New users joining Twitter are presented with a list of recommended people to follow in categories like entertainment, government, gaming and politics. It's telling, then, to see what kinds of people Twitter has deemed influential enough to merit a spot on these lists, which often act as a gateway to hundreds of thousands of followers and widespread public recognition. Until today, if you were interested in "Technology & Science," you were likely given a list that was 100% male.
Twitter's imbalanced recommendation engine was first spotted by science and tech journalist Rose Eveleth on Wednesday. "Really Twitter, you can't pick a single woman to suggest?" she wrote.
Then, Verge science editor Elizabeth Lopatto tweeted Eveleth's message at Twitter CEO and co-founder Jack Dorsey — and actually got a response.
"Wow. Haven't looked at this in a long time," Dorsey tweeted. "Will investigate and fix. Thanks!" He @-mentioned Ed Ho, aka @mrdonut, Twitter's VP of engineering.
Apparently, the Twitter team got to work. By Thursday morning, the list reflected the changes.
How Twitter recommendations work
Twitter's recommended lists are curated using a combination of humans and algorithms. Due to the nature of these algorithms — they adapt over time — the lists update frequently and show slightly different results when you refresh them.
But as we have seen with online harassment tools, internet search results and facial recognition that rely predominantly on machine learning, algorithms are not free from human bias, and that is almost always to the detriment of women and people of color. Simply put, many algorithms can't learn to be more inclusive without human input. It's short-sighted to believe that social problems can be solved by machines alone.
Which is why, in this particular instance, human engineers at Twitter needed to update the algorithm to ensure that it was surfacing more diverse recommendation lists.
"We've adjusted our recommendations to include more female account suggestions," Twitter product manager Kathy Porto Chang tweeted to Eveleth on Wednesday evening. "We'll continue to make our algos more inclusive!"
Twitter couldn't disclose exactly how the algorithm determined who was worthy of making the list. While the predominantly white male list Eveleth spotted is certainly representative of the industry as a whole, there are plenty of exciting female voices in STEM who use Twitter. Eveleth herself pointed out a slew of them when Chang inquired.
We checked the technology and science lists on Thursday morning. Some lists are still predominantly white men, but there are a few more women sprinkled throughout. We also spotted more brands, foundations, news organizations and even @Twitter itself in the recommendations.
Why it matters
It's important for Twitter to ensure its algorithms are updated to create inclusive lists for new users joining the platform. Diverse perspectives lead to diverse ideas, pulling people out of the homogenous bubbles many of us are trying to escape on Facebook.
Secondly, the recommended lists afford leading voices on particular topics real public legitimacy — not just a massive audience, but heightened visibility and influence on a vocal platform.
Twitter's updated recommended lists aren't a "mission accomplished" — but the change does mark a concerted effort toward righting small wrongs and installing features that make people of all genders and backgrounds feel safe and enjoy their experience on the platform. Twitter is all too aware of what can happen when they don't.