I make the full text of my blog available via the rdf & xml feeds. This is great for readers like me who like to read all their news in a newsreader and minimize the clicking. But it's bad for writers like me who like to know how much their stuff is being read, as the act of reading RSS doesn't increment any of the blog's counters.
Is there a silent majority, or even significant minority, of RSS-readers out there? In an effort to find out, some time early next week, I think I'm going to attempt to redirect, at least temporarily, the existing rss feeds to the ones provided by feedburner. This will give me some statistics and should be seamless for everyone, but it's possible that some newsreaders will detect the change and be unahppy.
Feedburner is “pre-alpha” software, so I won't do this for long, at least for now. If you are a reader who relies on either the rdf or the xml feed and find that this change causes you any grief in the next few days, drop me a note and I'll discontinue the experiment. (Or, if you think this is a lousy idea, please say so in comments, and I'll reconsider.)
Will this approach count the people who read your feeds through a web based aggregator like bloglines? Bloglines seems to download the feed once every hour to its own database and then serves up its downloaded copy to those who have this blog among their subscribed feeds. So you’ll see bloglines downloading the feed 24 times a day and not see the thousands (*) who read the feed that it downloads …
* OK, bloglines lists only 11 subscribers to discourse.net who have chosen to make their subscription a matter of public record; but _surely_ there are thousands (!) of more secretive subscribers whom your count will miss 🙁
Good point. The web based aggregators risk both over and undercount, depending on how frequently they grab the feed, and how many people read them. So there’s just bad data and different bad data….
Michael,
I saw your post here and wanted to comment. First, we’ve test many of the newsreaders with all of the various flavors of our burned feeds, and have only had issues with a couple of the very infrequently used readers (both of which also had problems with unburned feeds). The issue raised by anonymous reader is absolutely a good one, HOWEVER, a few of the aggregators, such as my.yahoo, and now I believe bloglines, are beginning to report the number of subscribers to a feed in their http requests…this IN THEORY will enable us to give more accurate counts. Additionally, for other aggregators that poll quite frequently (and may potentially poll from multiple IP addresses), we are not counting “If-modified-since” requests as “new” subscribers even if the IP address is new (on the assumption that it’s a client or aggregator that has previously requested the feed and is checking for updates).
Finally, there are a couple of rogue newsreaders out there that do NOT correctly make if-modified-since requests and those can skew the results upward a bit as well, but we’re aware of which readers these are, and will work to have them change their behaviors as we move out of pre-alpha and into a more stable environment. Please let us know how you like the service, and what additional features you’d like to see. We have had an overwhelmingly popular first couple of days, so we’re very excited about the possibilities. We’ll undoubtedly have a couple bumps in the road here as we’ve just launched, but we’re working furiously to see how the service is used and how we can improve it! Feel free to email me directly with any questions or requests!
sorry, i should additionally point out that we are not CURRENTLY counting the number of subscribers given by the few web aggregators that report these (in their http headers) in your stats. Look for the stats pages to change demonstrably as we move from pre-alpha to alpha in the near term.