Facts and Fairness: A Review of “Character Limit: How Elon Musk Destroyed Twitter”

A deep dive into a new book that Taylor Lorenz has called “a masterclass in investigative reporting”.

As this a very detailed review, here’s the gist in three paragraphs:

1. A new book dropped this week, written by two New York Times reporters. It represents something like the second draft of the history of the Twitter buyout, and will likely settle many a mind on the subject.

2. The book was well-written stylistically, and I both enjoyed and agreed with chunks of it. But it sells itself as being fair and accurate, and I didn’t find it to be consistently either.

3. This should matter, regardless of one’s stance on Musk. The quality of journalism should never depend on how journalists feel about their subject. All misinformation has a cost, and we rely on journalists to reduce and not add to it. 

Of course, you don’t need to take my word for any of this. Indeed I’d prefer it if you didn’t! While I’ll lay out my findings as clearly as I can, I am not the ultimate judge.

---

Corrections: As always, we reward corrections. Spot something wrong, misleading, or unfair? Use our anonymous Typeform or drop a comment in this post’s dispute doc.

Disclosures: I once worked at Quora, which is an X competitor in some vague sense. I have no personal financial relationship with Musk or his companies outside of a nominal $4/mo creator subscription fee. I’m also a long-term paid NYT subscriber.

---

Table of Contents

I. Why Write This | II. My Biases | III. Journalistic Standards | IV. Thesis Issues | V. Sourcing Issues | VI. Honesty Issues | VII. Narration Issues | VIII. Fairness Issues | IX. Curiosity Issues | X. On Musk vs. Agrawal | XI. On Moderation & Community Notes | XII. On Severance (Twitter Executives) | XIII. On Severance (Twitter Employees) | XIV. On Mass Layoffs | XV. Conclusion

---

I. Why Write This?

Though I’ve only ever once looked for a specific opportunity to do so¹, I’ve written a fair amount about Elon Musk and his companies over the years. Mostly because my bent is to point out the ways in which even prestige journalism can break down, and no other subject so routinely produces examples useful to that end. Even when looking for other things to write about, stories about him tend to angrily ripple through my feeds and chats.

My forever point: no person should be less above criticism than Elon Musk.

Journalism can involve judgment, and often should. Especially of the very powerful. But readers can’t stably judge or improve that which they don’t fully understand. Approaching coverage with anything less than genuine curiosity can only ever muddy the waters and make true accountability more difficult.

My experience is that lots of journalism about Musk is a Gish Gallop. The stories very rarely go to depth on any single charge in a way that would help anyone understand it in its full context. Typically it’s more a carousel of new claims, each framed as symptomatic of some old and concerning pattern.

Of course, some claims are true! But this isn’t that helpful when readers have limited ability to tell which are which! What books like this need is ironically the very thing that these authors almost entirely left out of their narrative—a tool like X's Community Notes to sticky-note disputes to specific claims. As such, this review is my version of that: a collection of issue-specific notes for readers to ponder and judge.

(For more on why I think it’s important to correct the record no matter the subject, see here.)

---

II. My Biases 

Note: I’ve largely stopped referring to journalists by name in my writeups, in part because my focus is more on systems than individuals, and in part because I don’t mean to hijack their search results. So I’ll be referring to the two authors here by their initials, KC and RM.

This isn’t my first time covering these two reporters. While I try to approach with a clean slate, I’m not sure that’s ever 100% possible. As such, some disclosures:

• RM. I’ve reviewed his past Musk coverage at length. For representative examples see here (Thai cave rescue) and here (Musk Foundation). Note that he repeats some of the same mistakes from his prior Thai cave reporting in this book, as eg. on page 24 where he advances the (very misleading!) position that Musk’s mini-submarine would never have fit through “the first fifty meters of the [cave] dive”.²

• KC. While I'm only really familiar with her Musk coverage in relation to this present story, I remain deeply concerned with how she and her (non-RM) co-reporter treated teen Mason Sheppard in their coverage of the big Twitter hack back in 2020. Detailed recap here. Though, as I only ever heard back from her co-reporter, I don't know how to allocate fault for that one.

Similarly, I can’t tell you which decisions each of these two made in writing this book, nor their editors. As such I take few hard stances on point-by-point responsibility, other than that both of their names are on the cover.

---

III. Journalistic Standards

Here’s a KC quote from a book interview she gave this week (starting at 40:25, emphasis mine):

What are the ways that you hold someone like Elon accountable? He has so much money, he has so much power, he has so much fame. What are the checks and balances on a person like this? And it turns out the best way to hold him accountable is to get his Twitter followers angry with him.

