I am a web developer that specializes in web-based applications. I have over 10 years of experience in this area. Here are my thoughts on some of the issues that are being batted around:
20 character limit, exporting, and sharing- The storage space excuse may be a bit weak because it's not a lot of data at first glance, but it's not just a matter of throwing a hard drive into the server. There's redundancy, backups, and database efficiency that must all be factored in as well. If the limit isn't just for business reasons, I'm confident it will be scaled when possible. I don't have the link, but I'm pretty sure I read that your character data would be retained and exportable for up to a year after you stop paying.
Google isn't a charity. Ads and marketing data pay for those "free" services. I'd imagine ads in character builder would create more nerdrage than what exists now.
Export has been bumped in priority. That's a good sign. Sounds like they have a process. But no matter what process you have, software development is still a technical art. All sorts of stuff can happen in the implementation phase that will throw your deadline off. I think this happened to the WotC developers and they were forced to go live with a mostly complete product and a list of post-launch priorities. Iplay4e-style character/campaign sharing is probably one of those features as well.
Silverlight vs. HTML- I really wish people would shut up about HTML5. It is not as magical as the hype sounds. The big features that get touted are fairly limited. You'll get the ability to play some video without a plugin and the
canvas element. The canvas is cool, but doesn't have anything like the capabilities of Flash or Silverlight. Other HTML5 improvements will make it easier to develop rich applications, but in the end it's just evolution and refinement of the current HTML spec.
Silverlight offers a mostly cross-platform environment that can more easily implement the complex UI that I've seen in the videos of the character builder. It's not impossible to do in HTML, but I wouldn't want to do it. Silverlight also uses all the .Net languages and constructs that the dev team are familiar with which is a big plus for a small department.
I think character builder is just the tip of an iceberg. What everyone isn't seeing is a more expansive and flexible back-end infrastructure. The part you interact with right now may be in Silverlight, but I'd bet money that it's calling on web services to do all the heavy lifting. Web services are platform-neutral, leaving a developer to write the UI and some glue code to get an application working. This makes iPhone, Android, HTML, etc. implementations practical. It could also make it possible for 3rd parties to utilize the services in more ways than we've seen with Iplay4e (The problem with MasterPlan was that it was making local copies of the data).
Cost- I think my six-ish dollars a month is well spent for what I'm getting now and will increase as time goes on. Just the other day I spent $4 on a coffee so I really don't understand why people think a subscription per person is so expensive. There will always be hard-luck cases, but in general most people that can buy the books in the first place can afford the subscription if they want it. Managing electronic transactions will only get easier as they become more widespread so I don't think the lack of credit card argument has much of any merit.
In the end, I think WotC is on the right track and my speculation will be proven to be accurate. There isn't a piece of software on this planet that doesn't have a list of bugs and a list of desired features. The problem is that eventually you have to ship something. We'll see how it all pans out tomorrow.