The Margins' 5 Point Plan to Fix Social Media

Demetricate. Chronologize. Slow down. Explain. $$$$.

Ranjan here. This week we’re talking ideas for social media regulation.

Senator Josh Hawley just unveiled the Social Media Addiction Reduction Technology Act (or cutely named "SMART Act"), which contained a plank that would limit users to 30 minutes a day on social media platforms. There were simple ways for users to circumvent this, but, needless to say, the bill was controversial.

I've been thinking on how to use regulation and/or legislation to improve social media consumption for a while. In the spirit of the SMART Act, and the Democratic debates, this is The Margins’ 5 point plan for fixing social media.

(Strong disclaimer: My co-host Can has issues with a few of these points, but has agreed to co-sponsor the legislation. Because that’s the spirit of compromise needed in this political era.)

#1 - Demetricate

This is the single most attainable plank of our plan. Remove all numbers from social platforms. The friend and follower counts, the Likes, the RT's, the views. They're all gone. A quick story on this:

It was during the run-up to the 2016 election that I really started feeling uncomfortable with Facebook. I learned about the Facebook demetricator, a Chrome extension that removes all numbers from the Facebook webview. Experiencing your feed without numbers is such a fundamentally different thing. You evaluate every bit of content, not in the context of external approval (how many Likes or the numerical influence of the creator), but simply in the act of its consumption.

The moment I knew just how insane the metrification of social platforms could be, was when my daughter was born. I posted the obligatory birth photo. I'd periodically check Facebook on my desktop, but I had the Demetricator installed, and it was weird to not see how many Likes the post got. I had long before deleted the Facebook iOS app, but I was curious (translation: thirsty) enough that I'd log into the mobile web version on my phone, just so to see the Likes racking up.

Now, I don't consider myself that influencer-y. I certainly enjoy creating things and the related external approval (especially newsletter views!). But that experience seared into me the drug-like power of the Like.

It's not just about you finding validation. It changes how you see other people and things. I still cringe when I hear things like "that person only has 1.2k followers", “that wasn’t really viral, it only got 200 RTs”, or “can you believe that video got 10 million views". Let's be clear, this is a new phenomenon. We didn't use talk about this stuff like this before and say, "The Full House finale had a 14.6 household rating and 25 percent audience share". The validation of media, and even the worth of people, via metrics is central to this entire challenge, and it’s the lowest hanging fruit of all.

This is clearly not crazy, because Instagram is already experimenting with hiding Like counts. So get rid of all of the public-facing numbers. Advertisers, or anyone dealing with sponsored content, could have access to their data, but for all organic usage, let's make it a numbers-free space.

2) Chronologize

Turn off the algorithmic ranking and make everything Reverse Chronological order (the newest item is first).

For this one, let's talk about Twitter. They've had a decade-long battle between Team Algorithm and Team Reverse Chrono. There were experiments like Top Tweets as early as 2010, but the default was always reverse chrono.

Now, I know that I tend to view all business decisions in the context of a stock price chart, but it was shortly after Twitter’s post-IPO crash that they started injecting tweets from accounts you don't follow into your timeline. In the Spring of 2016, they announced they'd be launching a default algorithmic feed (just a few months before a certain earth-shattering event..hmmmm). Then we reached the period where you had to dig into your settings to turn off the algorithm. They’ve gotten a bit better with a toggle switch in the top right between “Latest” (reverse chrono) and “Top Tweets” (algorithmic:

Thanks to Twitter's odd development cycles, their Mac app (which has now disappeared, but is supposed to be coming back) was still default reverse chrono, with no ads, and no feed-view metrics, even after the big switch to algorithmic in 2016. Switching between my desktop Mac app experience and phone was a perfect controlled experiment.

I can honestly say there was a palpable emotional difference between consuming an algorithmic feed and a reverse chronological one. On my computer, the experience was, while not quite serene, a calm one. It was just the people you followed, what they tweeted, in order. Meanwhile, on my phone, the world was ending and everyone was terrible, every time I opened it. The switch has even changed the way the people I follow tweet.

Mark Zuckerberg has said that the Facebook experience would be significantly worse without the algorithmic ranking. Back in 2014, the average user had 1,500 possible pieces of content, but they would only see around 100. The justification is that we need some system to sort through that sea of content.

Let the experience be worse, if being better simply means more addictive. Fried chicken is better than quinoa. But that doesn’t mean I should eat it every day.

Keep the feed reverse chronological by default. Every time a person logged in, they could manually opt into an algorithmically ranked feed, but they'd receive a cigarette label style warning on the dangers of algorithmic consumption. And it wouldn't be some American text-based package warning, but instead, one of those insane European ones:

3) Limit Sharing Velocity

Add some restrictions and friction to sharing.

All these platforms were built to encourage the "lizard-brain share", where you re-broadcast some piece of information with a viscerally charged tap or click, accelerating the spread of that content. It's why the guy who built the Retweet functionality said it was like "handing a loaded weapon to 4-year olds".

While the first three planks address problems for the individual level, this is the most important at the macro or societal level. This is how lynch mobs are born.

This one is also very achievable; WhatsApp is already working on it. Ahead of the Brazilian elections, they began experimenting on reducing the number of people you could forward a Whatsapp message to five users.

