We Need a Better Algorithm

In the late Nineties, Yahoo was the Internet. Technically it was a human-compiled directory of web sites, but it was essentially the home page of what was then a very new and exciting space. When I was looking around for a new job after business school, I discovered Salesforce on Yahoo.

Here’s how it worked: If you had a web site, you submitted it to Yahoo, and if someone there thought it was worth including, they categorized it accordingly. The company hired librarians and book store employees to do this job. It was hailed as the most significant effort at organizing knowledge since Carolus Linnaeus invented the modern system of taxonomy.

Then, of course, came Google. Google didn’t use people, it used math — or rather, it used math to formalize the collective intelligence of people. You typed in a subject, and its PageRank algorithm recommended a web site to you based on the number and quality of links to that web site. It was a brilliant algorithm.

Google eventually left Yahoo in the dust. It grew enormously popular because it was simple and it worked. There was no public outcry about Google’s algorithm. No one accused it of making the world a worse place. There was no documentary decrying its evils, a “Social Dilemma” for its time.

“Google didn’t use people, it used math — or rather, it used math to formalize the collective intelligence of people. ”

But then things took a turn for the worse. Social networks exploded in popularity, which was fine enough, but their algorithms had a very different goal: engagement. Suddenly the algorithmic priorities shifted away from authority and validation, towards hype and virality.

Now, that was fine when people used Facebook to connect with old high school friends, or used Twitter to post about what they had for lunch. But it became extremely problematic when these platforms transformed into the primary source of news and information for millions of people.

“Suddenly the algorithmic priorities shifted away from authority and validation, towards hype and virality. ”

Which brings us to the grim situation we find ourselves in today. In a recent Wall Street Journal article titled “Social Media Algorithms Rule the World. Good Luck Trying to Stop Them,” Joanna Stern writes that:

“People are shown things that appeal most to them, they click, they read, they watch, they fall into rabbit holes that reinforce their thoughts and ideas…They end up in their own personalized version of reality.”

So what can we do about this? I have a few ideas.

First, insert authority back into the algorithms. Stern suggests one option is that “the platforms get serious about deprioritizing the outrage, anger and conspiracy, and prioritizing the trustworthy, thoughtful and reputable—even if they know it means less engagement.”

Your feed can’t just be simply what your network is sharing. Amassing a large following doesn’t make you by default an expert on anything. In the early days of Google, people were constantly trying to game the system by purchasing dozens of links to their websites; the algorithm eventually figured it out.

Second, create more accountability. If the social media companies are indeed becoming a main source of “what’s going on in the world,” there needs to be consequences for spreading false information. We’re finally starting to see some of this happen now, but there needs to be way more investment in content moderation.

Third, fix the business model. Google works because its business model demands that user searches return relevant information. Again, this gives them an incentive to fight against people trying to game the system.

The social media companies have no similar incentive. If anything, they make more money when people game the system by driving more engagement. That’s the problem with a business model that runs slot machine psychology and variable rewards.

Social media will never be free of tribalism and anxiety. But getting rid of a business model that actually depends on tribalism and anxiety seems like a really good place to start.

Prioritize authority. Create more accountability. Fix the business model. Then maybe social networks can go back to doing what they do best: connecting people.

“Social media will never be free of tribalism and anxiety. But getting rid of a business model that actually depends on tribalism and anxiety seems like a really good place to start. ”

We Need a Better Algorithm

“Google didn’t use people, it used math — or rather, it used math to formalize the collective intelligence of people. ”

“Suddenly the algorithmic priorities shifted away from authority and validation, towards hype and virality. ”

“Social media will never be free of tribalism and anxiety. But getting rid of a business model that actually depends on tribalism and anxiety seems like a really good place to start. ”

Fresh subscription stories delivered to your inbox, weekly.