I still think JSON was a bad idea, that we should have worked harder, all of us, to make a single simple text-based data sharing language. And, if you ever doubted my sincerity, you can believe it now because I now have done the work to make my programming environment able to both send and receive JSON.

A couple of weeks ago, as part of the FeedHose project, I made it so that it could transmit JSON.

And over the last few days I've written a JSON parser.

For those people who are familiar with Frontier (the OPML Editor is just a distribution of Frontier, which was GPL'd in 2005), the JSON support is designed to work with the internals that have already been built around XML. There are two main routines for that, xml.compile and xml.decompile. There are two new routines, xml.jsonToTable and xml.tableToJson, that work exactly as the equivalent XML routines. They produce the same structures, so that all the techniques we have for walking structures work whether the data came to us through JSON or XML.

I've put up a web app for testing the JSON support in the OPML Editor. If you want to help, find bits of valid JSON that it doesn't correctly deal with. And find invalid JSON that it accepts.

http://json.scripting.com/

 10/23/2010; 1:36:03 PM.   .

Steve Ballmer is being widely quoted as saying that the next version of Windows will be the riskiest bet Microsoft has ever made. That's a pretty strong statement from a company that has made some big bets over the years.

My first reaction is "Sure, yeah whatever you say Steve" (with my eyes rolling). But then I thought about it, and surprised myself with the idea that yes, there is a big bet that they could make, and it could pay out big. The only questions are: 1. Does Microsoft see it? (Seems doubtful.) 2. Even if they do, can they execute? (Seems more likely.)

First let's start with a few assumptions.

1. It's 2012.

2. We love storing our data in the cloud because it's always available no matter where we are, no matter what computer we're using, whether it's a mobile device, laptop, desktop, tablet, or whatever. What's making it possible for all these different form factors to co-exist is the cloud, and the fact that we no longer have to synch data between devices, they all can access the only copy of all the data, the one in the cloud.

3. As computer professionals we know it's not wise to put all our data on someone else's servers. Perhaps users don't know that, but by 2012 perhaps some will. All it takes is one major outage on Google's servers, or Apple's, or Amazon's -- to spread the word that even though cloud computing is wonderful, and has huge advantages, it also has pitfalls.

Okay so that's the future context. Now, imagine you're running Microsoft. What are your strengths and assets? What makes you unique?

1. The OEM business. Apple always goes it alone, Google is just getting started with OEMs. Microsoft has OEMing in its DNA, for decades.

2. A familiar user interface. Whether you like it or not, there are a billion people who, more or less, know how to use Windows.

3. Lots of apps. Not as big a strength as it once was, but still substantial. And with Apple closing its channel more and more, Microsoft has the possibility to zig to their zag, and make their channel even more open to entrepreneurship. And they aren't starting from anywhere near zero.

4. They already have a version of Windows you can install on a server running in the cloud on Amazon EC2 or Rackspace. So they have a strong start at Windows as a cloud-based OS.

Okay, now what is Windows 8?

It's less than Windows 7. The UI is simpler. The media players are out. Its job is just to let your data have a place to live that you own, that you pay for, that you fully control. I would actually cut a direct deal with Amazon and create the ultimate cloud server that a user can manage themselves, that you pay for by the month. Need more services? Pay a little more. Need even more? Pay even more.

Now this would have to be promoted very crisply. With lots of influential individual developers non-disclosed and on-board long before it's announced. But because the basic softrware already exists, on Rackspace and EC2, you can start the bootstrap right away.

You also produce a simple, beautiful client that runs on netbook hardware, and runs on every other platform you can get software on. Go ahead and develop iPhone and iPad apps. Let Apple reject them, and when they do, make it public issue. Hell, make it a public policy issue. (BTW, I would partner with Facebook on this. Let Joe Hewitt run the project. Don't know who he is? Find out.)

Bottom-line: a really simple, user-administered, cloud OS is the next killer.

Now for the analysis...

Microsoft has gotten so boring. And it's not enough to say you're taking a big risk with Windows, you have to actually inspire people with a vision. It's not the risk that's important, it's the inspiration. Think about the commercial Apple ran when they re-introduced the company in 1997.

Now, back to the beginning -- do I think this is likely what they're doing? I don't. I think in all likelihood Ballmer has seen demos that impress him, but they are the old dogs and ponies they've shown many times before. Guys like Ballmer go for eye candy, just like most VCs and a lot of reporters. But they don't make good company-scale bets. Why do I think this is what they're doing? Well, because I don't think Ballmer has tech in his blood. He's a salesman. And one more thing -- Ray Ozzie just left. If they were doing something smart and bold, I doubt Ray would have gone.

If they went for it, would it be easy? Yes, actually it would, and that's the problem. Microsoft wouldn't need most of the OS engineers they employ. So you can bet that no matter what the user experience is, they will be rewriting major portions of the OS in this new release. Because that's what their engineers know how to do.

 10/23/2010; 11:54:22 AM.   .