Windows Live gadget architecture explained

At the Microsoft TechEd conference in Auckland recently, I got a chance to sit down with George Moore - GM for the Windows Live Developer Platform. He is of course based in Redmond, but was over in NZ for TechEd.

Here is an edited transcript of the interview, which covers the Windows Live gadget architecture in detail and also explores how to develop Windows Live gadgets using technologies like Atlas.

RICHARD: This morning you gave some very impressive numbers about Windows Live. They indicate that the Live platform has already gained a significant user base (much of it from previous MSN-branded services). A lot of the Live products are still in beta though (e.g. Mail, Shopping, live.com itself). So where would you say Windows Live is at currently in terms of product rollouts - will things start to come out of beta when Vista is launched, for example?

GEORGE: Well actually you're starting to see a number of the websites losing the beta designation and the trend looks alright. I can't say the specific date when everything's going to be out of beta, but certainly it's going to be this year - ahead of Vista.

RICHARD: Gadgets, or mini applications, are becoming increasingly important on the Web - they can be added to personalized start pages (like live.com), to social networks (MySpace is a popular platform for mini-apps), and even in normal blogs/webpages. You also showed one today that was integrated with Windows Live Mail...

GEORGE: The Live Mail example was to illustrate the gadget architecture. "The basic kernal pieces are identical and used and reused across the Windows Live services" In fact Windows Live Mail is built on the same gadget architecture, as is Spaces. So the basic kernal pieces are identical and used and reused across the Windows Live services. The important point is not that we're asking developers to go out and build gadgets because we think it's a good idea. We're self-hosting them ourselves - actually building our own services out of that same architecture. We're dog-fooding those pieces and we're achieving fairly massive amounts of scale using the same architecture.

RICHARD: Perhaps the most exiting part about gadgets is when future technologies start hosting gadgets - Xbox, mobile phones, television/Media Center. Can you tell us a bit more about the larger, future, vision for gadgets for MS?

GEORGE: It really depends upon the capabilities of the browser. I mean obviously you're not going to run gadgets as they exist today with Ajax inside. In mobile alone that's going to lack a significant Javascript interpreter and the xhtml request pieces and all those components. So in many cases it's dependent more on the capabilities in the underlying browser. In the case of Windows Media Center, you're running IE at that point - so there's no reason why you couldn't run a gadget in that case anyway. It'd be a ten foot UI experience as opposed to a two foot experience. But architecturally there's nothing different because you're just running on top of IE.

RICHARD: So that's part of your vision, to have gadgets running on all sorts of devices - not just the PC...

GEORGE: Well certainly having the user have access to their content from any number of different devices, is certainly part of the vision. We've been talking about that for almost a year now. The technological means of how that data is rendered will be dependent on the underlying capabilities of the device and where the device is predominantly used - and ten foot vs two foot UIs in mobiles and intermittedly connected vs always connected... that sort of thing. And that goes hand in hand anyway with the Live Anywhere vision around the gaming - and all those pieces as well. 

RICHARD: Is a lot of that outside Microsoft's control - in that it needs other parts of the infrastructure to come together. For example with mobile phones, the platform isn't quite ready yet.

GEORGE: For traditional mobile phones, yes. I mean you have fairly anemic browsers in those mobile phones today, but for Windows Mobile phones increasingly you'll see more capabilities being added into the browser. But that's just the normal evolution of things, that devices get smarter, more transisters, making use of the transisters, that sort of thing. 

RICHARD: The Vista Sidebar is an interesting development. As Vista hasn't yet been released, most people probably don't fully understand how Windows Live and gadgets will interact with Vista and the sidebar. I wonder if you could walk us through the high level of how that Web/desktop integration will work once Vista is released? 