While she clearly meant this as a neutral observation more than some admission of tactics, there’s a way in which this articulates exactly how many journalists approach their Musk coverage. The end goal is often inspiring some kind of purportedly needful course correction, but approached through stirring targeted outrage. While this may work when done with care, there’s an obvious trap here: in the aim of fitting claims into this optimal packaging, one might easily leave out too much of the truth.

Lots of people dislike Musk for various reasons, including many journalists. This is naturally fine in itself, so long as it’s for things he’s actually said or done. But how do people know what he’s said or done? In part through his interviews and tweets, but in larger part also by what they read about him. The latter is not less powerful, in that it tends to color how one parses the interviews and tweets they consume directly!

That in mind, both authors’ beats at the New York Times are bent towards covering Musk and his companies. Given the NYT’s cultural prominence, their coverage has a not-tiny impact on how the world perceives him. 

The NYT of course claims to appreciate all this deeply. To quote the big boss there:

Independence … demands that we reflect the world as it is, not the world as we may wish it to be. It requires journalists to be willing to exonerate someone deemed a villain or interrogate someone regarded as a hero.

Well put, truly. But is he describing some ideal or actual policies he holds all his journalists accountable to? 

Relatedly, both of the authors’s NYT bios include the same boilerplate: 

I’m committed to protecting my sources and writing fairly and accurately about the companies I cover.

RM also reiterated this more personally in a Threads post back in June, in response to unfavorable disclosures about how his work in covering Zuckerberg and Facebook was viewed by some inside the company:

My job isn’t to make these folks comfortable. And I’ll do what I can with my fair and accurate work to make them less so.

In KC’s own words, RM is “universally unpopular with CEOs”. Maybe this is because of his preternatural ability to speak truth to power with fairness and accuracy! Or maybe, you know, it’s not. But I’ll let you decide. 

---

IV. Thesis Issues

How did the book’s early readers (ie. reviewers) understand it? Well, let’s look at a few snippets highlighted by their publisher:

The definitive account of how the world’s richest man, in a fit of unbridled vanity and arrogance, took over and destroyed our digital town square. - John Carreyrou

...the definitive business book of the 2020s — a meticulously reported tale of tech-industry hubris, narcissism, and egomania collapsing in on itself… - Max Read

Sounds strong! 

Punchier yet, here’s the closing bit from the book’s intro (page 8, emphasis mine): 

To those users who have hung around, one of the most important modes of global communication has become practically unrecognizable and now serves the interests of one man. What was once called the digital town square is becoming Musk’s mirror.

Oddly, the book is never super clear on what it means that Twitter was “destroyed”. If anything, it offers a detailed articulation of just how many problems existed on and at Old Twitter that have simply survived all attempts at taming them. While they present these problems as having worsened, even there they don’t make much of a scientific case, nor do they show any real interest in any counter-narratives.

Apart from losing its old name and bird logo (RIP), the app looks about the same today and is being used roughly the same way by roughly the same people. It apparently has more daily users than ever, albeit with a stalled growth rate. There are mostly just fewer ads now, along with fewer links, more features, more checkmarks, and more doom. While those are meaningful and mixed changes, this very melodramatic intro sets up what I see as the real premise of the book: a set of primarily vibes-based values arguments.

I dunno, I’ve read the book and this guy seems basically right! The story to me is that Twitter was notoriously ungovernable, and then was bought by a notoriously ungovernable entrepreneur, and the authors broadly just dislike the latter's changes.

Note what KC admits in that same podcast interview I mentioned above (at 15:44):

I mean I joke with [RM] a lot that we wrote this sort of dramatic book just so I could sneak in my thesis on content moderation in the middle.

She laughed as she said it, so I do take it as a joke. But also I’m not sure it’s entirely just that. The book really does read as a missive that sells “wild man destroys cultural institution” at least in part as a vehicle for advancing the authors’ own thoughts on the values and tactics they feel Twitter/X ought to have pursued. And this would be fine if they made that more clear in the book itself. But they didn’t.

Anyway, as to that claimed destruction, if we break it down by category:

• Financial health. Yes, X is selling fewer ads now. But this is at least partially offset by X’s 25% purported ownership stake in xAI (worth $6bn as of last May and expected to grow substantially), which the book never mentions. Also, more damningly, no real room is given to considering that the watchdog reports that initially scared off advertisers may have been, uh, less than reliable (much more on that in section IX). While surely Musk has not had, like, the best possible response to advertisers, did he fire the first shot? And if those reports were sketchy, do journalists not share some blame for boosting them uncritically?

• Moderation. Yes, X has a different philosophy here than past Twitter (well, kinda). But there’s, stunningly, not a single reference to Community Notes in the book’s narrative! (It devotes about a single page, out of 468, to its predecessor program, Birdwatch, but then leaves that subject entirely in past tense.)

• Community. Yes, X has let some controversial voices back on the platform, including many I personally find troubling. But as the book is clear about, this was also the prior CEO’s plan (more on that later as well). Musk just settled on a different idea of what moderating problematic content ought to look like (eg. scaling Community Notes). While I do indeed find myself today in more collisions with tweets I dislike, that isn’t to say that the platform has been “destroyed”. And most of those tweets are annotated now! Plus I can annotate them! And anyway bad tweets aren’t new. Many prominent users, including one of these authors, regularly referred to Twitter as a “hellsite” (here, here, here) years before Musk’s acquisition. Yet most journalists are still as addicted to it as ever.

There’s a way in which the book, intentionally or not, is part of a larger attempt at self-fulfilling prophecy. Call a thing “destroyed” to popularize that narrative and maybe you help it happen. But the old Twitter is hardly “unrecognizable” in X today. 

---

V. Sourcing Issues

While this was likely a publisher decision, one frustration I had with the ebook is that there’s almost zero on-page sourcing for anything (ie, no footnote numbers). Apart from two or three asterisk footnotes, the rest were all endnotes. Not only did this make fact-checking difficult, but it also obscured just how many claims have no disclosed source. And I don’t mean cases where they obviously had some anonymous insider source to protect, but also just assertions of ostensibly public facts.

As an example, in presenting what they suggest was an escalation point in the Zuckerberg-Musk feud, they recount a 2016 mishap (pages 35-36, emphasis mine):

...Facebook had contracted with Musk’s company and spent $200 million to launch a satellite called Amos-6, which was intended to bring mobile internet—beamed from space—to parts of Sub-Saharan Africa. [...] Despite Musk’s disagreement with Zuckerberg, SpaceX did not turn down Facebook’s business. It needed the money and attention. [...] The company loaded the Facebook satellite onto a Falcon 9 rocket [...] Two days before the scheduled launch, SpaceX’s crew was running a test of the rocket’s engines when it suddenly exploded in a ball of flames, incinerating Facebook’s investment.

Four problems here: 

1. There are zero sources for this in the endnotes.

2. The authors seem to have confused the cost of the satellite (~$200m) with that of the launch itself (closer to a quarter of that). Something they’d probably have caught if they looked up a source!

3. The satellite was insured, and SpaceX launched its replacement for free.

4. The satellite wasn’t actually Facebook’s! Nor would they have contracted the launch with SpaceX.³ They just had an agreement to lease some of its bandwidth. And this lease was also insured. Facebook’s investment was not incinerated! 

Of course this isn’t a terribly important anecdote, and whether the explosion or its aftermath changed Zuck’s and Musk’s relationship for the worse only those two could say. But this is sloppy work, and it getting into print unsourced sets a ceiling on the level of editorial scrutiny that was applied pre-publishing. Any amount of checking, even of the event’s Wikipedia page, would have caught this.

---

VI. Honesty Issues

In recounting what they portray as a Musk meltdown weekend shortly after his Twitter purchase went through, they make a strong though oddly-worded claim (pages 320-321):

And he responded to a tweet that featured a quote from a white nationalist. Musk was always unfiltered, but his weekend felt especially manic.

In keeping with the last section, there’s no quoted source for this, nor any further clue about what the tweet was. Note also the artful vagueness of “responded to a tweet that featured a quote from”. Did Musk endorse it? Acknowledge it? What did it/he say? Well, my googling led me to this prior RM article. In it we find a source paragraph that was reworked for the book (fine to do in itself imo), which included a similar version of this line, barring one major difference (emphasis mine):

He posted, then deleted, a tweet engaging with a quote from a white nationalist.

And then later in the same article (links theirs, the first of which is dead);

On Sunday night, Mr. Musk responded to a tweet featuring a quote from a white nationalist, before deleting the post and moving on to squabble with Mr. Dorsey…

Ok, so the reality of it. First, the Musk reply tweet was to a quip wrongly credited to Voltaire. Musk simply agreed with an innocuous quote in his mentions not knowing that it had been misattributed. When it was flagged to him that it actually came from a white nationalist, he deleted his comment. (The original context of the interaction was a user praising Musk for being open to criticism, which was also left out.)

Second, the book combines the phrasing of both these sentences, despite both of them failing to tell the real story! What we’re left with is the implication that, in some special mania, Musk knowingly engaged with a white nationalist quote. But he didn’t! And RM very much knew this! What his motivation may have been here, or the influence of KC or the book’s editors, I can’t say. We simply know, from his original reporting, that he knew. 

Anyway, the book doesn’t source this, leaves out the essential context, repurposes a weasel-worded description, and drops the previously admitted point about the tweet deletion. However this happened, this is wildly inconsistent with every journalistic ethic I’m aware of. It’s a fundamentally dishonest output, and RM, being on the record as knowing the real story here, bears some minimum responsibility. 

It’s not that Musk never tweets controversial things, including some that I personally take some issue with. But this level of deceptive framing can’t possibly produce a healthy conversation! (I note too that there’s a “damned either way” dynamic here. If someone is going to get slammed by the press for their actions no matter the context, would we expect their sensitivity to…increase?)

---

VII. Narration Issues

As a prelude to the next section on general fairness, one important thing in journalism (in article or book form) is separating your own thoughts from those of your subjects. While we may on some personal level find more empathy or alignment with one side, we can’t adopt one team’s POV and language as our own and then expect people to find the end product to be scrupulously neutral!

For the most potent example of this objectivity collapse, let’s start with a description from page 250: 

To the Twitter employees, these outsiders [Musk’s transition team] could not be trusted, and they came up with a nickname for the intruders: “the goons.”

I reckon it’s fine that they included this in general. While we don’t get a sense of just how many Twitter employees saw it this way, surely some did, and I assume that some subset did use the term “goons” as a derogatory shorthand. The problem is that the book then keeps repeating this term in its narration, not as a quote but as the authors’ own gloss! They do this 15 times! Also interspliced with “henchmen” (2x) and “lackeys” (5x). It’s hard to take this as journalistic neutrality! And mind you that they knew this book was likely to be screencapped heavily, where the original reference to “goons” being an insider term would be lost. 

Going a step further, they also slip in physical characterizations that are markedly different depending whether the character is considered positive (or at least neutral) versus actively villainous. In describing the former, their eyes are “cutting” (pg 1), “piercing” (pg 205), or “doe” (pg 239). But then when we get to the full-on narrative foils or Elon himself? They’re “darting” (pg 4), “beady” (pg 134), or “shifty” (pg 358).

This is bad pulp fiction narration, not prestige journalism.

---

VIII. Fairness Issues

The book teems with jabs and barbs, both in the service of denigrating their chosen villains and justifying their good guys’ grievances.

From page 211:

Savitt’s past had another twist that made him intriguing to his new clients: a previous run-in with Musk. Tesla had hired Wachtell [Savitt’s firm] to defend it in a shareholder lawsuit stemming from its 2016 acquisition of SolarCity, but when the firm lost a motion to dismiss the case, Tesla fired Savitt and his team, showing the lawyer how mercurial Musk could be.

It’s “mercurial”…to fire a firm who lost the motion you paid them to win???

From page 281: 

Those on the calls weren’t sure what to make of Musk’s proxy. [Antonio] Gracias came off as arrogant. At one point, after a sales leader from Tokyo on the call introduced himself, Gracias volleyed back some words in Japanese and asked about a local restaurant. He had lived and worked in Japan for a bit, he said confidently.

Banal personal pleasantries are…proof of arrogance? And why wouldn’t he speak “confidently” of his own life experience? Are they suggesting he made it up? If not, what job is that word doing here?

From page 292:

In group chats, some employees joked that the costumes seemed a little too on the nose. Others asked how many fifty-year-old men dress up for parties with their mother? 

If you’re really considering this to be relevant commentary, at least use an Arrested Development joke here. C’mon! 

From page 421:

At one point, [X CEO Linda Yaccarino] held up her phone to the audience, and some eagle-eyed observers noticed she didn’t appear to have the X app downloaded on her home screen.

While this was unsourced and I had to google it, I ended up finding this image. It’s of a single screen on her phone. Most people have multiple screens of apps, with rather haphazard organization as to which appear on which and in what order. This is a Reed Richards reach, for what journalistic purpose?

---

IX. Curiosity Issues

Here’s another banger quote from that above speech from the head of the NYT:

Independence … calls for carefully conveying ambiguity and debate in the more frequent cases where the facts are unclear or their interpretation is under reasonable dispute, letting readers grasp and process the uncertainty for themselves.

Not to go too Ted Lasso, but one good safeguard here is curiosity, as it by its own virtue leads quickly on to humility. There are a great many things we can’t know for certain, and the more curious we are the more we’ll be reminded of this anew. Human motivations in particular are often legion. We can rarely narrow them down to just one from the outside, and we do well to follow the above advice and present a full picture so as to let readers “grasp and process the uncertainty for themselves”. 

Of course, this is less effective to your goals if you're trying to weight the scale towards a particular narrative!

From page 297:

Then Musk’s paranoia emerged. “Did Twitter read my DMs?” he asked. He appeared convinced that the former executives had been snooping on him during the lawsuit.

