Service oriented architecture is gaining traction. Proponents claim that it promotes reuse and innovation through recombination. To understand this, one should read the concept of modularity. One book that I would highly recommend is Design Rules by Baldwin and Clark. They define modularity as a design structure in which tasks and parameters are interdependent within modules and independent across them. Their most interesting contribution (IMHO) is how the process of modularization occurs through design rules. Design rules clarify the architecture of the system, describe the interface and procedures for integration and testing of modules.

When we apply the above to SOA, we have to first create modules that are of the right granularity. This is a difficult task. We should then look for a shared set of design rules. While there are several module providers and listing agents, what is missing is the rest of the design rules. Until we have those, we will not get the results -- reuse and recombination. In the case of mashups, platform providers like Google and Yahoo! bake the design rules into the platform itself. This way, module providers should focus on creating the right set of modules and allow third parties to innovate on the platform. The platform provider gets to create and manage the design rules.

programmableweb.com (my data source for mashups) has made major changes to the presentation of the mashup matrix. The data source has also improved in quality and access. The current network has 449 nodes. Since this clutters the picture, I have rendered nodes having more than three mashups. As always, the color coding is as folows:
Microsoft -- Sepia
Yahoo! -- Light magenta
Google -- Blue
eBay -- Melon
Government -- Gray
un affiliated -- Red

To keep a platform viable it is important to have a vibrant ecosystem with complementors releasing a variety of products that work on the platform. With iGoogle (is Apple OK with this? :-)), Desktop and Gadgets, Google allows users to access that are developed by third parties. According to this site, Google has over 4800 gadgets available to users. In order to keep this network of complementors growing, Google has started Google Gadget Ventures to financially support development of popular gadgets receiving over 250,000 page views a week (that makes 9% of them eligible for funding). This funding should act as both a quality control and growth mechanism for Google's platform. Google's product APIs support the ecosystem by providing toolkits and forums to developers. www.programmableweb.com has recently started tracking this and other platform vendors separately.

A recent article in the WSJ confirmed some of the stories I have been hearing on outsourcing to India. The first generation of outsourcing was based on cost arbitrage and sending mundane back office work out to India. This worked well for the most part because the coordination requirements weren't that high and the presence of cheap and competent engineering talent in India. Some cracks are beginning to show now. As the article suggests, the nature of work that is being sent out have become more complex and require a lot of coordination by the outsourcer. In addition, the talent that is available is not ready to deploy without some training. While this can be done, the question is at what price? For some of the more value added and complex work hiring, training and retaining engineering talent has become very difficult. With turnovers reaching 25% in some cases projects are at peril. In addition to this, salaries are growing at a 30% rate over the past few years and bringing total compensation packages at par to the rates in Silicon Valley. This poses a clear threat to Indian vendors. At the lower end of the work spectrum there are cheaper alternatives (Vietnam, Philipines,..) developing and at the higher end the cost arbitrage doesn't work out.

As a result, at least two things can happen. One: outsourcers can learn to share knowledge and coordinate better by using some of the emerging technologies. Two: Indian vendors can move into the higher value added work and become part of the innovation network and not just a low cost partner. I'm planning to explore this through case studies this summer.