I started playing with AJAX before it got its silly name and true to X I used XML in my application. At least at first.

What I discovered is that flicker caused by processing data and simultaneously writing results to document can be completely removed if I do a Google and fetch javascript object instead (what later became JSON ) and use that. Difference was so vivid, I dropped XML almost completely.

Fast forward to autumn. There’s a debate after our regular weekly talk and a question comes up. Why would Google or anyone else for that matter choose to use something else than XML?

Relying on personal experience I offered speed as obvious and most important reason. I couldn’t back it up, but since I’ve seen it with my own eyes, I didn’t doubt it either. It wasn’t the sole reason. I find dabbling with javascript objects simply easier than handling XML and I’ll care about XSLT when javascript in Safari can access it.

I got reminded of our discussion when I read this article . Read it, if you haven’t yet, because it’s well worth it. Anyhow, I decided to make a new test. I wrote a simple test page , that fetches sample json file and basically equal xml sample and measures how much time it takes to fetch same data.

Result?

After 15 minutes of wild reloading in different browsers (Safari 1.3+, FF1.0, IE6) and different local and non-local servers, I can’t see a difference that couldn’t be attributed to latency. It seems browsers have moved on and there’s little reason to avoid using XML today.

So use whatever you like most.