• home
• tech life
• management
• speaking
• glossary
• archives
• about

A Del.icio.us Interview

The original headline for this article hit me last week in a post-Turkey haze… laying on the couch… staring at the ceiling in the living room — “Why Del.icio.us is more important than Google”. Let’s hear it for SENSATIONALISM! Woo-hoo!

It’s hard to compare the two… one is a web service and the other is a company, but they do have a common goal — they strive to manage the endless pile of information that is the Web. They are both viewed as doing a successful job of this as measured by their ability to provide their users with relevant information… quickly.

I believe where Del.icio.us succeeds that Google does not is buzz latency. Like weblogs, Del.icio.us’s social bookmarking system does a fine job of identifying buzz quickly. A quick glance of the popular page and you get a pretty clear idea what Del.icio.us’s 30k+ community cares about.

Yes, Google indexes 8 billion pages and, yes, it serves up the results of queries to those indicies to, well, The Planet Earth, but Google chews on their large bites of the web relatively slowly. A monthly Zeitgeist reports tells me what The Planet Earth cares about, but I could pretty much guess that the most popular retail query on Google was Ebay. I was surprised that the #2 male celebrity query was Matt Drudge, but I don’t actually care.

I do care that Joel on Software is gathering the Best Software Essays of 2004. I’m also oddly interested in how to fold a shirt… free graph paper you say? Well, sure. These are topics I learn about from my anonymous del.icio.us peers… and that’s just the tip of the iceberg.

Del.icio.us more important than Google? Nah. It’s an AND, it’s not an OR. Can’t really write that article, so a different tact. We know there are several thousands Googlers in Mountain View… what about Del.icio.us?

Joshua Schachter is the decidedly low-profile author of Del.icio.us and he graciously agreed be interviewed via email.

RANDS: At whatever level of detail you prefer, what have you been up to since you graduated from Carnegie Mellon?

JOSHUA: I spend most of my time pressing the buttons in front of the big glowing thing. Occasionally, gunfire is heard.

From outward appearances, del.icio.us is the work of a single person… you… is this the case?

I do all the coding and other heavy lifting, but a cast of thousands contribute ideas.

Given there appears to be no revenue generated by your projects, how do you afford to eat?

I have a day job. I only work on del.icio.us one evenings and weekends. It’s not that expensive, just rack space and ALL MY SPARE TIME.

Do you expect to have to charge for del.icio.us at some point?

I don’t think charging is realistic. I probably could put ads on it to cover the bandwidth costs. I’m not really trying to make a business, just have fun.

Where did the idea for del.icio.us come from?

I had built a single-user bookmark system a few years ago. It had tags but otherwise wasn’t a lot like the current system.

I’d like to nominate del.icio.us for “Best Use of a Non-Dot-Com Name” — is there a deeping meaning to the name?

Not really. I’d registered the domain when .us opened the registry, and a quick test showed me the six letter suffixes that let me generate the most words.

In early discussions, a friend refered to finding good links as “eating cherries” and the metaphor stuck, I guess.

I somewhat regret using the domain name, because it’s almost impossible to discuss or verify without sounding silly. I’ll probably have to rename it at some point, presumably as something ending in -ster or -zilla or whatever.

From looking at del.icio.us from the outside, it appears you first design an architecture, throw it out in the wild, and then continue iteratively developing based off community feedback.  Is this a correct observation?   If so, how do you know when you’ve got enough of a product to throw it into the wild?  Is it a conscious choice?

I develop in the live system directly. I get pretty much immediate feedback about what works and what doesn’t, and I’m not above backing out a change I’ve made if it ends up working badly.

Usually I stay up really late hacking, and then as soon as I think I’m done implementing (but not debugging) I fall asleep. Then in the morning I fix all the damage.

Again, at whatever level of detail you prefer, can you give a high level architectural description of del.icio.us?

I’m using mod_perl, HTML::Mason, and MySQL. Pretty much standard stuff. It’s having some trouble keeping up with the load lately, so I’m starting to look into rearchitecting. The tags don’t map to standard RDBMS very well.

What’s the most interesting statistics you’ve found during the care and feeding of del.icio.us?

I built it to have an interesting dataset to look at for recommendation and so on.

Some folks have used it for interesting stuff. There’s a group of folks who use it to collaboratively find posts on one site, and they all use the same tag, and the feed for that is read by Livejournal…

I am a bit surprised that some people really just wanted to emit RSS feeds and nothing more.

I did have the expectation that users are much more clever and would use it for things that I did not imagine beforehand. I rarely tell people they are “using it wrong” or whatever (aside from the hierarchical categorization versus tags thing.)

Who do you consider to be your target user?

To start with: myself. I tend to collect lots and lots of links and need help managing it. The original predecessor system was built to manage tens of thousands of links I’d collected over the course of running Memepool.

After that, it’s a matter of listening to users, observing what users do (which is distinctly different,) trying to understand what people’s goals are, and what functionality can be added and what the interface cost is.

How many users?

30k or so.

A whole crop of third-party del.icio.us add-ons have appeared.  Do you have time to check these out?  Any favorites?

I love the nutr.itio.us bookmarklet, and I’m planning to integrate something very much like it into the site itself.

How do you approach user interface design?

Lots and lots of iterations until something feels right. Avoiding features until the interface for them is apparent. Seeing how users use the existing features to do things I didn’t expect, and then making those things easier.