Tweets from the past, or: Filters and serendipity
In his new book «Where Good Ideas Come From», Steven B. Johnson at one point tries to defend the Web against its critics by asserting: «Filters reduce serendipity. [...] But beyond bookmarking, filters are a second-generation addition to the architecture of the web.» The critics he is referring to usually argue that the filters we employ to identify relevant information on the web make it more unlikely that we stumble upon something we didn't actively seek - serendipitous discovery.
It's true that most filters we currently have reduce serendipity. However, they don't necessarily need to. In fact, I believe that really good filters should actively introduce serendipity.
How would a serendipitous information filter look like? Just adding an element of randomness is not enough - there is simply too much information for this on the web.
A simple way to have a little serendipity (or at least what I understand as it) in your «information diet» is to revisit information - something the filters and systems we currently have are not designed for. I have written 177 posts on this Posterous; I have saved 1778 bookmarks on Delicious and posted well over 3000 tweets. There was a reason I posted all these information bits, but I have forgotten about most of them shortly after they were online. They were replaced by new information that came my way.
I have started to "revisit" at least part of that information. I know, the idea of reading through your blog post or even tweets from a year ago sounds somehow stupid. But I keep stumbling upon interesting stuff I have long forgotten; and some of it even has a relevance for what I'm doing or thinking about right now. Especially valuable is leafing through old notebooks. (Of course, this is also partly a vanity and reconstruction exercise: remembering the many things that one has done and seen).
New ideas usually emerge when we start making connections between existing pieces of information. But the current systems - or, most of them - make it difficult to add "historical depth" to this networking exercise.
I'd like to have information filters that actively help me with this kind of revisiting. I don't know how they could look like. But I'm sure someone will figure it out. Or, more likely, someone already has, and I just don't know about it.
