Twitter Lists: Categories and Conversations


Anyone who’s spent any time with me in the past week or two already knows that I’m not a huge fan of Twitter lists, but I’ve come to realize that I’ve (once again) left many people with the impression that I’m more anti- on the topic that I actually am. In this post I’ll cover what I like about lists, what I don’t like, and also toss a little bonus experiment at the end.

As a starting point, I think that Twitter lists are a pretty decent feature that’s getting way too much attention because Twitter just got a big fat check (and because Twitter hasn’t actually added an obviously user facing feature in a good long while).

I also think that it’s very Twitter-like in that it’s unclear what lists are really for, and that the most interesting parts of the feature will likely come from what’s built on top by Twitter’s users.

That said, we move on to the good…

For me as an individual Twitter user, lists look like the feature that I’ve wanted for a year or more. Once client app support is well baked in, lists will be a relatively simple way to group and filter my tweetstream in place.

I’m not a promiscuous follower, but with even a couple of hundred people twittering away at me it’s easy to miss stuff from the core group of people that I really want to follow closely. With lists I can create a view for the 20 people that I want to keep close tabs on and just flip to that filtered view when I want to. [Yes, I know, Tweetdeck sort of does this, but I really don’t like the narrow multiple column experience.]

For some organizations (The New York Times jumped right on board, for example) lists are a nice Twitter-native way to organize and showcase the contributions that their members are making on Twitter. The lists feature pulls data that had lived on the Web (which NYT staffers are on Twitter?) and moves it into Twitter itself. Handy.

And finally for Twitter itself, there are a couple of clear benefits.

For one, the suggested user list and the hullabaloo that surrounded it can finally be put to rest. Twitter can feature a random user-created list every day of the year, and if people are pissed about being included or left out, they can take it up with the listmaker, with Twitter’s compliments.

For another, Twitter now has many, many people basically tagging one another, in Twitter’s database, with useful little snippets of information. If Twitter wanted to, say, be able to segment out its users based on their interests for some mysterious reason, those lists can serve as neat little topical tags.

And the bad, which isn’t really so bad…

Now by “bad,” I mostly mean “weak,” which is my one word answer to the question “but what about Twitter lists as a way to discover people who tweet about topics that interest you?” My longer answer is that I don’t believe it’ll work for discovery long term, for a variety of reasons.

First issue: for the most part, people are making big lists.

I don’t know whether it’s because they’re concerned about offending people by omission, taking a completist approach to covering all aspects of a topic, or something else entirely, but add just a couple of these lists and you’re suddenly following 150 more people and enjoying Twitter overload. This list making behavior could change, but it looks like a real issue for adoption at the moment. [Update: I was wrong about list subscriptions appearing in the main stream, so this point can be chalked off as poor observation on my part.]

Second issue: list rot.

We’re all excited about making lists now, but how much time are we planning on dedicating to maintaining those lists? Part of Twitter’s appeal is that it’s a low effort, low friction tool, and I’ve got to wonder whether the current batch of lists will reflect the interesting new people who join Twitter (or those who stop using the service) three months from now. Again, this isn’t an inherent problem with lists, but it’s a real concern with regards to usage.

Third issue: once you’re following a list, there’s no way to tell which tweets in your timeline are associated with a given list—who is this person, and why are you seeing their tweets, again?

You start following a music list and suddenly you’re seeing this @schlarb dude twittering about his breakfast. To find out why you’re following him you’ve got to click through to his profile, check which lists he belongs to, and then try to recall which of those lists you subscribe to. Yet again, this isn’t an insurmountable problem, but it’s a UX flaw that has a meaningful effect on the experience of Twitter lists right now. [Also: note that Chris Schlarb is an incredible musician and an interesting guy, so you probably should follow him anyway.]

This third issue also points to the central concern that I have about lists as a discovery tool: the way that Twitter lists are most commonly used right now encourages us to sort people into single topic buckets on Twitter, when people’s tweets are as eclectic as the people themselves.

Take everybody’s favorite example, the venture capitalists. I follow a fair number of VCs on Twitter, and with a couple of exceptions I’d say that their tweets are about the business of venture capital and the areas/companies they invest in maybe half the time.