The engineer tried to answer as best he could. The company made it extraordinarily difficult for employees to access a user’s direct messages, but some workers were allowed to do it when responding to reports of abuse or subpoenas. As far as he knew, no one had read Musk’s messages—but it wasn’t impossible.

Was this “paranoia”? What alternative motivations might we imagine here for this question? Or, better than imagine, what do we know about the subject?

I mentioned the big 2020 Twitter hack in passing in section III. One less-discussed aspect of the hack, that to my knowledge I was the first to report, was that Twitter quietly took down an archival tool during the fray that included DMs. My sense was that this was likely done for a reason, and indeed it was later confirmed (by KC herself) that dozens of accounts had had their DMs accessed and/or stolen.

Now let’s recall to whom one of the more famous hacked accounts belonged. If you were going to peek at DMs while you did your crimes, would Musk’s not interest you? Given the presence of massive security gaps around DM access in 2020, would 2022 Musk not have reasonable residual concerns here? It wouldn’t require paranoia to ask this particular question! 

More concerningly though, let’s move on to page 302. 

He had read reports that usage of racial slurs had increased dramatically on Twitter since Musk’s takeover, as trolls were emboldened by the new ownership to spew hate.

Which we can match with a longer sister claim on page 392:

Activist groups that study online platforms, including the Center for Countering Digital Hate and the Anti-Defamation League, found that slurs against Black Americans had tripled, while antisemitic speech was up more than 60 percent since the change in ownership.

These claims were acutely painful for Musk’s Twitter at the time, and went a long way towards setting off his ongoing advertiser showdown. Are they true? Well, both are sourced in the book solely to KC’s own prior NYT co-reporting, which doesn’t itself link out to any sources. It simply names the groups that published the findings. 

Even so, let’s start with the slurs. We can tell from the numbers quoted in said article that the underlying “research” comes from this CCDH blog post. It found that in the month of November 2022 (Musk’s first full month of ownership) there were 3,876 daily references to the n-word⁴, compared to a 2022 average of 1,282. There’s a big causality problem though, which we see in this chart from the CCDH’s sister blog post:

This chart is very spiky, with the November “Musk Bump” average being about dead even with the spike that started about a week before his purchase closed. And that spike is also only slightly larger than the one in March/April 2022, which crucially had completely evaporated before Musk first announced he was buying Twitter. Clearly something drove both, though perhaps not the same thing, and it’s not intuitive that either had any connection to Musk. The CCDH predictably expressed no interest into investigating this mystery.

This curious incuriosity is an even bigger problem with the second claim here, about antisemitism, where we have a much clearer idea of what exactly was being measured. The underlying ADL report is similarly unlinked, perhaps in part because there was no report. It was just a tweet!

Per said tweet, the ADL looked at “over 529K tweets mentioning ‘Jews’ or ‘Judaism’ both before and after Elon Musk’s takeover” and found that 163 in the after group contained antisemitic sentiment, vs 101 for the before. So there’s our ~60% increase. 

While any increase is abhorrent, there are some obvious questions about methodology here. The ADL reduced ~529k tweets to about ~1,100 for closer analysis. Ok, but based on what? How do we know that this subset was representative? The ADL seems to have never produced an actual publication or shared anything about their process.

More to the point though, why are we focusing here on such a narrow time window? As it happens, relevant things happened during that window that had little to do with Musk’s Twitter buyout. Surely there have been longer studies done since, with public methodologies? If they indeed exist and come to rhyming conclusions, why wouldn’t a book written well over a year later point to them instead?  

As a final extension of this point, here’s one more related claim from page 425:

In a meeting with the sales team the following Monday, two days after the SpaceX launch, [X CEO Linda Yaccarino] slammed Media Matters for its report, claiming that the organization had manipulated a Twitter feed so that ads would show up next to the white supremacist content.

The resulting trial is still active. The curious reader can find the gist of X’s arguments at paragraphs 9-13 here. In reviewing the filings to date, it doesn’t seem like Media Matters has yet to meaningfully rebut those questions about manipulation.

Curiosity is good, even when—or perhaps especially when—it leads you into complexity and ambiguity and very long and winding warrens of hard-to-parse new facts. That’s where the good stuff is. It just doesn’t sell as well as moral certainty.

---

X. On Musk vs. Agrawal

I’ve removed the subject’s name from these excerpts (taken from pages 83-84, emphasis mine). Who does it sound like the authors are talking about? 

[Person A] wanted to restructure and streamline the company, getting rid of several leaders in the process. … [Person A] wasn’t sentimental. He believed he needed to do what was best for Twitter, and that meant radical change to invigorate the company. …

[Person A] didn’t need to stroke egos by handing out C-suite titles, and he didn’t want to waste time sorting out who was making bad choices. He wanted a few effective lieutenants, and clear accountability for screw-ups. He suggested thinning out Twitter’s top ranks to just three executives, including himself. 

The person on the other end of the phone call was overwhelmed. The timelines [Person A] proposed for firing executives and revamping the company were aggressive. [Person A] had a lot he wanted to do, and he was eager to act.

Of course Person A here is Twitter’s pre-Musk CEO, Parag Agrawal. In what’s really the most fascinating part of the book, it presents the differences between him and Musk as being more interpersonal than ideological. Agrawal wanted to accomplish many of the same things, in broadly similar ways. He wanted to radically transform moderation policy (see all book references to “Project Saturn”), rethink Trump’s account ban (page 147), close non-core offices (page 97), and significantly pare down the company’s headcount and spending (see references to “Project Prism). He just never got a chance.

What I find odd is that the book is explicit about this parallel. Twitter of 2022 had to change, and the above was the planned direction of that change. Musk then emerged as a wild card, and many of these plans were disrupted, adapted, and/or thrown out. Even so, Agrawal’s counterfactual Twitter 2.0 might not have looked so different from Musk’s X. Not the same, no doubt. But also not, shall we say, unrecognizable from it. 

Sadly this book is not titled “The Victories and Stumbles of Twitter, Now X, Owing to Diverse Causes, Many of Them Long Endemic”. Which is a shame. Because that book would have been more useful in my view, and could have kept an awful lot of the same content, just with the least defensible bits taken out.

---

XI. On Moderation & Community Notes

Forgive the long excerpt here. It’s important. Taking from pages 173-174:

Inspired by Wikipedia, which allowed anonymous internet users to write and edit its corpus of encyclopedia entries, [Keith] Coleman had dreamed up a similar system for Twitter. He called it Birdwatch, and it allowed Twitter users to fact-check controversial tweets and add notes to them with more context. The notes could debunk misinformation or correct the record as fresh headlines emerged. Wikipedia’s army of editors enabled the site to move at light speed, and the site was often updated within minutes after news of a celebrity death or an international conflict broke.

Coleman’s idea was that Twitter could do the same, letting users fix misinformation on the platform faster than its content moderators could. In January 2021, he began rolling out Birdwatch, which quickly became popular with Twitter’s leadership. Alongside Dorsey’s nascent investment in Bluesky, Birdwatch showed that Twitter was opening its gates to users, letting them have more control over their experiences on social media.

It’s a bit devastating that we only get about one page on this, and a bit unbelievable that the book then never returns to what Birdwatch grew into: Community Notes.

Under Coleman’s guidance, X Community Notes, though still in its early days, has unlocked positives that no number of former human moderation banners ever could. Because it allows for contextualization at scale, quickly and democratically—including on Musk’s own tweets, even after you’ve seen them. It’s not a perfect system yet, but neither are Coleman and team done with it yet. There’s a lot to admire already.

Twitter has always contained multitudes, including lots of misinformation and hatespeech. These things are ugly and destructive and we should wish and fight for less of them. Musk’s present philosophy is that they’re better combatted than banished, where instead of relying solely on human enforcement of moderation rules (which still exist) the community can cut monetization and distribution using Community Notes, which has the additional upside of showing why a given claim is problematic.

I don’t understand how a book, that at its core is deeply concerned with the issue of content moderation, could just reduce all this to a single past-tense page. There were over half a million Community Notes contributors as far back as May. The book’s epilogue references an update from the same month, and it’s not like Community Notes was a super-secret project during the period they wrote it. They had ample opportunity to cover it. Why didn’t they? 

(Also, when is the New York Times going to offer something comparable to combat misinformation on its own platform? I’m a subscriber, and a general believer that they produce a lot of good work. But certainly not all of it is good. And in the rare occasions that they do accept corrections, they may not always note them. Can we expect a book about that too?)

---

XII. On Severance (Twitter Executives)

From page 229: 

The billionaire chafed at the idea of paying the Twitter executives’ multimillion-dollar severance packages, known as “golden parachutes,” and he wanted to wage a scorched earth campaign against everyone he thought had wronged him over the course of the deal.

Golden parachutes help grease the wheels of an acquisition by counteracting the natural impulse of the target company’s management to fight the deal to save their jobs. While the buyer doesn’t necessarily need their support, it’s generally preferable. So top executives tend to get a generous pay option, often in the tens of millions, that activates in the case of an acquisition as a reward for rowing in a helpful direction.

