Home » Uncategorized » Filtering Twitter like Email

Filtering Twitter like Email


Recent discussions on the AoIR list, kicked off by Bernie Hogan — on mundane use of Twitter have gotten me thinking about other kinds of Twitter use I don’t like, and what to do about them.

All sorts of applications, on the web and soon on mobile devices, are starting to realize that Twitter doubles nicely as a free broadcast channel for their in-application status messages. If you, for example, “Just secured a safe house in Berlin, Germany. #spymaster” then all your Twitter friends know about it, whether they are playing Spymaster or not. I don’t want to unfollow someone who I like and find interesting, but I only want the interesting parts, not necessarily their gaming activities.

Quasi-biomedical devices are starting to Twitter too — they are just novelties at this point, but there’s a potential here for turning Twitter into a passive stream of medical data that you may in the long run prefer to be private. Examples so far include heart rate monitoring, baby kicking (h/t @askpang) and my favorite, publicly quitting smoking.

But these applications using Twitter changes Twitter from a conversational medium to the ultimate activity stream. And unfortunately, any app who’d like to can pollute your stream.

Bandwidth in the Twitter stream can be seen as a Commons [1]. It’s a resource that we all share, and each of our tweets consumes a little bit of it. The individually rational choice is to use up as much of the bandwidth as we’d like — promote your web game, your mobile application, your blog, whatever. But if every application pollutes the Twitter stream, then pretty soon it’s a cesspool that’s not of use to anybody. The collectively rational choice then is to conserve the bandwidth, but good luck getting anyone to do that, because it’s counter to their own individual motivations.

This sounds an awful lot like email or usenet. You’d like an uncluttered inbox and uninterrupted conversation, but spammers — free riders in consumption of your bandwidth — want to use up some of your bandwidth.

In these situations, we use filters. ISPs and users deploy spam filters. Mailing lists get routed into their own folders. Some people get flagged as “high importance.” Filters are the internet’s version of remediation when the stream gets polluted.

If Twitter (the company) doesn’t want Twitter (the service) to collapse under its own weight by treating all streams equally, then something has to be done.

Twitter seems content so far being a monolithic, inclusive source for tweets of all kind, conversational, application, or otherwise, and letting all the filtering happen downstream in third-party applications. Any why not? It’s worked well for them and a huge number of apps now populate the Twitter ecosystem.

So far I’m not completely pleased with my options. Application creators sometimes sort of allow filtering in a basic way, e.g. by person, but I want filters that match the power of email filters. Something like, “All tweets with some hashtag #xyz go into folder XYZ unless the tweet is from @myfriend, in which case it goes to my inbox.” This way I can choose which tweets to read now, to save for later, or to ignore completely if I like.

Why doesn’t Thunderbird support Twitter? It supports email, RSS and usenet, so why not Twitter? The structure’s pretty similar. Think of a tweet as an email with a sender and a subject, but no recipient or body. Is anyone interested in building this?


[1] Of course, Twitter is not exactly a commons. It’s not an open protocol like email or usenet (NNTP). It is interesting to note that Twitter, like Facebook, are in principle government-like (and FB explicitly thinks of itself this way), but so far Twitter has been as hands-off as Facebook has been hands-on regarding control of cultural expression.

[2] Update: Fred said nearly the same thing a year ago!



  1. Fred says:

    Amen. My thought on the matter are here. I’ve sort of given up on Twitter implementing anything like this, but I still hold out hope that a third-party client might take the initiative.

  2. Fred says:

    Twitter is a very interesting problem when you think of it as a channel – and for those of us with any following volume, the problems are very much in line with those identified in early information theory (Shannon, etc.) In the updated equation I suppose our time would replace bandwidth capacity. The most interesting problem is how to not miss important tweets. Not those important in a global sense (i.e. if I miss one tweet about how Michael Jackson died I’ll probably see another), but those important situationally – like a friend who only tweets once in a while checking in. I want to follow these high volume people but I don’t want to miss my close friends. My best idea was “rooms” or something like that. But should we come at the solution through HCI, or should we be going back to Shannon and thinking about the core equations and how to apply them in this case.

    Either way, interesting stuff, and glad to see we’re thinking along the same lines. I propose that NERD solves the problem!

  3. […] Continue reading here: Filtering Twitter like Email « Scott Golder […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: