Monday, May 07, 2007

JavaOne 2007 (Day 1 – CommunityOne) Keynote with Tim O'Reilly

Update: Tim Expands on a lot of the topics he introduced in this Keynote in a podcast from EdTechLive 2007.


Tim starts and says that his first Open Source talk was in 1997 about the community aspects of open source. He talked about the history of the IBM personal computer and what we could learn about the shape of innovations - the open source paradigm shift

He asks how many in the room use Linux? How many use of us in the room use Google? He points out that this is one of the most widely used Linux applications but is not Open source. Technology has moved in some way. What does that mean? What does this mean in the context of the web? What are the killer apps of the internet? What do they all have in common? They all have the network as the platform. Sun were ahead of their time with the tag line “the network is the computer” (John Gage) They spotted the “network effects” phenomenon. These days most sites are built on Linux or free BSD. Few are open source. (Live Journal only) They are services and not packaged applications. They are data aggregators. He says he had a big argument with Richard Stallman about this. Tim's point was that if you had the source for Google you still couldn't run it. It's a set of servers, processes, and the data. These are big changes. This is a return to theme of community. Network effects for user contribution are the power these days. So what drives them?

This led Tim to web 2.0. He says that the focus on s/w licensing was a red herring. What mattered more is a modular architecture (Web Services) and innovative collabrative development using new tools. Users re now contributors. Web 2.0 companies use this in new ways. Alpha geeks showed the way.

So what does Tim mean by “Web 2.0?” He says it is like the “law of conservation of attractive products” (Clayton Christiensen)

Firstly the value moved from h/w to s/w. IBM didn't realise it and lost out to Microsoft. This is the world we grew up in. Dell are the best exemplar of how the h/w layer is now commodity. Intel could carved off a proprietary piece of this stack as did Microsoft.

So we move up the stack. Did it go to Linux and Open Office? No, instead the pattern repeated. Now s/w components have become commodity but there is still a system level lock in: proprietary info. What makes google so powerful? What makes all these new companies so powerful? It is not community per se but the design of theirsystems to harness network affects to get better the more people use them. Google's pagerank was the breakthrough in this.

Take craigslist as another example. It is built through user self interest. It is the 7th most traffic-ed site on the internet with but only has 25-26 employees.

Web 2.0 = “Bionic software” Now companies are leveraging the 6 million-human computer. It is all about building s/w powered by people. Think about Flickr and the tag cloud. This is a UI element generated dynamically solely by the users using it. Del.icio.us and digg too are powered by people. Compare digg vs wired news in site hits. Digg blows past them very early on in its life. Similarly LastFM don't ask you to do anything. You install their plugin and it listens in on your listening habits. They are instrumenting your PC. Pandora tried the same kind of thing with free music but they did this by analyzing you the user's likes algorithmically. This is not as good as simply looking at what you do.

This is all about the cornucopia of the commons. These companies are building on top of Open Source. Yahoo pays folks to build their directory. Wikipedia is another example. They harness the self interest and automatic behaviour of their users. They have an architecture of participation.

They use this to build massive databases. Google maps uses Navteq. They (Navteq) spent the money getting the map info. They are an example of the “intel inside” of the web 2,.0

There is also the concept of the “perpetual beta” (which Microsoft call “live software”) They all grow dynamically. “How cool is it that I can finish it tonight and push it live to the world tomorrow”. This is the converse to shipping the CD. Rails and Spring have helped this. Few people can build these sites quickly. Think of the two-pizza teams in Amazon. There is a real critical need for instrumentation and live feedback.

Tim closes with some questions for Sun:

  1. How do you help developers in this new style of dev to harness collective int.

  2. How do you help them create “live software”

  3. How do your thinking about OS and open standards in the context of data lock in?

Tim Bray – I have a problem with the terminology. It's not “content” It's my life. This is a symptom of an attitude problem in Web 2.0.

Tim O'Reilly – But users do generate that data. These are changes that are profound. Lots of us don't like the names, but there is something going on. If your system doesn't get better the more people who use it, you have a problem. This is really powerful.

Rich Green – There is a gradient to this which we feel less confident in defining. What about the phone? When we're on the phone are we chatting? Or are we generating content?

TO – But lets think about how would a phone company work if they were Google? And what if Google were the phone company? Tim elaborates on this in his blog post "What would Google Do?"

RG – Is the difference the storage of the data?

TO – So how do you make it easier for authors to generate their own mashups? How will Sun do that. Todays it's the experts who do it. It needs to get easier.

TB – I spoke to some bankers in Canary Wharf in London last week. He went in with a Web 2.0 message. As Bill Joy said, the smartest person is always somewhere else. He is not in the room with you. You (the big banks) need to join this info revolution. This is a phenomenon that is important for business, not just for the users out there.

Ian Murdock – As Jonathan Schwarz said a year ago – computers are not the commodity, it's the computing.

TO – Johathan was right. But Sun offered it and no one took it. Amazon offered it for 0.0x cents and they took it. The margins are getting tight in this area.

IM – Sun needs to go after the developer to get in that way.

TB – There is another effect. The degree to which you need to be a rocket scientist to make a contribution has come down but you still need to reduce the friction and the barrier to entry.

TO – I totally agree. Now you can contribute without being a programmer.

RG – Access is a problem. This is a big priority for Sun.

Audience questions:

Q: “Is there a limit to what Sun will open source?”

RG – Everything (edge cases aside) Open sourcing our IP under licenses conducive to sharing and community is good. More eyeballs = good. This is our business model. We are working through getting everything out there.

Q: “What is web 3.0 likely to be?”

TO - To me this is a meaningless term. Most transformative changes will be when we stop typing. This is not just speech interfaces (Google are offering 411 services to get better at speech, not to get into telephony) Gestural interfaces will be big (Wii), and lots of instrumentation (e.g. Insurance companies billing on your for your insurance based on the info of your car's location and speed.) Our computers will start to learn from us not from just what we tell them explicitly.

TB – It is very hard to road map the web side. I don't know what it's going to be but I know it will be surprising?.

Q: “What is the future at Sun for Java and other languages on the JVM?”

RG – People associate the JVM and java as the same thing. They aren't. JRuby has been big for the past year. There are 100+ languages on the JVM. We are adding a lot more. Pioneering steps are to be announced to bring far more indviduals into the fray. This is a very active play for us right now.

TB – Languages in their native form are not be blown off too (e.g. Ruby and Rails.) What runs on which VM is not that important. Loving the developer is the big thing

IM – As developer platforms of choice are moving up the stack, are the Operating Systems still as important? This week we are going to talk about “why the OS matters” The layer of abstraction is great until you need to see what goes on underneath.

TB – I would impress on us to see what we can learn from Wall Street. The first clicks are becoming the battle ground. Google and Yahoo are trying to give you your results inline in the results page. You don;t need to even go to the actual linked site to get your info. The massive databases will fight for who gets the user. The game is far from over but it is the game.

Q: “Flash and Heroic Attempts in Ajax vs Java?”

RG - Come back tomorrow...

No comments: