Open Source and Web 2.0
Pages: 1, 2
Working for a Buck an Hour
O'Reilly next asked Schwartz about Sun's new model of renting out CPU time at a rate of one dollar per CPU hour. Schwartz explained that the spectrum runs to the big banks that need 10,000 CPUs to run simulations. Traditionally, this has required long and complicated contracts. Initially, Sun thought this was going to be the bulk of the market. They did recently get a rush of demand from oil and gas companies in Texas needing to perform computationally intensive calculations after Hurricane Rita.
Schwartz thinks the idea of paying for computing service makes sense. He thinks this will be another business in which the long tail pays off. There are a lot of people who have short-term needs and a limited amount of money. There might be a researcher with a small grant who needs access to a grid of computers to help with his or her research. Having access to a grid is an attractive solution.
From Sun's point of view, comparing services to selling hardware, "the economic opportunities are much more attractive in the next generation of the business than in what we've seen to date." Mitchell Baker said there are a lot of cost savings that can be realized with this model. "The cost of control structures is very high with the 500-page contracts and the centralized infrastructure and needing to control it. We haven't really explored what the savings are with self-service at the infrastructure level."
Baker pointed out that it is sometimes hard to see the extent of the potential savings from our current perspective. She said that in the future, "It may well be that we look back at our current model of proprietary software with each copy of the license where we make whole new laws so shrink-wrap licenses are enforceable; we worry about piracy; and we build a giant infrastructure based around a center trying to control. I think as that begins to change to customer self-service, people being able to do more things, we'll see how prohibitive and what a drain on the current system the cost of enforcing control is."
Product and Distribution
In response to a question on the recent Sun-Google announcement, Jonathan Schwartz reiterated his belief that "The power of open source is not code. Very few of us know how to modify it." He noted that Java, Google, Flash, and many of the other popular frameworks and applications do not have source that is open. He argued that the power is in the community and in the distribution of the product. He said that the benefit to Sun of partnering with Google is to be able to take advantage of their distribution.
Mitchell Baker pointed out that "a well-distributed lousy product is not enough." She said, "The code or the product, and the user experience, and the quality of what you are actually distributing matters." Baker reminded the audience that we sometimes forget that "the average consumer does not know the difference between the browser, the internet, and the search box."
O'Reilly steered the conversation back to "a destructive period in the '90s when user experience was damaged by a pissing match between Netscape and Microsoft." He asked if we are facing a return to that and if this might be a roadblock to all of the good things that are starting to happen with Web 2.0.
Baker responded that there were two problems in the '90s. First, there was the Netscape and Microsoft fight over new features, and second, there was the demise of Netscape. This was followed by a long stagnation in Netscape's browser with no new features. She added that it is "easier for vendors to develop for one mediocre product. It is not easier for users."
O'Reilly observed that Microsoft is a great company that is best when it has strong competition and that now that it is being threatened on multiple fronts, he expects them to do their best work. He said that there are two evils: "hypercompetition is destructive to the users and you can have no competition."
Baker replied that the Mozilla Foundation is cognizant of hypercompetition. She said that they spend a fair amount of time thinking about what will happen if they add a new feature or a new capability. What will be the experience of an Internet Explorer user who doesn't have access to this feature? Content and websites won't adopt the new Mozilla features unless there is "graceful degradation."
On the other hand, she explained in answer to a question from the audience, Mozilla won't just add a feature because Microsoft has. She said that they received a lot of complaints when they didn't follow the ActiveX model, even though it has turned out to be a good decision. They recognize that a significant portion of the market has IE and that users benefit from some level of feature compatibility.
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 O'Reilly Network