An Exercise in Aggregation

A few weeks ago I collected a ton of posts regarding the Iraqi elections. I did this for a few reasons. The elections were important and I wanted to know how they were going, but I could have just read up on them if that was the only reason. The real reason I made that post was to participate in and observe information aggregation and correlation in real time.

It was an interesting experience, and I learned a few things which should help in future exercises. Some of these are in my control to fix, some will depend on the further advance of technology.
  • Format - It seems to me that simply posting a buttload of links in a long list is not the best way to aggregate and correlate data. It does provide a useful service, it being a central place with links to diverse articles, but it would be much better if the posts were separated into smaller groups. This would better facilitate scanning and would allow those interested to focus on things that interest them. It would also be helpful to indicate threads of debate between different bloggers. For example, it seems that a ton of people responded to Juan Cole's comments, though I only listed one or two (and I did so in a way that wasn't exactly efficient).
  • Categorization - One thing that is frustrating about such an exercise is that many blogs are posting up a storm on the subject throughout the day, which means that someone like myself who is attempting to aggregate posts would have to continually check the blog throughout the day as well. Indeed, simply collecting all the links and posting them can be a challenge. What I ended up doing was linking to a few specific posts and then just including a general link to the blog with the instruction to "Keep scrolling." Dean Esmay demonstrated how bloggers can help aggregation by providing a category page where all of his Iraqi election posts were collected (and each individual post had an index of posts as well). This made things a lot easier for me, as I didn't have to collect a large number of links. All I had to do is post one link. Unfortunately this is somewhat rare, and given the tools we have to use, it is also understandable. Most people are concerned with getting their voice out there, and don't want to spend the time devising a categorization scheme. Movable Type 3.x has subcategories, which could help with this, but it takes time to figure this stuff out. Hopefully this is something that will improve in time as more enhancements are made to blogging software.
  • Trackbacks - Put simply, they suck for an exercise like this. For those who don't know, trackbacks are a way of notifying other websites that you're linking to them (and a way of indicating that other websites have linked to you). Movable type has a nifty feature that will automatically detect a trackback-enabled blog when you link to it, and set the site to be pinged. This is awesome when you're linking to a single post or even a handful of posts. However, when I was compiling the links for my Iraqi election post, I naturally had tons of trackbacks to send. I started getting trackback failures that weren't really failures. And because I was continually updating that post with new data, I ended up sending duplicate pings to the same few blogs (some got as many as five or six extraneous pings). I suppose I could have turned off the auto-detection feature and manually pinged the sites I wanted for that post, but that is hardly convenient.
  • Other notes - There has to be a better way to collect permalinks and generate a list than simply copying and pasting. I'm sure there are some bookmarklets or browser features that could prove helpful, though this would require a little research and a little tweaking to be useful.
Writing that post proved to be a most interesting exercise in aggregation, and I look forward to incorporating some of the lessons learned above in future posts...