In Twitter’s case, Musk opted to chuck these parachutes by immediately firing the core executive team once the deal cleared. As the book relates, they meant to resign, but didn’t do so in time. Was firing them fast legal? It hinges on whether the firings were justifiably “for cause”, which would require either “gross negligence” or “willful misconduct”. X’s / Musk’s arguments to this end are represented on PDF pages 10-11 here and PDF page 3 here. 

This aspect is still being litigated, via at least two separate suits (main case here). Barring any settlements, the courts will be tasked with deciding whether Musk’s concerns were strong enough or whether he has to pay up. In general this seems a fair one to defer to the courts. 

Though it is worth highlighting one aspect of this (from pages 247-248):

In a call days earlier, Edgett had briefed [Twitter board members] Taylor and Pichette on a proposal from Savitt, in which the outside lawyer asked that Twitter consider paying his firm $95 million for their work. White-shoe law firms like Wachtell sometimes charged what they called “success fees,” lavish honorariums they were awarded only if they were victorious in their cases. The success fees vastly exceeded their already robust hourly fees and could enrich the partners for years to come.

…[Twitter board member] Lane Fox responded to Edgett almost instantaneously:

O 

My 

Freaking 

God

She broke her sentence into a haiku-like format for extra emphasis. Lane Fox never dreamed the total would be so extravagant. But the board knew there was no time to waste. If they didn’t pay Wachtell before the deal closed, it was unlikely that Musk would pay them at all. And didn’t the lawyers deserve a reward for trapping the rule-flouting executive into the transaction?”

While it’s good that they included this, it’s also symbolic of why Musk is fighting those severance payments.⁵ This wasn't a $95m invoice per se. The true (also hilariously inflated) invoice was about $18m. This was a $95m request (later reduced to $90m), of the same sort an unusually ritzy restaurant might suggest above-and-beyond your bill. Faced with a, say, 500% optional service fee, you might be more tempted to pay if you're using someone else's credit card! But that person will likely not be happy!

(For an excellent rundown of this whole invoice thing from an ex-Wachtell guy, see here.)

---

XIII. On Severance (Twitter Employees)

This was the hardest section to write, as layoffs affect the rank-and-file in far more acute ways than executives, and there are often complications around visas and benefits that deserve our empathy. I’ve been there. It sucks, deeply.

That said, the book makes its most controversial claim on this subject in drive-by form, never stopping to really dig into the legal complexities. And then later it fully undermines that earlier gloss, and never explains how we’re supposed to parse this.

To make sense of it all, we need to divide between two distinct groups: (1) those who were laid off involuntarily early on, (2) those who took Musk’s later offer to take severance as an opt-out from staying on at “hardcore Twitter”. 

For the first group, a bunch of lawsuits and arbitrations sprang up almost immediately, many of which are still ongoing. One core plank that you’ll find in some was an idea that they were obligated to get more than Musk’s Twitter offered, in line with the more generous prior severance policies of Old Twitter. 

(Note that I’m skipping over a bunch of the complexity around legal questions of standing, forum, class-action vs. arbitration, etc. That stuff is all important, though I think less central than just “did the merger agreement explicitly obligate New Twitter to pay anyone according to Old Twitter’s promises?”) 

For the sense, from page 235:

Workers panicked about the impending deal, fearing layoffs or cuts to Twitter’s cushy benefits. Executives tried to reassure them by pointing out Musk’s contract required him to leave pay and benefits unchanged for a year after buying the company, but the promises rang hollow. Employees knew Musk didn’t always honor his agreements.

While I can’t speak to what the employees were told, it’s worth noting that this isn’t quite what the contract actually said. Or at least it’s not all it said. 

While it’s true that section 6.9(a) (see page 51 here) said that benefit plans, including severance, would not get downgraded for a year after the acquisition, it’s additionally true that section 6.9(e) on page 52 can be read as making it clear that this clause was effectively optional.⁶ You can also further consider the force of section 9.7 on whom exactly was party to the contract. Anyway, what the courts will ultimately determine is not whether agreements will be honored, but what the agreements actually said.

In a world in which X’s reading of 6.9(e) and 9.7 are upheld, who owns the blame here? If an Old Twitter employee asked their bosses in good faith what the severance situation would be like post-merger, they deserved a fair answer. If indeed they were given a gloss of 6.9(a) without any caveats about 6.9(e), someone did wrong by them.

Setting that question aside though, I was a bit gobsmacked in reading this later take, on page 299:

One [severance option presented to Musk] was a bare-bones version that covered what Musk was legally obligated to pay, and the other was slightly more generous. The severance was uninteresting to Musk. He told Pacini to let Bjelde pick what he thought was best.

Two things here:

1. If Musk was indeed legally obligated to pay according to Old Twitter’s policy (as the book suggests above on page 235), how is his max legal obligation now lower? What changed? (The minimum in context here is per the WARN Act, or 60 days for Californian HQ employees. Which is far less than what Old Twitter paid.)

