Comments for post Github reviews as a way to improve code quality?
thesp0nge writes: Hey antirez this sound interesting, would be a nice feature in the startup I'm planning to launch in the second half of 2011.
Can be a good deal, since I'm planning to connect user's github repo to provide code review facilities, to provide also some rating and commenting (in a social network style).
It can be also interesting to add a "Translate comments to Issues" feature to pull all the social interaction made on codesake.com in action to github repo... (of course only the report owner will see this button).
What do you think about it? Does it make sense?
toryburch writes: One of the reasons the current
Philippe Rathe writes: In addition to the proposed rating system, It would be nice that each tagged commits could be updated with user's ratings. By default it could take the last tagged commits user's ratings. So ratings could evolve only on tagged releases.
I would be also nice github as a "similar projects" feature so you could compare portential projects you would like to use using those ratings.
Anonymous writes: Maybe just read the code?
Brian Mitchell writes: @Marak, Great point. Github has excellent APIs but even more so, it's really just a URL. Sounds like a good project for those interested in to take on as a side project.
Anonymous writes: Why not add a 140 character "request" text field into the code quality rating that other users can "rate" based on how much value they think it adds. Users that originate a high number of high value fix requests over various projects can have their votes weighted higher.
This would incentivize:
People to give higher quality requests in order to level up.
People to be the first to request in order to be originators.
high value account holders to minimize spam in order to maintain their reputations.
It would dis-incentivize
Low value account holders, the most likely spammers, to post spammy ratings, as they won't be weighted much.
Can you game it? Sure. A high value player can build up several accounts and then sell them. But that can be fixed by tying the ratings to your profile's code quality score as well. So to become a high value player would take more of an investment and increase the cost of gaming the system. Thoughts?
Marak writes: If you want this functionality, you should build it on top of the github API and solve all the hard problems yourself.
Rankings and Ratings do not belong on github.com
HJ writes: Completely agree with honest guy. Two stars in performance would tell you nothing about where your code is lacking anymore than "it's too slow!" would.
Of course people can always open an issue or fork the project to help out.
Honest Guy writes: If your goal is to educate these programmers instead of scaring them away from open source you have to encourage serious reviews under which people put their real names.
Anonymous voting is simply not the right tool for the job.
antirez writes: @Miguel: yes it's a problem indeed, I happen to be the guy running the largest digg-alike site in Italy and I know a lot how bad this problem is...
Miguel Pilar writes: The problem with such rules is that they will eventually be gamed. See Stack Overflow's current situation, it'll become a circlejerk and/or spamfest.
antirez writes: @Honest Guy: what I hope is that in general the programmer's community is able to resist to spam a lot more than other communities. At least I hope so, but indeed this can be a real problem. You can still allow voting only if you have an account older than that, with at least N followers, and so forth? Not sure.
Honest Guy writes: If they ever allow anonymous voting i will make sure to spam Redis with 1-Star ratings into oblivion to punish you for suggesting this atrocity. ;)
mekdigital writes: +1