The first talk of the day was a conversation between Tim O’Reilly and Evan Williams.
Evan started off by forming a company in his home state of Nebraska, then moved to work for O’Reilly Media for nine months but says he never liked working for other people. A little later on he formed Pyra, which after a year had Blogger as its main focus in 1999. They ran out of money in the dot com bust, had some dark times and he had to lay off a team of seven in 2000. He continued to keep it alive for another year and built it back up. Then Evan started talks with Google and sold Blogger to them in 2003, continuing to run Blogger at Google for two years. He eventually left Google anyway, says that it was partially because of his own personality (working for others), and also because within Google Blogger was a small fish in a big pond. Part of the reason for selling to Google in the first place was that they had respect for them, it was a good working environment, and also they would be providing a stable platform for Blogger to grow (eventually without Evan). But in the end, he felt that he’d be happier and more effective outside Google.
So he then went on to start Odeo at Obvious Corp. Because of timing and the fact that they got a lot of attention, they raised a lot of money very easily. He ran Odeo as it was for a year and a half. With Jack Dorsey at Odeo / Obvious, they began the Twitter project. Eventually Evan bought out his investors when he realised Odeo had possibly gotten it wrong as it just didn’t feel right in its current state.
Tim asked Evan what is Twitter and what Web 2.0 trends does it show off? Evan says its a simple service described by many as microblogging (a single Twitter message is called a tweet). That is, blogging based on very short updates with the focus on real-time information, “what are you doing?” Those who are interested in what someone is doing can receive updates on the Web or on their mobile. Some people call it “lifestreaming”, according to Tim. Others think it’s just lots of mundane, trivial stuff, e.g. “having toast for breakfast”. Why it’s interesting isn’t so much because the content is interesting but rather because you want to find out what someone is doing. Evan gave an example of when a colleague was pulling up dusty carpets in his house, he got a tweet from Evan saying “wine tasting in Napa”, so that its almost a vision of an “alternative now”. Through Twitter, you can know very minute things about someone’s life: what you’re thinking, that you’re tired, etc. Historically, we have only known that kind of information for a very few people that you are close to (or celebrities!).
The next question from Tim was how do you design a service that starts off as fun but becomes really useful? A lot of people’s first reaction in relation to Twitter is “why would I do that”. But then people try it and find lots of other uses. It’s much the same motivation (personal expression and social connection) as other applications like blogging, according to Evan. A lot of it comes from the first users of the application. As an example, Twitter didn’t have a system allowing people to comment, so the users invented one by using the @ sign and a username (e.g., @ev) to comment on other people’s tweets (and that convention has now spread to blog comments). People are using it for conversation in ways that weren’t expected. [Personal rant here, in that I find the Twitter comment tracking system to be quite poor. If I check my Twitter replies, and look at what someone has supposedly replied to, it’s inaccurate simply because there is no direct link between a microblog post and a reply. It seems to assume by default that the recipient’s “previous tweet by time” is what a tweet sender is referring to, even when they aren’t referring to anything at all but rather are just beginning a new thread of discussion with someone else using the @ convention.]
Tim said that the team did a lot for Twitter in terms of usability, by offering an API that enabled services like Twittervision. Evan said that their API has been suprisingly successful, and there are at least a dozen desktop applications, others that extract data and present it in different ways, various bots that post information to Twitter (URLs, news, weather, etc.), and more recently a timer application that will send a message at a certain time period in the future for reminders (e.g., via the SMS gateway). The key thing with the API is to build a simple service and make it reusable to other applications.
Right now, Twitter doesn’t have a business model: a luxury at this time, since money is plentiful. At some point, Tim said they may have to be acquired by someone who sees a model or feels that they need this feature as part of their offering. Evan said they are going to explore this very soon, but right now they are focussed on building value. A real-time communication network used by millions of people multiple times a day is very valuable, but there is quite a bit of commercial use of Twitter, e.g., Woot (the single special offer item per day site) have a lot of followers on Twitter. It may be in the future that “for this class of use, you have to pay, but for everyone else it’s free”.
20% of Twitter users are in Japan, but they haven’t internationalised the application apart from having double-byte support. Evan says they want to do more, but they are still a small team.
Tim then asked how important is it to have rapid application development for systems like Twitter (which is based on Ruby on Rails)? Most Google’s applicationss are in Java, C++ and Python, and Evan came out of Google wanting to use a lightweight framework for such development since there’s a lot of trial and error in creating Web 2.0 applications. With Rails, there are challenges to scaling, and since Twitter is one of the largest Rails applications, there are a lot of problems that have yet to be solved. Twitter’s developers talk to 37 Signals a lot (and to other developers in the Rails community); incidentally, one of Twitter’s developers has Rails commit privileges.
Tim says there’s a close tie between open source software and Web 2.0. Apparently, it took two weeks to build the first functional prototype of Twitter. There is a huge change in development practice related to Web 2.0. A key part of Web 2.0 is a willingness to fail, since people may not like certain things in a prototype version. One can’t commit everything to a single proposition, but on the flip side, sometimes you many need to persist (e.g., in the case of Blogger, if you believe in your creation and it seems that people like it).
So, that was it. It was an interesting talk, giving an insight into the experiences of a serial Web 2.0 entpreneur (of four, or was it five, companies). I didn’t learn anything new about Twitter itself or about what they hope to add to their service in the future (apart from the aformentioned commercial opportunities), but it’s great to have people like Evan who seem to have an intuitive grasp on what people find useful in Web 2.0 applications.