At the end of long saga well-covered in hilarious fashion by Matt Levine, Elon Musk has purchased Twitter.

He then began doing things.

One of them was to tweet ‘fresh baked bread and pastries are some of the great joys of life.’ On that I hope we can all agree.

His other decisions were less obviously great. More of a mixed bag.

The response by many news sources and also individuals has been that everything is The Worst and a giant dumpster fire and what he has done will destroy Twitter, and is so much worse than we expected, oh no.

While it is true that the details have not exactly covered Musk in glory, it all also seems like the kind of thing those same sources would say no matter what. One can assume that everything that happens is being intentionally framed to make it look as bad as possible for Twitter and Musk. Any blame that can be assigned will be, etc.

While looking at what is happening, one can also usefully look at the coverage of what is happening. Notice how media sources describe events. That tells you how they describe events in these spots, and how you should expect them to in the future.

So I decided to gather up the things I’ve seen since Elon bought the place, and my thoughts on them, and try to put them into a semi-coherent format. Here you go.

Oh Elon

One of Musk’s first actions was to fire a bunch of top Twitter executives ‘for cause’ which if it stuck (which it won’t) would allow him to avoid tens of millions in severance payments. And by some weird coincidence Musk finished the transaction exactly in time to fire a bunch of Twitter employees right before a bunch of their stock options trigger. If anyone involved was surprised by these moves, that is on them.

By now, it should be clear that Musk has zero interest in honoring laws or norms or contracts when dealing with people unless he likes them, and that his approach to the law is ‘make me.’

It also should be clear Musk has zero interest in stopping, before showing zero interest in honoring norms or his contracts or laws, to ask what would happen next, or wonder whether he will inevitably lose the resulting lawsuits.

He did not ‘think things through’ when trying to get out of buying Twitter. He is not about to start now. That is not how this type of thinking works.

People are rightly drawing parallels to Trump’s ‘blatantly violate contracts and no-pay people all the time and dare them to sue you’ except they think that ‘do it to rich people who will obviously sue you’ makes it somehow different. Trump also gets sued all the time and has for decades. Sometimes it is a surprise – he thought this person would not sue and they sued anyway. Sometimes it is very much not a surprise. Doesn’t matter. Once you have the ‘do it anyway’ algorithm running, it does not get a filter for ‘they will sue you and win in this case.’

I have known about this phenomenon for a long time because my father once worked for such a person, who would never pay their bills because what are you going to do, sue me? Worst case you force me to pay the bill. I have since encountered others. It is unfortunate that such systems often succeed in business. It is clear that they often do.

Firing the Rest of the Staff

Elon wants to reduce Twitter’s head count a lot. This has highlighted that a lot of SV/VC types have the thesis that headcounts in tech companies are drastically too high. For example, here’s Paul Graham:

A common response is that this violates the Efficient Market Hypothesis or other versions of ‘corporations wouldn’t be systematically making big mistakes.’ Which to me is Obvious Nonsense, of course they would if there are dynamics leading to that and it is the default path. That doesn’t make it easy to do something about it, and get rid of the 50% or 75% (or 90%?!) you don’t need while the rest includes everyone you do need.

The thing about employees that are worthwhile for a tech company (or anyone who doesn’t have a lot of ‘on the line’ style workers) is they are almost all super duper valuable, and not having one that you need is super duper terrible. There are real costs to hiring people who aren’t worthwhile, the drag effect is real, yet it is the easier and safer mistake to be making.

So how do you fix this problem once you know you have it?

Presumably you fire half the staff, although not the originally rumored 75%. As Musk took action to decide which half to fire, the process was described as ‘Trump-style management’ in the press.

“It’s like Twitter’s culture has been completely turned inside out overnight,” one employee said. “Mass trauma event over here.”

“We’re all working for the Trump White House,” the worker said, comparing the atmosphere to Donald Trump’s administration, where tweets from the president announcing policies that hadn’t been discussed internally could come at any time.

One Blind post from a Twitter worker, viewed by The Post on Wednesday, said simply, “This level of silent treatment is totally unprofessional.” Another Twitter employee replied, “It’s not silent treatment it is psychological warfare.”

You could, if you wanted to, report it like this, I suppose.

This happened on Friday. This is how they did it.

I agree that an eight hour window of ‘wait for it’ is absurd (also, shout out to an email address literally called ‘peoplequestions’). Also, I notice the thing where you are checking two email accounts – if you get a personal email then You’re Fired, if you get a company email then you’re not. So you can either doom update your personal email or hope update your work email (or, if you want to be fired, you can do the opposite, some people understandably want the severance under these circumstances). I would be sending some update email to both addresses no matter which outcome.

