As always seems to happen, people looked at the list which already had the problem of being grossly skewed towards men, and noticed other odd little things. Like, “hey, I follow this science person who has more followers than so-and-so, why isn’t she on the list?” Clearly, there were errors of omission, and oddly, they all seem to skew away from women and people of color. Or maybe not so oddly — it’s striking how often that seems to happen.
And then people started thinking about the premise. I recall from my phone conversation with the author that I’d told her that follower count was a poor measure of influence — I mentioned a few names to her that I thought used Twitter really well, but that I had no idea what their follower count was. It turns out that @bug_gwen and @docfreeride, for instance, don’t have enough followers to have made the Top 50 list.
Wait a minute — so why are we using follower count as a metric? I don’t pick who to follow by looking at the number of followers they have! If I did, I’d be following @JustinBieber, who has 54.7 million followers. There is a kind of bandwagon effect, in that you’re more likely to encounter someone on Twitter if they have lots of followers, because it is more likely that you’ll be following someone who follows the popular gang…but that’s also a poor reason to consider someone a good person to follow. There are a great many people who have huge follower counts because they have a fanbase in another medium — books, for instance, or television shows — and are absolutely abysmally bad at Twitter, and I don’t need to name names, since you’ve all probably got a few in mind already. (there are people who are good at multiple media: @NeilTyson deserves to have a lot of followers for his good use of Twitter, and I’ve found @NathanFillion to be hilarious and enthusiastic).
But, unfortunately, what it means is that Science picked an easy metric to measure, rather than a difficult one that might tell you something significant, and it was also a metric that amplified biases in the pool. We already know that people are less likely to pay attention to women’s voices, so when you use a parameter that is dependent on how many voices listen to you, you’re already screwing that part of the population. And then I suspect the filter goes to work: you start searching for popular twitter accounts by looking at who the most popular twitter accounts follow.
I do not blame the author. She was trying to track down a quantifiable measure and used the ones at hand, and was also trying to address a specific contrivance, the claim that high Twitter follower counts was somehow indicative of scientific failure. She didn’t invent the Kardashian index, so don’t blame her: blame Neil Hall, who came up with the K-index in the first place.
And now, of course, we get a useful backlash. People have started compiling lists of active Twitter users who also happen to be scientists and women. Here’s one from Paige Brown Jarreau; one by Erica Check; another by Victoria Herridge. I also posted a list of women scientists on youtube a while back. It seems these are relatively easy to find. Instead of referring to arbitrary lists of people assembled by an arbitrary metric that has a built-in bias against certain kinds of people, you’ll find that there are other lists built by advocates to counter those biases.
I am reminded of the time that atheists conference organizers were complaining that they couldn’t find any women speakers, so they kept mining the same old small group of old boys for their rosters, so Jen McCreight compiled a list of awesome female atheists to make it easier for them…and to kill that excuse. And then they complained that they could only find white atheists, so Greta Christina put together a list of atheists of color for them.
Isn’t it weird how invisible people suddenly become apparent if you just look for them?
It also has a snowball effect: You follow one, you read their conversations, you follow the people they’re having conversations with (and sometimes you end up in a full circle with them having conversations with people you were following already).
And you get broader different perspectives and you learn a lot.
Also, another little thing: as I was looking through those Twitter lists, I discovered that the statement in your Twitter profile is really important in helping others figure out whether they want to follow you. Follower count…not so much.
I added quite a few because they sounded interesting and seemed to share some of my interests.
The biggest problem about using follower numbers as a metric is that a large number of the users who follow people are spambots. The more followers you have, the larger percentage of followers are spammers, it’s kind of exponential, but not exactly.
And there’s still the problem of having to wade through crap to get to the good stuff. Until Twitter allows reasonable, easy-to-use, and effective methods to curate one’s feed, the signal-to-noise ratio isn’t sufficient to draw me onto the medium. Got too much of that non-confrontational estrogen vibe, I guess.
As to invisible people suddenly becoming visible when you look – why, it’s almost as if they were there all along!
To me, it seemed so obvious that the methodology (starting with celebrity scientists on Twitter, people those celebrity scientists follow on Twitter, people in lists of “scientists to follow online”) was going to reproduce the biases that give rise to things like patterns of citation that overrepresent the work of men and underrepresent that of women that I’m frankly baffled (1) they didn’t try to come up with a better methodology, or (2) at least acknowledge up front this likely explanation for the dearth of women (and non-white scientists) on the list.
It’s a weird inverse of that thing scientists sometimes do of trying to take their explanatory strategies way too far from the domains where they work. Instead, we have a reluctance to notice that the biases and patterns of social interactions that create a particular effect in one situation could maybe be causing a similar effect in a situation that is right next door.
Or maybe it’s just that people don’t really believe the empirical evidence and they are sure, dammit, that their evaluations are objective.
When you’re at the top of the privilege pyramid, it’s amazingly easy to not see people who don’t look exactly like you who are also at the top of the privilege pyramid.
FUCKING AUTOPLAY ADS! VERIZON, I AM NEVER BUYING YOUR PRODUCTS!!
And yes, I’ve reported them through the system. This new ad provider sucks balls. They are not following FtB’s policies. Egregiously and consistently.
/thread derail.
Sorry, but I do blame the author of the article. There was a chance to do a thoughtful analysis and really understand the subject, but nope, that was too damned much work.
Journalism–including scientific journalism–in this country is dead. Maybe the Daily Show would consider taking on science journalism–they certainly couldn’t fuck it up any more.
A bit of a missed opportunity for Science. I guess what they were going for was a quick review of scientists having an impact on Twitter. But they should have either (i) delved deeper to find out who was having the most impact/persuasive influence, or (ii) written about the follower metric but explore what that meant rather than simply taking it as the gold-standard measure of importance.
How about
number_of_followers * number_of_tweets?Or the number of retweets? The number of tweets @ them?
sounds like the article was meant to draw in readers easily because the subject was not really who were the influential scientist but twitter. Was it a “high class ” puff piece, a list for hufpo, a missed opportunity . The subject raises many question that would be revealing with further expansion but not alas quick or short.
Is it an example of the trouble journalism and publishing are facing in this transitional period of electronic communications?
uncle frogy