GEORGE: The gadget architecture can actually "The gadget architecture can actually render to any number of different technologies. It can render to DHTML, to Avalon, etc." render to any number of different technologies. It can render to DHTML, to Avalon, etc. So it would be up to the gadget author to detect if they're running on a Vista machine, if they choose. They could just write a standard straight-up vanilla gadget that renders to dhtml - and it would work on live.com and Vista sidebar. However they could detect the fact it's running on Vista and say: I'm on a more capable machine, I want to do something fancy with Avalon, some translucent GUI effects or whatever it is - or I'd like to be smart about being intermittedly connected... so it's really up to the imagination of the gadget author as to how far they want to push the boundaries, of leveraging the underlying hardware on the platform.

RICHARD: So the sidebar is quite a key element of vista - it seems like it's being positioned as a crucial intermediary between the Web and the desktop. Is that correct?

GEORGE: Well it's a handy docking space for people to be able to have ready access to bite-sized bits of information. That's the whole vision around the gadgets anyway. 

RICHARD: What are the most innovative Windows Live products in your opinion currently? I know it's early days, but are there one or two products that you think point to the future and are really innovative in nature. E.g. you mentioned Messenger Bots this morning, which are like automated agents. That seems pretty innovative to me, but I wanted to get your pointers on other interesting Live products/apps worth keeping an eye on. 

GEORGE: Well certainly the increased capabilities of running gadgets in Spaces, as Spaces continues to evolve and grow in that sense. It's interesting to think about Messenger and being able to de-compose parts of the Messenger experience. So things like the chat window itself is logically separate from presence, which is logically separate from the contacts, and things like that. So I think you're starting to see a generalised trend that this componentization metaphor is essentially de-composition of things like Messenger into its component hostable parts. 

The first example of this is the Contacts gadget, where we're essentially taking out the contacts information on Messenger and enabling it to be hosted standalone on a site. So when you say innovative features, you really have to differentiate between the finished services that we provide - Mail, Messenger and Spaces - and what capabilities we can provide for APIs inside vs the infrastructure pieces. The underlying pieces around why the Contacts gadget, for example, runs independent of Messenger and doesn't require the Messenger client to be installed on the user's machine. 

We want to provide this "We want to provide this a la carte, you choose whatever APIs you want and push together - that's just how Windows itself bootstrapped." a la carte, you choose whatever APIs you want and push together - that's just how Windows itself bootstrapped. Because we can never predict [what people want] in various geographies worldwide - given users tastes, demographics. And the inventiveness of developers worldwide is astounding. We can't predict this, we don't want to be in the business of writing every single application. You will see us writing [apps/services], just as in the case of Office (which is a very horizontally available, broad set of applications). We do the same thing for Messenger and Mail, but there's lots of opportunities to do all sorts of innovative things using the base infrastructure.

RICHARD: Darryl Burling (a Developer Evangelist for Microsoft NZ) mentioned today at TechEd about being able to create "your own gadget-enabled website" using Atlas - which will enables developers to create rich interactive web pages. Can you explain a bit more about this? What kind of websites do you envisage people making?

GEORGE: If you go to atlas.asp.net, you can download the Atlas SDK. It works if your site is hosting ASP.Net, and there's a PHP version as well, and you just basically install this on your site and then users who visit your site essentially download the necessary client-side components - in order to do effective rendering of whatever you scripted/wrote.

So there's technically and legally - "...technically the bits are on atlas.asp.net for you to download." technically the bits are on atlas.asp.net for you to download. Legally there is a Go Live license that lets you  legally start to build businesses upon that if you so choose. Even ahead of a formal release, at which time it will become a supported Microsoft product.

RICHARD: Finally, is the Windows Live rollout closely aligned to the Vista rollout?

GEORGE: Logically they're disconnected. If you followed the re-org mails from a couple of months ago, there is alignment now in the management team. Steve Sinofsky's group is responsible for both the Windows Vista user experience piece as well as the Windows Live user experience piece - so there's unification from the organisational perspective, but not from a scheduling perspective.

Summary

To conclude the interview, George said they are very interested in feedback regarding things like the Contacts gadget - they routinely read the forums and blogs for feedback. He also noted that they can't predict every scenario that matters to people - so if you have ideas or specific scenarios for the Windows Live platform, you should contact the Windows Live team and let them know.