Apparently, they were going to impose this across the world in Jan 2019, but somehow the WhatsApp group with about 15 Indian relatives is still mostly forwards like this, so I’m not sure exactly how this worked out:

There are a number of ways to implement this. Maybe, you limit the number of times any specific tweet could be retweeted or replied to. My own personal dream is that, to share a piece of content, the user has to write a Margins newsletter's worth of analysis on why they are sharing it. Instead of a maximum 280 character count on Twitter, maybe we impose a minimum character count. You all can share whatever you want, but it's gonna cost you in time and effort.

Whatever the solution ends up being, just slow down a person's ability to share.

4) Explainability

There is an increasing push to have companies explain the decisions made by their algorithms. Explainability is more often brought up in life-or-death arenas, like whether someone gets a loan, or even more apt, an algorithmically informed sentencing in a criminal case.

But this is also important for an algorithmic content feed. I should be able to click through on any post and receive a plain-spoken explanation of why I was shown it. "You are seeing this because"....

I was going to get more into this one, but while writing this post, I learned that Facebook has already begun implementing exactly this.

I'd personally love to see explainability like "you are seeing this content because you were deemed vulnerable to totalitarian propaganda", but this is a great start. Well done, Facebook.

5) Regulate the employee incentive structures.

The first four planks are all product-oriented, but as a big believer that underlying business models are the true driver of all things product, we need to address the real villain behind the curtain.

We need to implement regulation on the incentive structures for employees at big social media platforms.

That includes the product managers, "partnerships" people (fun fact: MBAs at big tech firms will never say they work in sales), ad salespeople, managers; everyone. I still marvel at this epic April 2019 Bloomberg piece on YouTube's north star, and how a very specific and strategic business shift clearly led to NaziTube.

This may sound like heavy-handed, nanny-state type stuff, but I worked on a bank trading floor through the 2000s. From what I lived, two critical factors that helped catalyze the financial crisis were monopolistic control of capital transactions amongst the big banks, combined with distorted individual financial incentives.

  • If you were a big corporation, you simply had to do stuff through the big banks. Capital raising, currency management, all financial services - the big banks were the only ones with the credit, risk and reputation ratings that allowed you to do business. This meant, no matter their behavior, they had a baseline of revenue and profits.

  • Meanwhile, sitting on top of those massive pools of capital, were a bunch of traders (including myself) that were incentivized to take as much risk as possible. You got paid based on performance at the end of the calendar year. You didn't think of any external consequences, and if you lost money, you went somewhere else. There were even cultural quirks like, the bigger your losses were, other banks would often want to hire you more, because there “had to be some reason you were allowed to take so much risk”.

The entire culture changed. The Volcker Rule banned short-term proprietary trading within banks. The bonus structures changed to where you got paid in almost all stock, much of which could be clawed back. At the larger bank-level, capital requirements were increased. At both the micro and macro level, regulation transformed the compensation culture of bank trading floors, which changes the nature of the revenue composition, and products themselves. Most of my friends still there complain about it. That's how you know it’s working (sorry, fellas).

The analogy isn't an exact one, but my point is, to make any of this work, there has to be a change in the compensation culture at the big social media platforms. Compensation (stock or cash) based on increased usage and engagement targets has to be somehow changed. I'm not sure what that exactly looks like, but I am certain that as long as this isn't fixed, even with the implementation of the first four planks, we’re still going to be facing the same challenges and no SMART Act will be able to save us.


The World's Best Chicken Comes From Hainan - If you follow me on Twitter, you may have noticed I have a bit of a chicken obsession. It’s primarily for the fried kind, but having lived in Singapore, Hainanese Chicken Rice is in my chicken canon. It’s something that has long been lacking in NYC, but a new place just opened up in Midtown called Gai that was so good, it led me down a long chicken-reading rabbit hole.

Above I linked to the best overall food-travel writing on the dish I found, but I also wanted to pull my favorite line, from this piece (about the new spot, Gai):

While the dish has roots in the wenchang chicken of Hainan, it was created by Chinese immigrants in Southeast Asia. Fanatics call it the ne plus ultra of chicken, and this dish is an exercise in chicken — not fried chicken, not vinegary chicken, just pure chicken. (It’s like a composer layering a single note on itself, if that note were really good chicken.) 

After Trump cites Amazon concerns, Pentagon reexamines $10 billion JEDI cloud contract process - This Washington Post article on the current AWS / Pentagon / Trump conflict reminded me just how complex all this shit is. This Oracle VP playing to Trump’s love of the conspiracy and hatred of Jeff Bezos is technically doing his job, and doing it well. But what level of salesmanship crosses the line between fiduciary duty and societal malfeasance?

Oracle has lobbied Trump aggressively on the matter, hoping to appeal to his animosity toward Amazon as well as former Defense Secretary Jim Mattis, who angered the president when he resigned last year over the administration’s foreign policies. Oracle Executive Vice President Ken Glueck, who runs the company’s policy shop in Washington, said he created a colorful flow chart labeled “A Conspiracy To Create A Ten Year DoD Cloud Monopoly” that portrayed connections between Amazon executives, Mattis and officials from the Obama administration.

That graphic made it to Trump’s desk and led to a discussion between the president and his aides, people familiar with the matter said. In April of last year, Oracle co-CEO Safra Catz also raised the issue directly with Trump at a dinner at the White House.