I’ve been thinking a lot about ways to organise weblog posts, and I think I may have something.
In the search for workable ways to organise chunks of content on a weblog, I looked at how I organise things in the real world, to see if there was any existing system that could be adapted. I turned to my CD collection, which sits in racks that cover one wall of our flat. The CDs are organised alphabetically by artist, starting at the top of the far left column. There are several things that can be observed about the system I employ: firstly that the alphabetical sorting is quite informal—artists are grouped by the first letter of their name, but I’m casual about exact alphabetic sorting. I don’t worry about Radiohead coming before REM or Kraftwerk before Lenny Kravitz. Relatively simple name-grouping is sufficient for me to find what I’m looking for, without increasing the overhead of returning the CDs to the wall or adding new purchases. Secondly, there are spaces in the racks between each alphabetic group, again to allow me to add new CDs without having to reorganise the entire collection. Thirdly, I notice that, within the alphabetic groups, I tend to place albums by the same artist together and, in some cases, organise these albums in order of release.
I’ve observed that this system is about guiding the eye. It operates on the basis that I’ll see a CD on the wall and then have some idea of where the one I’m looking for is in relation to the one I’ve seen. Say I’m looking for Placebo, and the first album to catch my eye is by Leftfield, I’ll instinctively move to the right until I see something like Portishead, and then look around there.
A key difference between this system and the system used in record stores is that there is nothing to guide the eye to the right album other than its position amongst other albums. To find that Placebo album in the store, first I’ll find a section called something like Rock And Pop, then a subsection called something like L To Q, then for an index card that says P, then another that says Placebo. Only then will I have to examine the cover of any particular album. There are no such visual aids in my collection, largely because my collection is considerably smaller, and also because of the overhead in creating and maintaining such a system. But it should still be possible for someone other than me to find a CD in my collection without me having to explain or otherwise annotate the system.
This idea of referencing things by the things that surround them is interesting. It’s made me change my approach to the weblog posts problem. Rather than forcing things into a structure, I’m thinking more about how to form relationships between posts so that you can find what you want from what you’ve got.
Individual posts on a weblog already have a relationship with each other. The fact that all posts are dated means that it’s possible to take any two posts and observe that one was created before the other. Of course, this doesn’t tell us a great deal: even though we can see the order in which the posts were created, there is no binding context. These two posts could be about anything, and not necessarily about the same subject as each other, so their relationship in time is fairly meaningless.
If posts have been arranged into categories, then some context will have been provided. You look at two posts within the same category, not only can you see the time relationship but also the subject relationship between them. I could file this post, for example, in a category called Weblogs, together with my previous post on the subject, and this would show the natural progression between the two. But herein we find the problem with post categorisation: the categories have to be broad in order to be useable, but this means that any two posts within a category may not be so closely related. within the example Weblogs category, I could have written about weblog technology, such as we have been discussing, but also a jumble of other things: blogmeets; the Guardian’s competition; the fact that some weblog has redesigned or shut down; the release of a new content management system, or whatever. A post about a forthcoming blogmeet may be posted after this one and placed in the same category, but wouldn’t have any real connection with this one: it wouldn’t be an amendment, follow-up or addition to this article, so not really related by context. Simply placing Metallica and S Club in the Rock And Pop section of the record store doesn’t make the albums more similar.
So, rather than defining categories, I thought about doing something clever with search technology. Let the computer read the posts and form connections between the ones with common words. But then, this post now contains the word “blogmeet”, and therefore the computer may decide to connect it to others that are about blogmeets. So then I thought about methods to create metadata for each post so that an automated search may get better results, but the overhead of doing this for the entire archive is massive.
It wasn’t until I’d stopped thinking about it for a while and had gone off to look at some things on the web that an idea hit me. I was either looking at comments on someone else’s weblog, and it occured to me that weblog comments display stronger relationships than the posts themselves. We have a post from the weblog author, and then subsequent comments from others both on the same subject and ordered by date. The original post provides the context for the comments that follow, and in some cases earlier comments provide the context for later ones. Comments don’t behave like weblog posts: their behaviour and structure is closer to that of posts on a bulletin board. They arrange themselves into threads, and their context is dictated by previous content, rather than by categorisation.
These relationships mean that, if you start with any particular comment, you can easily find the post that went before it as well other comments before and after, and if you start with the post you can observe the subsequent discussion on that subject. This is exactly the relationship I’m looking to recreate between weblog posts.