2. Musk found this all uninteresting? He deferred on picking? He was ok with offering more than required? Yet he screwed employees out of what they were legally owed? Which is it?

Anyway, now we get to employees who were given the option of quitting with severance, which is also framed as a nice positive (per page 353):

As it applies to those who left on their own, the NYT also says they got their promised severance. 

In a rare moment of grace, he allowed the company’s human resources team to send out separation agreements to the departing employees and accepted the resignations of a few more workers who had waffled over the decision, allowing them to collect the promised severance.

---

XIV. On Mass Layoffs

One recurring feature of the book—and much of the contemporary reporting—was that the mass layoffs over Musk’s first month were chaotic and a bit brutal.

For argument’s sake, we can imagine two ways of approaching them:

1. Conduct a slow and methodical talent review where all line managers make their individual cases for the net contributions of each employee. While this is in many respects the nicer and more just approach, it comes with tradeoffs. First, no manager particularly wants this duty. Second, it will leave the company largely paralyzed for a long duration of time.

2. Just use crude measures to form a net that you hope will capture most top performers, though knowing that lots of them will fall through, that lots of intangibles will get overlooked, and that a number of critical people will be let go only to receive an “about that” text later. This is messy and unpleasant in many respects. Its main virtue is that it’s fast.

Anyone who has worked in tech long enough has been through some version of 2. It’s not fun. But I don’t know that it actually sucks less than 1. Imagine that you don’t make the cut. Would you prefer to say to your next interviewer “I was found through an extensive review to have been a below-average contributor” or “hey look it was wild, you heard about it, anyway it all had nothing to do with me really”?

Personally, I’d hope for the second scenario, taking my chances that I’ve shown enough value to make the cut, while also knowing that if I don’t make it then no recruiter is going to blink when they look at my CV. While others may think through all this and still pick the first option (totally valid!), that there was no room in the book for this being a tradeoff felt unhelpful.

(It may also be true, as the book represents, that some of those doing the cutting at Twitter were needlessly insensitive. While I leave it to readers to reach their own judgments there, I certainly agree that, to the degree that we’re getting the full pictures here, all personal unkindness is bad.)

---

Conclusion

One last quote from the book (page 60, emphasis mine): 

The press pilloried Dorsey and Twitter, regardless of what he said. And most lawmakers seemed like they wanted to ask only about specific content moderation decisions: Why was this taken down? Why was this left up? Why was my account being shadowbanned? They didn’t appear to care about the future of technology or solving the problems they complained about.

That first bit is an apt self-indictment of tech coverage. While many individual reporters are excellent, in aggregate the press tends to pillory specific targets in a predictable way, more or less regardless of what they’ve said or done. The civic outcomes of this are not good, for either side. 

The second bit is also as true for journalists as it is for lawmakers. Do they truly care about the future of technology, including at X? Do they want to see it saved or destroyed? While some have partially shifted to Threads and Bluesky, Eternal September will arrive there too by-and-by. And the problems will be just as hard to solve, and deserving of much richer and more nuanced discussion. Because they're messy human problems. And as ever, the fault, dear Brutus, is not in our caesars nor our algorithms, but in ourselves.

---

1. In the very early days of this project, I had my name put forward to cover a Neuralink event as an unpaid one-off because I thought the tech was cool and figured an outsider POV might be helpful. I was one of several selected. But as I spent more time thinking about how to chase the mission here, I decided that this kind of coverage wasn’t really on-brand.

2. This is just the tip of the iceberg in all that’s problematic in taking a universally pro-Unsworth position in that dispute, which is all covered in exhaustive detail here.

3. I suppose it’s possible that Zuckerberg had some influence with Spacecom in their decision to go with SpaceX, though it’s not intuitive why he would get involved and my cursory googling doesn’t suggest this was the case. 

4. I wish the blog had also been clearer about whether they filtered for uses of the hard-r n-word vs. the soft r, not because one is inherently less racist when used as an epithet, but just to make sure that they accounted for friendly use by Black Twitter users. Because otherwise what we’re measuring in part is just spikes for events that draw in larger numbers of Black users. If it’s the hard-r alone, we can safely assume there’s something more racist going on.

5. While it’s unclear to me what precisely Musk knew about this specific payment at the time he made his decision, it’s at the least symptomatic of the deteriorated relationship. While this hardly makes the departing executives cartoon villains, there is a seemingly valid legal question here, and it’s good that we have impartial courts to weigh and decide appropriately.

6. There’s also a separate question of whether Musk signed this section of the merger agreement at all, though I haven’t come across any further updates on that one.