What Elon Musk fails to understand about Twitter
“Algorithm” matters, but what matters a lot more is what we’d like the algorithm to do
Elon Musk has made many mistakes running Twitter, but the biggest? Assuming his new toy was a technology company.
Whether the billionaire CEO ever really wanted to fix the platform he’s now in the process of ravaging remains a mystery.
It’s just as likely he was trying to troll the libs with a buyout offer that, to his horror, was accepted. But he has certainly performed as if he wanted to fix it, and he has focused most on what everyday people might understand least: code. Lines and lines of code.
Before he owned the site, Musk hoped to put the code on GitHub for all to see — to open-source the algorithm, as lovers of industry jargon would have it. How this would help was never terribly clear, as strings of numbers and commands in a vacuum don’t always tell us how a system works in the real world, with real-time data flowing in.
They also don’t tell us why the numbers and commands look the way they do, what policies and principles the engineers were trying to put into practice.
Once he owned the site, Musk demanded that employees print out their last 30 to 60 days of code — soliciting physical reams of inscrutable “ifs” and “fors” that he promptly decided they should shred.
The revised directive was that “anyone who actually writes software” email him “up to 10 screenshots of the most salient lines of code” they had written in the past six months.
This fixation probably comes across as wacky to laymen and computer linguists alike. Yet it’s actually in line with the way the rest of Silicon Valley would-be unicorns approached their tradecraft for years, before they knew better.
Organising the world’s information
Companies like Facebook, now Meta, set out to solve almost impossibly large problems with the tiny instruments of bits or bytes: their missions statements, “bring the world closer together” and “organise the world’s information.”
The leaders eventually realised they would need English (and Spanish, and Hindi, and so forth) in addition to PHP and Java so they could write terms of service that users would understand. But still, they approached the rule-crafting task as engineers.
They thought they had coded a better, or closer, or more organised world. When something broke, such as Alex Jones hawking two-for-one Sandy Hook conspiracy theories and harassment how-tos, or neo-Nazi websites surfacing high up in search results for the Holocaust, either they tweaked the algorithm or they tweaked the rules behind it.
Until fairly recently, they thought little about the very human set of principles they would have to conjure up to make the hard choices about speech and safety that their projects demanded. And they thought even less about the human set of problems that algorithms more often reinforce than they resolve.
Extremely hard-core
Musk has basically literalised this mindset. Everything isn’t treated as if it were code; everything, to him, is code. He chucked out his trust and safety team, dragged more than 50 top-notch Tesla engineers into his takeover and elevated the “best” engineers already in-house who could stomach his mandate to go “extremely hard-core” above anyone else, while others exited en masse.
This is how we’ve ended up with three different colours of verification check marks and zero thoughtful changes to how the rules are made, how the rules are enforced, or how any of that is communicated to the public.
As far as what the rules are? Those tricky prescriptions so vexed the self-titled Chief Twit that he gave up and started conducting polls whose results he abides by when feels like it.
The trouble is, Twitter isn’t about code so much as it is about people. “The algorithm” matters, but what matters a lot more is what we decide we’d like the algorithm to do — as well as how we behave, which, with a culture war on, has been poorly of late. The problem has never been programming so much as it has been policymaking (plus, of course, moneymaking via advertising).
This is truer than ever now, when users are fleeing the site not because they don’t like how it looks, or because they wish their feeds surfaced posts more relevant to their interests, but because the new guy in charge acts like a teenager-tyrant with a grudge against journalists.
All this works both ways. People are Twitter’s greatest liability, but they’re also Twitter’s greatest treasure. The site has always been a conversation with seemingly everyone in it — which is the very reason everyone has felt they have to be in it.
The users are the value, and if the users leave, the value leaves too. Musk has missed his chance already, but if he were as smart as he thinks he is, he would have realised what he’s running isn’t a technology company so much as it is a humanity company.
Washington Post
Molly Roberts is a noted technology columnist