There’s a nice overview of the proposed APML standard over at masternewmedia.org. Though the article did very little to change my generally negative view of the standard I have to give them credit for creating the best “laymen” explanation that I’ve seen thus far.
For those who don’t know APML is a standard that attempts to capture and harness your attention data between sites. So in theory it should be able to see you searched for Kelly Clarkson music on Amazon and then use that to suggest Kelly Clarkson music to you when you open iTunes. This is something I’ll probably cover in depth at some point so I’m not going to go through a point by point analysis of what I think is wrong with the standard but I would like to quickly throw out my top 3 objections…
There’s no vendor advantage and a significant vendor penalty: When you get right down to it a standard like APML is going to have to rely on websites supporting it to work in any realistic way. This in turn means that sites will have to freely give this attention data away in order for it to work. The problem there is that attention data is the most valuable thing most sites have and they’re being asked to give it away with nothing in return.
The APML supporters seem to think they can get users to demand support from individual web sites but I simply don’t think that is going to happen. If you look at the current supporters you’ll see that the companies supporting this standard thus far are all companies with virtually nothing to lose (Feed reading companies which already allow export through opml and new companies whose business model focuses on APML)
Until they come up with a way to sell this to sites themselves I don’t see it ever getting off the ground.
It’s too simplistic: This is probably my biggest complaint in that finding a way to harness attention data is a fantastically complicated task and the APML standard is built around applying an extremely simplistic method of solving this problem. So if I’m looking at a book on .Net does that increment my interest rating for C# and VB.NET or is that a separate thing? If I’m interested in football will I then start getting soccer info from U.K. based sites? If I search for Eminem does that raise my interest in him personally, rap, hip hop, or all three? These issues go on and on (the one’s I brought up here are just the simple problems)
There’s a lot more to say here but the bottom line is that the idea of arbitrary categorization and equally arbitrary ranking makes APML an extremely flawed system.
No set build process: To a certain extent this point should be included in the above paragraph but this is something that is so clearly outlined that I thought I deserved its own treatment. To quote from the blog post linked to above…
Likewise, as for "private browsing activities", such as pornography, the services that implement APML can be set to ignore certain browsing habits.
“can” being the operative word there. A standard that can be implemented in several different ways isn’t really a standard. Imagine RSS feeds in which the feeds listed weren’t the most recent feeds but just feeds chosen at random. It would be useless. Same applies here.
I could go on and as I said before I’ve actually spent some time on this and someday plan to post my issues with the actual format itself (as opposed to the theoretical problems listed above) but that will be for another day. For now lets just say I admire what the APML folks are trying to do I just don’t think they are doing it in a way that will create any positive change.