Elon Musk needs to make Twitter better
He'll succeed or fail based on product improvements, not moderation decisions
The Elon Musk Twitter saga ended as it began: with a very rich, right-of-center Twitter user buying a platform whose social influence exceeds its financial value, decapitating its top leadership, and adopting an antagonistic attitude toward the rank-and-file staff.
Musk also penned an open letter to Twitter advertisers promising, among other things, that under his leadership the platform will not become a “free-for-all hellscape.”
Many of the responses to this promise emphasized the idea that there’s no amount of free speech nostrums that will eliminate the basic dilemmas of content moderation. My favorites in this genre are Nilay Patel’s “Welcome to Hell, Elon” and John Herrman’s “Twitter’s Day of Spite.”
But both pieces, like a lot of Twitter coverage, are written from the perspective of someone who kind of hates Twitter. As someone who really enjoys Twitter, part of what makes me relatively optimistic about the Musk Ascension is that Elon Musk enjoys Twitter, which didn’t seem to be the case for Twitter’s prior board or even its top executive team. If you think of Twitter as a problem, then Musk almost certainly can’t solve the problem. But if you think of Twitter as a product with tremendous value but also a lot of functional limitations, then Musk plausibly can make iterative improvements to the product.
If Twitter gets better, more people will use it. And if more people use it, the value of the network increases. And at that point, Twitter becomes an online services company with a proper flywheel of product improvements driving user growth, which drives network effects, which drives revenue, which drives product improvements.
This won’t require ending all content moderation, but it will require acknowledging that content moderation isn’t the central solution to Twitter’s problems.
My general sense is that abstract debates over free speech and social media tend to be overblown — there’s no big push to make Apple put porn apps on the App Store or for Meta to allow NSFW Instagram accounts. There are some perennial Insta controversies at the margin, usually stemming from the arbitrary convention that women’s nipples are naughty and men’s aren’t, but no apocalyptic battles are being fought over this because it has no partisan implications. By the same token, Musk has vowed from the beginning of his Twitter saga to reduce the number of spam bots on the site, a straightforwardly speech-hostile stance. What Musk wants is content moderation decisions that command broad consensus rather than reflect a niche progressive view.
But the actual partisan implications of these changes are often unclear. Nobody in the Republican Party, including Elon Musk, is in a rush to get Donald Trump back on Twitter because his account is embarrassing to the GOP. Selectively moderating some of the most extreme alt-right accounts while letting hammer-and-sickle Twitter fly makes the left look crazier than it is and the right look saner.
Twitter is, despite its aggravations, a source of great delight for me. I interact with a wide range of people, have made a lot of virtual friends, and benefit from the wisdom of not only academic specialists and beat reporters, but also obsessive hobbyists and practitioners in a range of fields. I would like the software tools that Twitter gives its users to be better so that more people can get more out of it and build the value of the community. And I think on some level this really is more of an engineering problem than a political problem. Dramatically reinvigorating a tech company as old as Twitter strikes me as somewhat unlikely. But there’s a 30 to 40 percent chance that a new owner with a history of engineering success can pull it off.
Content moderation changes will probably be modest
Outside of the business press, the discourse around Musk and Twitter has, I think, tended to underplay the specificity of his interest in content moderation. But it’s clear that a key precipitating event was Twitter banning the conservative satire site Babylon Bee over a March 15 piece headlined, “The Babylon Bee's Man Of The Year Is Rachel Levine.”
That’s a mean joke that violates the standards of my own community. In the broader American community, though, the Bee’s view is far from marginal. In a 2017 Pew poll asking whether a person’s gender is determined by their sex at birth or if it can be different, “determined by sex at birth” won 54 to 44 with a huge partisan split. They haven’t re-polled that question, but an even blunter 2021 poll asks whether greater acceptance of transgender people has been good or bad for society. In that poll, “good” won with 38 percent, but 32 percent said it was bad with a hefty 29 percent saying it’s neither good nor bad.
This isn’t to say the Bee ban was bad, just to note that Twitter picked a side in an actively contested issue in American politics and society. And maybe that’s good! Back in 1947, Branch Rickey and the Brooklyn Dodgers took a side in the actively contested social issue of racial segregation, and we look back and say they were on the right side of history. I assume that’s how the relevant Twitter staffers see things.
Musk evidently disagrees.
And you’re free to disagree with him in turn. But as a business decision, there’s a difference between stepping in at the vanguard of a controversial social dispute and enforcing widely agreed-upon social boundaries. If you read the conservative media, they don’t use the n-word or call gay men and lesbians “faggots” and “dykes,” so a space that censors those kinds of slurs won’t be seen as violating conservatives’ own community norms. But progressives’ accepted wisdom on how to talk about trans people is just very different from conservatives’, so enforcing the progressive norm is an inherently more controversial step than enforcing well-established American norms.
Trying to stitch Musk’s various statements into a first-principle philosophy of free speech probably isn’t going to yield anything coherent. But I do think he wants to limit censorship on things that can command a very broad consensus. Exactly how broad? I don’t know, and I don’t think Musk knows either. But broader than Twitter Trust & Safety demanded, which would mean different calls on things like the Bee or the Hunter Biden laptop story, but not a total revolution in moderation.
