NewsGator CTO Greg Reinacker writes:

"We’ve now implemented a persistent APML endpoint in our online platform. What this means is, if you’re using sync with NewsGator Online, there is a well-known URL that represents your APML attention data."

At first glance, this may strike you as something that only hard-core geeks would be interested in.  After all, it’s not as though tons of people are clamoring to get their attention data.  And by itself, your attention data doesn’t have much value since there’s not a helluva lot you can do with it yet.  But as more services support APML, the more valuable your attention data becomes. 

ReadWriteWeb claims that the keys to a killer web service are search, aggregation and conversation.  I’d add relevance to that list.  Relevance is key to bubbling up the stuff that’s important to you out of all the stuff you’re getting via search, aggregation and conversation.

Right now most services that provide relevance do so by learning over time.  They might, for example, notice that you often read articles about a specific topic, and then use that information to recommend new articles to you.  Or they might be able to recommend new "friends" to you based on the friends you already have. 

But you have to use a service long enough for it to learn about you.  Sites like Netflix and Amazon can’t recommend new stuff to you until they know about the stuff you’ve already bought.  I don’t know about you, but I don’t want to keep training all these web sites in the hopes that they’ll serve me better once they know more about me.  I’d rather have some sort of external profile I can give to these sites which tells them what I’m interested in.

That’s what APML is for.  Attention data via APML describes the things that are important to you, and once a service knows what’s important to you, it can do a better job surfacing relevant information without first having to spend so much time learning about you. 

Since we’re providing your attention data as a well-known URL, you can easily share it with other services.  Unfortunately, we’re still at the infancy stage here, so there aren’t many services that support APML yet.  My hope is that by being among the first to offer APML, we can convince more services of its value.