Citebite creator: Amazon.com S3 is "my secret sauce"

AFTER OUR article on the nifty Citebite.com service, this scribbler decided to have a chat with Ronald Reeser, Ruby on Rails programmer and creator of Citebite.com

He tells in this talk with the INQUIRER how Amazon.com was his "secret sauce ", what kind of hardware is used to un the site, his ideas about financing a project like this, the copyright hurdles, and more.

INQ: Thanks for this interview opportunity Ronald. So, let's start from the beginning: When did you have this idea? when was the site born? Was it out of personal need of such a service like most of the software out there?

RR: Citebite launched in the early summer of 2006. I came up with the idea a few weeks before that. I wanted to experiment with extending hypertext. HTML allows for linking only to the document (or anchor) level, which is kind of coarse; Citebite allows for linking to the character level of a document. This extends hypertext in a pretty fundamental way. I did not have a personal need for the service save for very occasionally, but I thought it would be useful for others and heck, it might even catch on.

INQ: What language did you choose to create it?

Ruby on Rails, but Citebite doesn't take much advantage of the Rails framework. Most of the heavy lifting is done using the Prototype JavaScript library.

INQ: What kind of server does it run on?

RR: Citebite runs as a Mongrel cluster with an Nginx front end. The pages themselves are all flat files served directly from Amazon S3. S3 is the secret sauce. Serving files from S3 is highly reliable, highly scalable, and dirt cheap. Even if the Citebite web application goes out to lunch - as you witnessed yesterday - the cited pages will always be served from S3. This should give folks some confidence that they can use Citebite links in their pages without having to worry about them going stale.

INQ: Ron, I'm a big fan of the "KISS" (Keep it Simple...S..eriously :) philosophy, and your site does that to an extreme. There are other challengers now with the same idea but a much more complex design and - some could argue - feature-bloat. How did you resist the temptation of bloating the site?

RR: Citebite was set out to do one thing, and it does what it does reasonably effectively and elegantly, so I have elected to leave it alone. I was offering a Bookmarklet and a Firefox extension for quite a while, but I got tired of maintaining them and they weren't getting much use anyway, so they were discontinued. [He later agreed to re-upload the FF extension after this scribbler fixed the version check so it works with FF 3.0]

I have made some infrastructure upgrades to Citebite - for instance, it did not initially store cite pages on S3, and I have also upgraded the web and application servers as best practices for Rails have evolved. I will continue to make infrastructure upgrades, but I don't foresee any major changes feature-wise.

I also moved on to other projects pretty quickly. My current focus is Wakerupper.com - unfortunately doesn't work in the U.K. or Argentina yet.

INQ: So how do you plan to finance and maintain Citebite, long-term? I have not seen any Google Ads anywhere. Is this effectively a hobby project?

RR: Citebite doesn't cost much to run and so will remain self-funded for the foreseeable future. An overwhelming majority of traffic to citebite.com comes from cited pages - not from people visiting the Citebite homepage. If the cited pages are where the traffic is, that would be the most effective place to stick the ad spots. However, placing ads on cited pages would necessarily entail placing ads on other people's mirrored content, which I imagine would excite the ire of content creators - and rightfully so.

Placing ads on the homepage is another option but the homepage receives relatively little traffic, so it would in effect be adding clutter in return for pocket change and to me that's not a good trade.

I have zero revenue expectation for Citebite, so I just try to control the costs and enjoy the ride. Over time, Citebite should yield an interesting data asset - i.e. if someone cites something in a page, one can presume they thought it was important enough to go to the trouble to do so, and knowing what people think is important...is important. Perhaps an academic will one day want to take a crack at it, who knows? For now the exit strategy is: no exit.

I think of Citebite as more of a 'pet' project than a hobby project. Hobby implies lack of commitment to me, and I'm not planning on walking away from Citebite the way I start and stop and start and stop learning to play piano every year.

INQ: Don't you think there should be a standard (W3C, IETF or whatever) for web clipping? Shouldn't the web browser be able to store clippings along with bookmarks, for starters?

RR: I do think there should be a standard, and I rather expect something like that to come along.

INQ: Ever lost any sleep about the legal issues involved in such a service? did you get any complaints about the "mirroring" of content?

RR: No. Citebite has been the subject of some criticism by web wonks, but I have actually never received a take-down request.

INQ: Related to the previous question, do you offer the chance of having content deleted?

RR: The unstated policy is that if someone sends a request that a cited page be removed, I'll take it down immediately. The reason the policy is unstated is because, so far, it has not been an issue.

Ronald Reeser, ladies and gentlemen... the web needs more like him. µ