I have a deliberate strategy that guides who I follow on Twitter, and the tools and techniques I use to curate my feed. I’m not sure how unusual it is; perhaps many of us do the same types of things but don’t share our thinking with each other. I wrote this post in case my thinking is useful to others.
My goal on Twitter is to broaden my horizons and help me to see the world through other peoples’ eyes. I explicitly do not use Twitter to learn about what’s happening in the world, stay in touch with friends and family, share photos, or discover articles or other content. I value control over what I consume, and I value quality over quantity.
To achieve that goal, I balance carefully whom I follow and how, to create a high signal-to-noise ratio in my feed. This article is a set of principles and opinions I’ve developed over time, some fundamental and some in response to circumstances that arise.
First, my choices related to whom I follow and how:
- Follow thoughtful people. I unfollow people who share content without taking a moment to ask if it’s real, intelligent, kind, and verifiable. I’ll give some examples in a later article of the type of sharing that makes me unfollow people. A random example of a person whose sharing has a superb signal-to-noise ratio is Manisha Agarwal.
- Follow no more than 100 people. I set an arbitrary limit of 100 after experimenting to see how much content people share. This forces me to prioritize. For example, it meant that I unfollowed some smart and worthwhile white people to make room for people of color.
- Following isn’t friendship. I don’t feel obligated to follow my friends. My priority is the content that people share, and the quality of content is only loosely correlated to whether I know someone. I want to listen outside my bubble, and if I follow my friends they’ll drown out the other voices. Many of my dearest friends aren’t on my following list. Choosing to use Twitter as a window to the world, and to disregard any sense of obligation to send social signals with it, is liberating.
- Don’t follow popular people. Popular people have a lot of followers. I’ll hear their best work secondhand.
- Don’t follow prolific tweeters. I’ve learned a lot from people such as @jasonlk and @girlziplocked, but I don’t need to follow them to do it. Again, their followers will retweet their best. Extremely prolific tweeters can be interesting to follow for a while, but when I start to feel a sense of exhaustion, I just unfollow them.
- Unfollow people who tell me what I already know. I don’t need my biases and beliefs reinforced by others. I do that well enough on my own. This inherently means that Twitter isn’t “comfort food” for me to read. It means I’m curating an experience that’s at least slightly disconcerting, which is good.
- Turn off retweets sometimes. Sometimes I find a person whose perspective is interesting, but their interest in something like a celebrity, or their support of their employer, isn’t. I can get the best of their original thoughts by disabling retweets and just looking at what they write themselves. I’d say I’ve disabled retweets from a third or more of the people I follow.
- Follow more women. I live in a patriarchy. Men hold most of the positions of power and influence. A lot of what they think and say comes from a place of bias and privilege and isn’t nearly as interesting, true, or valid as what women tweet. Listening to what women say has changed my perspectives significantly. If powerful, privileged men say worthwhile things, women are a great filter to pass it along to me through retweets. But if I listen to what men think about themselves, I hear delusions of grandeur and righteousness.
- Follow more minorities and nontraditional identities. Following people who adhere to different religions, people of color, trans people, and people from different worlds (not people in tech who live in America, for example) is great. I still get the bias that I’m following people who, like me, have access to Twitter. But there’s no way for me to escape all biases.
- Follow mostly individual accounts. I follow a few non-person accounts to introduce variety; for example, I follow @tiny_star_field to sprinkle little pictures into my feed and break up the monotony of text. But I don’t follow many non-person accounts that rate dogs or tweet out emergency kitten pictures. I follow enough people with cute pets already.
- Follow some people for specific topics. I follow a limited number of people who specialize in topics I’d like a little bit of.
I’ve seen various counterpoints to these principles as time has passed. For example, someone tweeted me to tell me I could follow my friends but mute them. That way, presumably, I wouldn’t feel guilty or risk offending my friends by not following them. I disagree; first of all, where does one draw the line at who’s enough of a friend to follow? There’s no right answer if following is about friendship and trying to make someone feel good about themselves, so I choose not to do that. Secondly, if I mute them, I won’t see any of their tweets, which isn’t what I want. It’s better to unfollow and rely on mutual friends to retweet their best to me. Again: I believe following is much better used as a curation and filtering mechanism. Trying to intersect that with a social signal is a losing game.