What’s the other half? Well, they twitter about their spouses and kids, the restaurant they’re at, music, the funny thing that they just saw…just stuff. It’s almost like they’re just regular human beings who have diverse interests and enjoy sharing those interests.

The realization that follows is that when one starts following a topical list, one can expect to get something like 50% on-topic signal and 50% off-topic noise. Sub-optimal experience, right there.

So what else can we do with lists, then?

Here’s one thought: what if rather than using Twitter lists to group people by category, we use them to group people by conversation?

If the goal is to find people who have interesting things to say about music, or food, or venture capital, why not start from a person who’s Twittering interesting stuff on that topic and link them to the people they’re talking to and talking about on Twitter? Seems like a decent way to find other people with good stuff to say on that topic, no?

Enter conversationalists. The idea occurred to me on Friday afternoon: the Twitter API allows you to create and modify lists, so why not have a dynamic, automatically updated “conversational list” of the twenty twitter users that you’ve @mentioned most recently? It makes explicit the community that you’re paying attention to, and offers people an easy way to add context for the list creator’s tweets by seeing what the creator is reacting to…it makes the experience a little deeper and richer.

And now, thanks to the awesome skills of Kevin Marshall, conversationalists have gone from a vague idea in my head on last Friday afternoon to working code on this Friday morning. And note that this timeline includes the four days that he apparently spent dead drunk in New Orleans, so it’s a pretty damn impressive accomplishment on his part.


Go, read a little more about it, create your own conversationalist, follow the conversationalists created by others. I’m curious to see how this idea plays out.

I don’t think that this approach is necessarily “better” than topical groupings, but it seems like an approach that factors in the very human feel of Twitter, and it’s therefore worth a few bits and a few hours to explore.

A number of people asked me about what I meant by being invested in “being wrong in interesting ways” after my blog post a couple of weeks ago, and this is a prime example. I may not be right about this being a worthwhile use of Twitter lists, but I’m certain that if I am wrong then we’ll all find some interesting stuff in the process: I’ll have been wrong in an interesting way.

  • Better than being interesting in the wrong way :-)

    Thanks for all the thoughts… all very interesting to me!

  • Great explanation…oh and for the record, it was 5 days in New Orleans (Sat. – Wed.) and I was at least reasonably sober while watching the Falcons lose…the rest of the time I plead the 5th.

  • Okay, your occasional sobriety is officially on the record, as is yet
    another “awesome work on the quick!” from me. :-)

  • Great idea. Why only daily though? Seem like you want the new users in your stream ASAP.

  • No particular reasoning behind daily updates for the lists, just an arbitrary unit. In the grand scheme of things, I think you're right: the shorter the refresh interval, the better…we could start calling it “constant context” or something. :-)

    It sounds like the idea started some wheels turning in your head, though, so I'm looking forward to seeing what you come up with.

  • Nice! I love the idea of dynamically generated lists. It would be cool if the tool weren't specific to @ replies, but was more general… sort of like Smart Playlists for iTunes, where you could specify what criteria defines the list (@ replies, favorites, posts containing links, etc). Here's a comment I made on Fred's blog a few weeks ago about the concept…

  • Excellent, glad you like it!

    Kevin and I have talked through a couple of possible extensions. The
    idea that Charlie O'Donnell mentioned in the comment you were replying
    to is the most obvious: we decided to go with a straight “most recent”
    for replies initially, but having some slots set aside for “most
    frequently replied to” is pretty straightforward and makes a lot of

    You're dead on that it should be reasonable to allow person X to
    create a list of the people that person Y is replying to, since the
    tweetstream is public anyway. It's less efficient in theory than just
    having person X create the list themselves, but people don't always do
    what you want them to. :-)

    Kevin actually built something that may fit what you want for the link-
    focused tweets: Does some nice stuff like
    calling out links that have been tweeted by more than one of the
    people that you follow, too.

    The more customizable stuff (views filtered by number of times
    favorited, topics, what have you) I think is really interesting, but
    I'm still wrapping my head around some of the simpler possibilities
    right now. :-)

  • Great stuff Whit!

    (and Kevin too)

  • Thanks! We'll be rolling out a few refinements and variations over the
    next few weeks, I'm looking forward to see what shapes this sort of
    thing can take…