O'Reilly Media founder and president Tim O'Reilly opened up OSCON 2004 with a look at what is currently on his radar. He began with an abbreviated version of a talk he has been giving lately on Internet applications. He then turned his attention to social software.
Tim began by asking what it means that Google, eBay, and Amazon.com are the new killer apps. He updated a quote from his 1997 speech Hardware, Software, and Infoware that "Free and open source software is the Intel inside of the next generation of software applications" by adding "--or is it?" He explored the implications of Clayton Christensen's law of conservation of attractive profits, the value simply migrates to adjacent levels.
To put this in historical context, Tim said that as building computer hardware became a commodity business with lower margins, the value went up the stack to proprietary software. At the same time, Intel showed that there was an opportunity to push value down the stack as long as you build a critical component that the next layer depends on. So the value was pushed out of the middle layer of building computers to the adjacent layers of software such as the Windows operating systems, Office, and to hardware components such as Intel chips.
Looking at that dynamic, people imagined that with the rise of Linux and open source we would replace Windows with Linux and Office with an open source solution. This was the dream of having inexpensive commodity hardware coupled with free software. For now, they reasoned, open source would leave the Intel layer alone because it is a hard project that has many technical requirements.
O'Reilly argues that instead of a stack that is capped by Linux and Open Office, we have another stack to consider with LAMP and similar open source software components occupying the middle. The value has once again moved to adjacent layers. Looking up the stack, eBay, Google, Amazon.com, and Mapquest are all building proprietary software on top of open source foundations.
Several of these applications have found a way to get users to create data that is used on their sites, which leads to lock-in by network effects and not by API. There are more than ten million user reviews on Amazon. That's not really software but the added value here is the data. Tim asked, once there's a critical mass of buyers and sellers why go somewhere else? In this grouping, the "Intel inside" position is occupied by Network Solutions, which controls domain names. All of the mapping applications are built on top of Navteq.
The platform is the Internet and not the PC. These applications are built on top of open source but are not themselves open source. Tim says that's OK because they have built tremendous value. More importantly, if we want to move open source forward, we have to understand that the whole model of what constitutes open source doesn't work. For example, you could give away the Google code and still not be able to implement Google. If we're thinking of openness we have to ask what openness means in that context: a world where an app runs on 100K servers and Richard Stallman cannot run it on his personal machine.
As you create your web-based applications, ask how you might build a participatory level around the data in the same way that eBay and Amazon.com have done. Tim left this topic asking who is going to control the key namespaces and who will integrate the entire open source stack. He suggests we think beyond Linux and ask who is going to be the Dell of open source and make sure that evrything works well together.
Tim next turned his attention to social software and asked how many people in the audience had tried Orkut. Most of the audience raised their hands. He followed by asking how many people kept using Orkut and very few left their hand in the air. He said that all the social software services are a hack because we haven't really reinvented the address book.
Tim showed screen shots from a Microsoft Research project that could answer questions such as who you communicate with around this particular topic. The question that follows is how we build tools for creating networks and managing our contacts. These tools could end up as part of Outlook and proprietary software, or they could become a connection between Orkut and GMail. "We have to Napsterize the address book and the calendar so that we own the data about our social network but we are able to query our friends about who they know."
There are benefits of thinking of software, as Dave Stutz suggested, above the level of a single device. Tim offered the iPod and iTunes for lessons for open source. This is the first application that has done seamless application from server to handheld device. The handheld is not just a bad copy of the web or PC interface. When you build an application, consider what the changes may be when you assume that it runs from a handheld all the way to the server.
Tim concluded his remarks with the announcement that O'Reilly Media will be producing the third annual MySQL Conference later this year in Santa Clara, California
Daniel H. Steinberg is the editor for the new series of Mac Developer titles for the Pragmatic Programmers. He writes feature articles for Apple's ADC web site and is a regular contributor to Mac Devcenter. He has presented at Apple's Worldwide Developer Conference, MacWorld, MacHack and other Mac developer conferences.
Return to the OSCON 2004 Coverage Page
Copyright © 2009 O'Reilly Media, Inc.