishy than commits. But either way - it'll probably be fine.
Loque 13y ago
Needs backlog with priority value, progress state... basically paper trail. Also, I like the idea of this project, but I hate the pitch. I don't know why. Maybe it is because programmers already go above and beyond, taking care of the stuff that no-one else thinks over. I would actually suggest investing time in looking to fund better development. Just my 2 cents - great idea, best of luck!
- unhush 13y ago
  Thanks for the feedback! The tech team at EFF is pretty overloaded all of the time; we've hired outside contractors before but this is a project where it seems more efficient (and potentially more fun per person) to crowdsource. The same goes for HTTPS Everywhere (https://www.eff.org/https-everywhere https://www.eff.org/https-everywhere), a browser extension that I maintain where almost all of the 10,000+ XML (ugh!) http-to-https upgrade rules are contributed by volunteers. I think EFF is in a rare position to experiment with crowdsourced programming tasks, because (most of) our contributors don't have any incentive to game the system and deliberately submit low-quality work, which decreases the usefulness of results you get from money-driven crowdsourcing platforms like Amazon Mechanical Turk. This particular experiment was far more successful than I predicted, which brings up the question of whether there is a systematic way to make crowdsourced web development effective.

collyw 13y ago

Agreed. I was trying to package some Python files yesterday, so I could easily install them via git / pip. That took a lot of commits before it actually worked.

ColdHawaiian 13y ago

I guess a better measure of contributions would be number of distinct features implemented and number of bugs fixed.

konklone 13y ago

We could also do PRs accepted, or issues closed, or other things a little less squishy than commits. But either way - it'll probably be fine.

Loque 13y ago

Needs backlog with priority value, progress state... basically paper trail. Also, I like the idea of this project, but I hate the pitch. I don't know why. Maybe it is because programmers already go above and beyond, taking care of the stuff that no-one else thinks over. I would actually suggest investing time in looking to fund better development. Just my 2 cents - great idea, best of luck!

unhush 13y ago
Thanks for the feedback! The tech team at EFF is pretty overloaded all of the time; we've hired outside contractors before but this is a project where it seems more efficient (and potentially more fun per person) to crowdsource. The same goes for HTTPS Everywhere (https://www.eff.org/https-everywhere https://www.eff.org/https-everywhere), a browser extension that I maintain where almost all of the 10,000+ XML (ugh!) http-to-https upgrade rules are contributed by volunteers. I think EFF is in a rare position to experiment with crowdsourced programming tasks, because (most of) our contributors don't have any incentive to game the system and deliberately submit low-quality work, which decreases the usefulness of results you get from money-driven crowdsourcing platforms like Amazon Mechanical Turk. This particular experiment was far more successful than I predicted, which brings up the question of whether there is a systematic way to make crowdsourced web development effective.