Musk might try large-scale layoffs
Where we really might see revolutionary change is in Twitter’s staffing.
After saying he might lay off 75 percent of Twitter’s employees, Musk has backpedaled to say actually he won’t lay off 75 percent of Twitter’s employees. This is good because layoffs on that scale all at once would be crazy. But there really might be large layoffs.
I’m not exactly sure why this happened, but roughly a year ago there was a substantial vibe shift in Silicon Valley which holds that most large technology companies are massively overstaffed. Multiple CEOs of privately held tech companies have voiced this critique of their larger peers to me. They’ve also criticized the venture capital community for encouraging excessively rapid headcount growth, but some influential VCs are now saying they agree with this. And there seems to be some competition to engage in the highest possible estimates of overstaffing. Marc Andreessen says the good big companies should lay off half their staff and the bad ones are worse.
Nat Friedman, the investor and former CEO of GitHub, says “many tech companies are 2-10x overstaffed.”
I know folks are accustomed to me opining, but this is an area that’s really way out of my depth. But what I’m telling you, journalistically, is that the issue here isn’t that Elon Musk thinks Twitter is overstaffed. It’s that Musk is part of a large community of influential technology industry figures who believe that all large technology companies are massively overstaffed, and the odds are decent that Twitter is going to be a guinea pig.
On its face, the idea that a bunch of different companies all independently made the exact same costly error sounds a little fishy to me. But there’s a coherent story you could tell about it. Say you’re in a field where a star performer is extraordinarily valuable but where even the best judges have difficulty telling exactly who the future stars are going to be. In that case, it’s completely rational ex-ante to have hiring practices that lead to the addition of a lot of people who ex-post are pretty useless. In pro sports you have limited roster slots, so the busts end up getting cut. But if you’re a company experiencing strong revenue growth, it’s easier — and better for morale — to give the busts something to do as long as nothing egregious is happening. If you have a bunch of years of good growth, this really could lead you to accumulate a tremendous amount of dead wood.
Now these high-end estimates — firing 75% of the team, overstaffing by 4x or even 10x — sound kind of implausible. But the basic thesis that “generally flush times in Silicon Valley let these companies keep plenty of busts who could now be eliminated with no real negative consequence” makes enough sense that sooner or later someone is going to see if it’s true. Why not Elon Musk?
Make the product better
One of my favorite changes to Twitter in many years has been the deployment of a feature I never use, where you can issue a tweet that is limited to people who are in your “circle.”
I don’t do this, but I do sometimes see tweets from other people who have included me in their circle. And I find these are generally very high-quality tweets. Letting people somewhat limit the distribution of their tweets encourages some people to be looser, funnier, spicier, more provocative, and more uncensored. Which is to say it lets more people tweet well. This makes my experience as a Twitter consumer better. And I also sometimes have good conversations all embedded within a circle. And to me, this is the spirit of the right kind of changes for Twitter. I hope that some of the people who’ve stopped meaningfully engaging with their accounts for self-defensive reasons will be tempted by circles to re-engage with the platform.
Something that I do use to improve my Twitter experience is a third-party service that automatically deletes my tweets after a few weeks. This should be a feature of Twitter where you can either mass delete your tweets or else schedule tweets for automatic deletion. That would not only be user-friendlier, but as a Twitter feature, it could be more powerful. Instead of being gone forever, tweets could be “archived” — placed in a private status where they are visible only to the person who tweeted them, so you and you alone can revive an old tweet of yours.
A potentially more niche thing is that when I was promoting my book, my publisher asked me if any celebrities or high value accounts followed me. I had to say that I wasn’t really sure — unless another account interacts with me frequently, it’s hard to know who’s following me. Making it easy to sort your own list of followers would make having a big list of followers more valuable. And if having more Twitter followers was more inherently valuable, then more people would try to do it.
Another third-party service I like is MegaBlock, where you drop in a bad tweet and it automatically blocks not only the author of the bad tweet but everyone who “liked” the bad tweet. If you have a big Twitter account, it’s important to block lots of annoying people. If you don’t block annoying people, then you’ll be constantly feeling annoyed when you use Twitter, which will make the experience of using Twitter annoying. “Nobody is allowed to be annoying” is not a reasonable content moderation goal. What you need is things like MegaBlock functionality, the experimental downvote button, and other product tools that let people be exposed to less annoyance. It’s of course true that some people experience harassment and threats that go beyond mere annoyance. But putting too much emphasis on trying to draw the line between harassment and annoyance is a dead end. Every user should have easier access to better tools that pull the entire chain toward seeing less stuff that annoys them and more stuff that pleases them.
Turnarounds are hard
It’s easy to imagine a world in which this works, and it’s also easy to imagine a world in which it fails. Once a company has a lot of employees and they’re doing things a certain way, it’s objectively very difficult to get rid of a bunch of them, change the work processes of most of the others, and also bring new people into a chaotic situation.
