Ethan Miller

Unless otherwise noted, all work licensed under : Creative Commons License

A Stream or a Reservoir

Mar 6 2009, 1:19PM

Here's a question that keep coming up lately: In the context of data flows, what should I treat like a stream, and what should I treat like a reservoir?

To explain what I mean by example: my email inbox is a reservoir. Although data streams in, it's clearly something that's supposed to be fully processed. It doesn't seem correct to let email float by without reading, then deleting or archiving (although it certainly happens).

First confusing case: Delicious - I started my Delicious account in Sept of 2004. I have basically been using it as a reservoir until recently. When I started using it I adapted my mental model for 'bookmarks' in my browser. I basically worked under the assumption that I was creating a finite list of resources that I would prune when necessary, and return to as needed.

In fact, what has become clear is that I almost never prune my collection, and I only occasionally use it to look up a link that I wanted to remember. It's really a stream. Like Digg it's just a way to comment on a link, take note of it, share it, and let it go.

Next confusing case: Google Reader (really it's any feed reader, but Google Reader has some features that highlight my confusion). I have been using it as a reservoir - I recently realized the reservoir model is actually stronger for my feed reader than it is for my inbox, for some reason.

I think Google Reader provides some confusing signals in this area. It very prominently shows you the number of unread items - it does this in several places. That's clearly part of the reservoir model - for me at least, it creates a need to see a nice fat ZERO. At the same time, Google Reader makes trends available to you - which compare the number of items posted to the number of items you've read. This implies that it's expected that some items just float by unread.

Since I'm always loosing the battle to keep up with stuff I'm interested in, the issue comes up a lot. I'm trying to treat my feed reader like a stream. If I succeed, Google Reader will permanently show "1000+" for unread items - bothersome, but I prefer this to un-subscribing from great feeds, and completely missing those posts.

On the other hand, there are certain feeds I need to capture 100% of the time (for example, updates on software tools I use, or friend's feeds). I know I can solve this with folders - but it doesn't really resolve my question about how I really want to treat these data flows.

Tags : information


blog comments powered by Disqus