Another person agreed with my “follow more women” advice, and said he tries to follow roughly the same number of men as women. But he misunderstood me. I don’t try to do that. Women don’t get an equal voice in our society. I need to follow far more women than men to equalize their voices to the level they should be at. Check who I follow and see. If you’re having trouble figuring out how to follow more women who deserve to be heard, perhaps start with a list such as this and go from there.
Here’s an example of why putting into action many of the points I’ve made thus far is working well for me. Someone once tweeted “I wish following was a spectrum, so I could follow @pmarca at 25% and get only his top-quartile tweets.” Well, here’s some unsolicited advice: if you unfollow him and follow thoughtful people who will retweet his best, you’ll get that. But in this specific example, because @pmarca is actually not a nice person at all, what you’ll get instead is an education on why a lot of what he tweets is narcissism and hatred. You’ll get your perspective changed. That’s what you should want, instead of the top 25% of his most skillfully-worded bigotry. Forgive me for being opinionated, I guess?
It’s not all about who I follow: it’s also about how. There’s a number of technological and other factors at play (such as political and business interests), and I need to negotiate these with care. Here’s how I do that:
- Use Twitter. Twitter is inherently more open and egalitarian than other social media networks, which don’t encourage third-party client software that might present a different perspective than the one they think they can make the most money from. It’s definitely the best (lesser of the evils?) of the major networks.
- Don’t use Twitter’s client. Twitter’s algorithmic choices are not for me. I want to choose my own timeline, not let algorithms do it for me. Algorithms are genuinely dangerous to human society, as we’ve seen over the last several years. I believe it is my individual responsibility to curate my timeline carefully. I cannot surrender this to a machine. Third-party clients ignore Twitter’s algorithm and present the full, unmanipulated feed of content. Or at least I think they do. They also don’t show Twitter’s ads. Sometimes I wonder how long Twitter will permit this. I hope they continue to.
- Use a sophisticated client with filtering. Clients such as Tweetbot (on Mac and iOS) or Flamingo (on Android) let you do things like muting a particular client by viewing the tweet’s details and clicking or tapping on the client software that sent the tweet: I’ve muted a bunch of clients like Untappd, SumAll, instagram, LinkedIn, nuzzel, and the dozens of client applications that companies use to automatically blast content onto social media. This significantly improves my signal-to-noise ratio. You can also mute keywords. I find it useful to mute political words, for example; I get my political news elsewhere. Here’s some of my keyword mute filters in Tweetbot: The first keyword is a regular expression that mutes tweets with too many hashtags; the others mute things I find are noisy or stressful. Muting political hot-button content keeps some of the shouting out of my timeline.
This sounds like a lot of work, doesn’t it! It is, but it’s my responsibility. And it works: I no longer lack control over what I read. (Or at least I have the illusion of control). Put another way: I completely empathize with this tweet from Asya:
Dear @Twitter I follow more than 700 people. That means I want to see what they share in my timeline and not what some people I don’t follow (but they happen to follow) share. If what those people share is so cool, my friends will retweet it and I’ll see it then. Thx.
The solution I found to that problem is purely technical, because the problem is purely technical. If I couldn’t use Tweetbot or something equivalent, I’d just stop reading Twitter. (I’d probably continue writing on it, though.)
Why go to all this trouble myself, instead of letting Twitter’s algorithms do it for me? Isn’t that what Big Data and Machine Learning is for? Ostensibly, yes, but as we’ve seen, absolutely no. There’s a deeper point here: I either take responsibility for my own consumption and learning, or I abdicate it to machines. And I believe that abdication to machines is amongst the most pressing dangers facing our society today. We have a President that was elected by foreign powers who figured out how to manipulate millions of people through the algorithms that they allowed to control what they see, hear, and believe.
Those algorithms are built and influenced by the inexorable demands of the company’s business model. I think the company is mostly helpless to understand, let alone overcome, the vulnerabilities to which their business model exposes them. If this doesn’t scare you, maybe it should.
My solution to this is purely non-technical: I built my own algorithm, so to speak, by following people with sound judgment, and a variety of experiences and worldviews. It’s old-fashioned, but I think it’s better than the new-fangled technologically generated news feeds that dominate the Internet today.