And Musk, though he’s succeeded in a range of fields, lacks specific experience with that kind of work, so you could easily imagine this turning into a tremendous faceplant.
But he’s clearly a smart guy, he’s done a lot of things well, and Twitter is popular enough that there are plenty of genuine Twitter enthusiasts with all kinds of relevant expertise who might want to work there as part of a potential turnaround. Crazier things have happened.
More broadly, I think left-of-center people working in media or creative fields or academia or political advocacy need to get used to the honestly quite banal idea that many successful and capable businesspeople have right-wing political views. The whole point of right-wing politics is that successful businesspeople should pay less taxes, so there’s a natural affinity there. What you want from the people who make the stuff you use is well-run businesses, which mostly means businesses run by rich and successful businesspeople, which mostly means businesses run by people who like right-of-center political parties that want to help them pay fewer taxes. That’s life.
I hope that as the actual platform owner, Musk will see some benefit to himself in avoiding too many partisan controversies. But it’s also the case that many of us with left-of-center views have benefitted over the years from the work of conservative businesspeople, just as people with right-of-center views have enjoyed art and educational and health services provided by liberals. Twitter is useful to me, but the product isn’t as good as it could or should be, and that’s limiting the growth and the utility of the network — a new leader who’s determined to improve the product rather than optimize moderation policies could make things much better.
> I’m not exactly sure why this happened, but roughly a year ago there was a substantial vibe shift in Silicon Valley which holds that most large technology companies are massively overstaffed.
Many, if not most, tech companies are extremely reluctant to fire engineers for performance reasons until they run into a cash flow issue and have to do layoffs. You’ll find this general consensus across tech forums (e.g., Hackers News). I believe this is the case for a multitude of reasons.
First, engineers are among the most expensive employees to hire and onboard. A lot of resources go into recruiting, including numerous interviews with existing engineers and managers. And many candidates get rejected or reject the company. This adds up to a lot of time and money (guessing high 5 figures) just to hire one engineer.
Further, it can take months for an engineer to get up to speed at an established company due to all of the proprietary tech and knowledge. And engineers grow in value for years as they pick up more tribal knowledge of the firm’s codebase and systems.
Second, engineers on the same team become non-fungible due to working on different projects. We certainly try to minimize this by rotating people on to different projects so that they can gain more tribal knowledge. Yet every team member becomes the canonical expert on different systems since they simply have worked on different projects.
Third, there rarely is any short term value to firing a low performing engineer. Yes, some of us are an actual net negative by worsening the codebase quality or breaking things that require other people’s help to fix. Yet that is rare. More likely is a “quiet quitter” that makes some minimal, yet positive contribution. You’ll find numerous self-reporting on Hacker News of engineers only working 10 hours a week at FAANG firms.
Fourth, engineers have short tenures, commonly jumping to another firm in two years. It’s an open secret that an internal promotion at almost all tech companies is harder than simply getting hired at that higher level at a comparable firm. There’s a lot of debate among engineers about whether that is due to a failure of internal promotion processes or a failure of the hiring processes. And of course the most ambitious and highest performers are jumping firms more frequently.
So in summary, engineers are expensive to hire, possess valuable differential tribal knowledge, are almost always a positive contribution to the team, and constantly leaving anyways. So there’s just no reason to manage out low performers unless the firm runs into cash flow issues.
I think there is a fundamental difference between left-wing and right-wing speech norms.
Left-wing norms are about the sentiment; what a ban on "misgendering" amounts to is a ban on expressing the position that trans women are men; it doesn't matter what language that you use to express that sentiment, you are not allowed to say it.
Right-wing norms are about specific words. The question right-wing people ask, often, is "what words can I use?". Their mental model is that you can say that a trans woman is a man, but you can't use (list of slurs) to say so.
To pick an issue where the badness of the sentiment is less controversial:
Right-wing norms on race are that you can say that black people are intrinsically less intelligent than white people (they mostly think that this is incorrect as to fact, but it's acceptable to say it), but you can't call them by a slur.
Left-wing norms are that the sentiment that black people are intrinsically less intelligent than white people is, in itself, outside of the norms.
I should add that this is consistent: if an atheist says "there is no God", then right-wingers do not take offence (they disagree, often vehemently, but they don't take offence); if they use what is intended as disparaging language, like talking about "invisible sky fairies", then right-wingers do take offence at that.
You'll often hear right-wing people asking what words they can use to say what they want to say - and they rarely hear the truthful answer, which is that there are no words through which it is acceptable to the left to express that sentiment.
You can hear this culture clash all the time: right-wingers objecting to people using George Carlin's seven words; left-wingers happily using all of them and objecting to The Bell Curve, which never once uses a racial slur, but expresses a view of black people that is utterly abhorrent to the left-wing mindset.
The only word I can think of where left-wing people object to the word itself rather than the sentiment is the N-word. I'm aware that there is some discussion about two versions of this word, but as a speaker of a non-rhotic dialect, I can't actually hear the difference. Other slurs you can usually quote directly or talk about (the use/mention distinction).