Web 2.0 day 1: Open Source Infrastructure

Email.Email weblog link
Blog this.Blog this

Robert Kaye
Oct. 05, 2005 01:46 PM

Atom feed for this author. RSS 1.0 feed for this author. RSS 2.0 feed for this author.

This morning the sold-out Web 2.0 conference in San Francisco started off with large crowds pouring out of the presentation rooms. This morning focused on Workshops, that here at Web 2.0, are informal discussions that attempt to get everyone's juices flowing for this evening and the next two days when the conference kicks into high gear.

The first presentation I attended was Marc Canter's "Open Source Infrastructure" workshop. Marc opened the workshop with this thesis: If we're now moving away from a closed Microsoft world to a world dominated by Yahoo and Google, we're still dependent on closed systems. How can we take advantage of open source and build open infrastructures? How can we open up existing services?

First to respond was Brian Dear, from EVDB. EVDB builds the Eventful project, which is an open system for publishing events. Eventful is based on the idea that "Events are hard to find on the web, and events are hard to publish on the web." -- this echos the very same sentiments I heard last month at the Future of Music Conference in Washington DC. Small bands have a hard time letting their fans know when they play gigs. And fans have a hard time finding gigs by their favorite bands. Eventful aims to solve this with the Simple Event Sharing service that builds on top of existing technologies and attempts to aggregate event information from anyone who cares to share this information with Eventful. Users can enter events directly, ping their servers via personal blogs or use the Eventful API to add and search events.

Matthew Mullenweg from WordPress chimed in about the Ping-o-matic project he is working on. Ping-o-matic is a weblog trackback ping aggregator that allows someone who posts a new blog entry to ping only one server, as opposed to pinging a whole host of services, which slows down the whole blogging process. Ping-o-matic represents a perfect example of an open source infrastructure project -- various blogging companies are working to set up Ping-o-matic servers to help the project along. This makes total sense, since Ping-o-matic is considering creating a non-profit company for the project -- this would keep this valuable ping data in the open.

Next up was Toni Schneider from Yahoo!, who responded to Marc's questions about Yahoo!'s recent acquisition spree of smaller companies like Upcoming and Flickr. Yahoo!'s aim is to keep all of these services unchanged and to relieve their smaller inundated companies of their burdens of hosting services that are gaining steam. Toni outlined how Yahoo is working on new open API to make many of their services open and accessible to the public. The most interesting thing, IMHO, that Toni said was: "Yahoo! doesn't care so much of official standards, but more about standards that are getting uptake". This is a subtle change from RFC standards (many of which are not in widespread use) to throwing their weight behind standards that are just now gaining mind-share. Of course, someone like Yahoo! throwing their weight behind a standard is a good way of getting a standard to gain more momentum.

The second thing that Toni pointed out was: "Services, not formats, are getting the grassroots following." I think he may be on to something there -- Flickr may be the perfect example for this. Flickr has gotten a tremendous following because it presents a useful service that sparks people's imagination, not because people care about the formats that make Flickr tick.

The last person to give us his views on open source infrastructure was Tantek Çelik from Technorati. Tantek's first point was the you control your own data. Currently email and your contacts are under your control and not some third party's system (unless you use Outlook, of course :-) ). True open source infrastructures should make sure that you remain in control of your own data -- much like how people can choose to control their own blogs, as opposed to having some third party lock up your data.

Tantek went on to talk about open data formats, APIs and services and how they must be open, free from royalties and must be publicly accessible. If any of these are not the case, then the system is not open enough to really qualify as an open source infrastructure piece. And echoing what Toni said, he mentioned that sometimes ad-hoc systems are more important that the accepted formal standards. Tantek went on to talk more about his microformat design principles, the community behind microformats.org and gave an example of a microformat system in the distributed review system called hReview. hReview is a great example of an open source infrastructure format -- it specifies a way to post a review about anything in a web page, syndication feed or an XML document. Technorati (and everyone else) can then find and index this new type of data, while everyone still owns their own review data. I like it!

From this workshop I hear that "Open, emerging/ad-hoc and useful without reinventing the wheel." are the keys for building open source infrastructures. I'm generally pleased to see the trend in these systems/formats towards opening them up and that larger companies like Yahoo! are throwing their full weight into this. Many people are realizing that closed or royalty encumbered formats are having a harder time taking off today when they face stiff competition from open formats that capture people's imagination. I think Flickr with their open APIs are a great example of this. I wonder what Flickr would've been had it not decided to be open?

Robert Kaye is the Mayhem & Chaos Coordinator and creator of MusicBrainz, the music metadata commons.