snuxoll.com

    There have been a lot of great ideas for support systems, whether they be simply an IRC channel or forum, or a bug tracking system like launchpad (or one of the other thousands out there). They all have their own pros and cons, and often times I've seen problems bounce back and forth between them because while each solution works, they aren't always what you want at that time, but they may be 10 minutes later for the same issue. So, after some idea bouncing today, I came up with an idea, a hybrid support system.

    So, now you're thinking I'm crazy, I know, but bear with me and finish reading until the end. If something is still unclear at that time you can feel free to leave a comment or send me an email (stefan AT nuxoll DOT eu DOT org). Now with my contact info out of the way, let's begin, shall we?

    Forums are a great tool for keeping an archive of issues and their solutions, along with bug trackers. The issue with these support methods, is they are generally slow, and it's painful to use them in a realtime manner even if you do post within a minute or so, because they weren't designed for such use.

    Realtime support methods like IRC have their own issues, it's hard to keep an archive of issues, even if you log a channel, because they get mashed together with other conversations. With tab-completion and nick hilighting it's generally easy to keep track of YOUR conversation in an IRC channel while other people talk about different things, and that's what makes it a unique place. But this is a double edged sword, while it's easy to keep track of coversations at that moment, it's a pain to do in an archive.

    Thus, I propose the completely unpatented snuxoll hybrid-support system. We need a way to keep everything related to an issue together (like on a forum or bug tracker), but we need to be able to have A) realtime chat and B) ad-hoc comments. This means we can't do a traditional IRC channel that just has logs posted, because issues get mixed together in one giant mess, but it also means we can't have a forum because you can't do realtime support on them.

    Thus, we need a ticket/issue system, combined with a realtime chat method (such as, say, Jabber group chat). You post a issue online by filling out a form, this will generate a new ticket/issue and you are automatically subscribed to it. Each ticket has their own groupchat room on a jabber server, and a logbot automatically joins it and adds the logs to the issue on the site. In addition, people that don't want to join the chat can simply post a comment on the issue online, and the logbot can notify the room of a new comment.

    Doing this would make it easier for everyone to contribute, and when an issue is solved you basically have a knowledgebase article compiled, all ready to be linked to when someone else has the same problem. So not only would getting and giving support be easier initially, but it would be easier to find solutions for people with the same issue.

    Moving on from the realtime vs adhoc support side of things, another issue is free vs commercial support. I love community support, I really do, I enjoy helping others and I like getting help when I need it. However, when you're trying to get help with some obscure or overly complex issue, not many people want to spend their time on your issue. One way the open source community has generally solved this issue is with 'bonties', small amounts of money that are put up for fixing bugs or aiding in finding a solution for a problem.

    In such a hybrid support system, a bounty system would be an easy thing to add, you simply tack on some money to your ticket, and the person(s) who solve your issue get to collect it. This would mean people could collect a little extra money on the side, and the person posting the bounty may get better support (or support at all, in some cases).

    I'm sure there's probably a few things I haven't thought through, and if you think I missed anything feel free to comment. But this post has gotten long enough and I think it's about time to check 'publish'.