Feedburner(http://feedburner.google.com) provides statistics about RSS feed of subscribers and reaches. This is interesting.
It is easy to understand that Feedburner can count visits (reaches) to a RSS feed. But, how does Feedburner get to know subscribers to a RSS feed.
In my understanding, each requests to RSS Feed URI is independent. There are no cookies or identity validation. So, how does feedburner know how many subscribers to a RSS feed?
The easy part is Google tell it the number of Google Readers, and so do the other Reader/Aggregators.
For individual users polling the RSS/Atom feed, there are http headers involved in the request, so users are tracked by IP address, and when behind proxies, a number of proxies include original IP in header, this helps sort between proxied sources.
Failing that you could read the FeedBurner help on that topic.
I wonder if it knows anything at all.
--- end sarcasm ---
Seriously, my sub numbers for my blog will jump from about 2k to about 3k at the drop of a hat.
Determining subscriber count is an inexact science at best.
It does rely on reporting from other services, and sometimes these services go down, or they change how they report.
Services like FeedBurner are actualy a proxy feed to your blog's feed. So when you use FeedBurner (or alike) users subscribe to a feed hosted on Google's servers that is fed from your feed.
Thusly people are really subscribing to the feed hosted by Google and they can then get statistics just as if you were visiting a site.