I delivered a presentation to the Innsbrook .NET User Group this evening regarding Silverlight 2. I've attached the PowerPoint slides here.

Silverlight Version 2 - Hazzard - IDNUG - March 2008.pptx (237.20 kb)

During the presentation, I got dozens of questions which I truly enjoyed. It makes the presentation so much more fluid when the audience is fully engaged in the topic and feels comfortable in asking hard questions. I found one of the questions that was asked tonight particularly provocative. I'll paraphrase it as, "How do you decide when you're ready to use Silverlight for a development effort?" In the arguably justifiable excitement around Silverlight technologies (or Flex/Air/Laszlo for that matter), what should a business consider when making this kind of move?

During my presentation, I listed three key elements for making a decision like this. They were:

• Readiness of your users to make the move. Part of this decision is in understanding the cost of adoption. In the case of Silverlight, the barriers are low but there are barriers to consider. Some portion of your user base is going to be unable to install the required plug-ins due to incompatibility, inability by the user or even fear. If you are in a corporate environment, you may have more control of the incompatibility and inability problems. In the larger Internet community, there will always be a percentage of users who cannot overcome the hurdles required to implement the upgrade no matter how simple we believe it to be. Can you afford to write off those users?
  
• Ability to deliver something to your users that they cannot get (easily) using another combination of existing technologies. This is what makes Silverlight so interesting in my mind. I can build web applications in a lot of ways. But driving C#, JavaScript, IronPython and XAML to the desktop with Silverlight is compelling because it's the stuff I already know. Being able to push that core set of competencies so close to the user, all sorts of possibilities open up. In the case of this decision point, the question is, "Is the richness I can achieve by pushing my skills and tools out of the server and into the user's world so unique and valuable that I should do it?"
  
• Shifting resources in a way that allows the business to focus more money or time in areas that need it. There is no doubt that Silverlight will shift some of the importance quotient ascribed to ASP.NET into the browser over time. As that shift occurs, other technologies in the application stack will probably move upwards. For example, the SOA strategy your company uses today may be focused on providing so-called middle-tier services or even back-office services. Many companies put services on the edge of their networks for B2B transactions. But few do that for B2C applications. Arguably, AJAX callbacks fit this mold to a certain extent. But companies rarely focus on AJAX callbacks with the design rigor that's applied to middle-tier services. AJAX implementations are often seen as point solutions related to specific GUI requirements. So the "services" that support them get the same treatment. But what happens when much of the rendering load shifts to the desktop and inside the browser? That second layer of services that we're accustomed to applying contract-first, transaction, reliability and security principles will move up in the stack to fill the void. It's no wonder that Silverlight 2 has support for JSON, REST, POX, RSS and SOAP service calls built right in. Microsoft knows that this wave of change is coming. So, back to the original question. When the presentation engine moves out of the core of the network, what advantages will it offer your company, if any? What skills will you need in that kind of environment that are different than the ones you use today? Will it force (or allow) you to change your hiring patterns? How will your security model be affected? Will you buy more servers or fewer of them? What kinds of capabilities will those servers have that differ from the ones you buy today? How will your geographic distribution model for content be affected? With the reduced rendering constraints, will you be able to virtualize your applications to a greater extent? The list of questions goes on and on for this one.

If you can answer 2 out of 3 of these topics in an affirmative way (meaning: it makes sense to move forward), then I think you are ready to try building a rich Internet client using Silverlight. My advice is to start small. Pick bits of content that can easily be delivered the old way and the new way from one code base. Do a bit of browser detection work on your servers and deliver some Silverlight content optionally at first. In other words, try not to require the automatic installation of the Silverlight plug-in if possible. Perhaps you should even allow the users to opt in to such a program. Measure the uptake rate on the new capability before and after adding the installation requirement. Once you get a feel for the impact on the users and your own infrastructure/staff, you'll have what you need to make sound decisions. At SnagAJob.com, this is exactly what we're doing in our product development processes.