As always, this site is maintained and moderated by relsqui, who can be reached at that nick by PM or at firstname.lastname@example.org. Hermiene and khmer provided a lot of css help, zigdon bought the domain and maintains the RSS feed, and Randall drew the logo source and inspired the tagline. Diet-Drew and heero reported major bugs; root/cerise hosted the site until its move.
The original #xkcd quote database was created on December 15, 2006 to distribute the bother of keeping a personal quotelist. It used the now-defunct RASH quote management system and was housed on the shell account server b.armory.com until July 2008; at that time, it was rewritten from scratch with a more convenient UI and new spam protection and moved to an account at nearlyfreespeech.net. In April 2010, it was updated again to track up- and downvotes separately and rank quotes according to a Bayesian average score rather than a simple difference.
How It Works
When you vote on a quote, the site checks to see if you already have a cookie saying that you're a human. If you do, you can proceed unmolested. If you don't, the site displays a captcha; when you pass the captcha, it gives you the cookie so you don't have to do it again. The same cookie keeps track of which quotes you've already voted on, and doesn't allow you to vote again on quotes you rated while you had the cookie.
The effect is that a normal user who just wants to read quotes and mark the ones which are or aren't funny will only have to do one captcha per session. When they lose the cookie or it expires, they can vote again on the same quotes, but they'll have to do another captcha. A user who wants to spam vote a quote, on the other hand, will have to delete the cookie from their system and pass a new captcha for every vote, which makes spam voting by hand very tedious and writing a script to do so challenging.
The code XKCDB runs on is not currently publicly available. While I don't object to releasing it in principle, the code prep required is not high priority for me and I'm reluctant to accept responsibility for maintaining an open-source project. Feel free to contact me if you want me to know you support this, just be aware that it may not happen in the near future. There are at least two quote management systems which are already open source: RASH and Chirpy!. I'm not aware of one which is currently being updated, though.
Banana Mode premiered for April Fools' Day 2012; the injoke it's based on is explained here. Search works normally in Banana Mode (i.e. it searches the real quote text, not the altered quote text), and it defaults to off except on any first of April, Pacific time.