Is this a crazy way to fire half of a distributed workforce? Again, I’d do my best to send every damn email at exactly 9:00:00, and I’d send them all to both emails. Otherwise? Yeah. Seems fine. What is the alternative? Everyone has a zoom meeting at some point during the day when they find out their fate, would that somehow be better? Everyone gets called into the office so half of them can be escorted out? In that scenario, if someone doesn’t show up but wasn’t going to get fired, do they get fired and someone else keeps their job instead?

There are rarely good ways to fire people who do not want to be fired. When there is a good firing, it is typically because the person involved either wants to be fired, or at least understands why it is happening and is fine with it. Firing half your staff all at once, when the staff already hates the new management? What were better options?

Musk is famously against working from home, and the pandemic is no longer a valid excuse. Yes, people are now used to it and want to keep doing it. Musk is happy to have those people quit if they care so much.

Firing decisions for engineers are claimed to have been based largely on code commitments. I am not convinced this is dumb, or at least as dumb as it seems at first glance. You don’t know the internal politics, everyone involved is plausibly hates you and is going to be lying to you in various ways to save their own skin and those of their allies. This might be the cleanest available data set.

There is a clear issue where those working on harder or more complex problems, or who write more efficient code, will have smaller code commits. So if it really is pure ‘count the lines of code’ rather than looking at the code, it is definitely not ideal, especially if relied on too heavily. Whether it is being taken too far, given plausible alternatives, I don’t know.

There is much mocking going around that over the weekend some of the thousand plus employees that got fired on Friday got asked to come back. Some because of mistakes, some because ‘they are necessary to build the features Musk wants.’ I am confident that there will always be quick regrets in some cases when firing this many engineers even if it is not this rapid. In which case, yes, eat some crow and hope they come back. Much better to admit one is wrong quickly in those cases than attempt to save face, especially with so many eyes looking so intently to destroy one’s face. Bloomberg says it was ‘dozens’ of employees total, or about 1% of those fired. That seems mostly fine if it stops there, and the rapid reversal is better than waiting.

There is a report that employees are being encouraged to build something and show it off to Musk.

Presenting the first replies, also the two genders.

Note that there is no contradiction between these two reactions.

I think this is good. Twitter already works and is stuck in a rut.Twitter is a place where there are lots of potential new features that could add (or subtract) a lot of value. Those features are mostly much more about ‘figure out what the right feature is’ rather than ‘this is a hard engineering problem.’

Having many employees throw stuff at the wall seems like a good fit to me. It gives Musk data. He learns who is good at such things and who is bad. He learns who is willing to do such things and who is not. I checked with a friend who I know to get good work out of engineers, and he too thought this was a good idea.

The part where staff are being asked to work an 84-hour week? Not as good. If sustained over time and the plural is appropriate here, quite bad. Elon works insane hours, and part of his formula (based on what I’ve read about his other projects) seems to be to think everyone else should also work insane hours.

Moderation and the Dreaded Algorithm

Headlines across papers said that accounts flooded Twitter with hate speech the moment Musk took over.

For now, Twitter’s head of trust and safety continues to swear that there were few layoffs in that division, that nothing has changed and the temporary surge (see below) has been dealt with without issue, that everything is operating normally, except that there were some necessary access restrictions due to pending layoffs that required some temporary triage.

They also insist that none of the moderation or safety rules have changed. I doubt they would lie about that. The biggest thing that has changed is who can be blamed for a given moderation decision, especially a decision not to act. Of course, that does not mean that they aren’t also doing additional things that they feel like doing that aren’t in the rules, or that they didn’t previously do things that they felt like doing that weren’t in the rules, or change the rules whenever they felt like it to satisfy the demands of the Current Thing. Anyway.

There is certainly a lot of assuming something very not new is new. Here’s an example from Jack, who founded Twitter.

This was the New York Times right after the takeover.

This explanation made sense to me.

This had little to do with impacting anyone’s user experience or actually spreading hate speech. The magnitude of all this was essentially zero. What this was, instead, was a scientific experiment by those involved to see if the effective rules had changed.

Twitter safety claims it was all dealt with and the accounts involved were banned.

Eric Weinstein thinks something did change, and it’s wonderful.

Derek Thompson thinks it is obvious things changed, except it’s worse.

The Kyrie Irving anti-Semitism thing (and the Ye anti-Semitism thing) are unfortunate timing. I do not see the link between them and Musk buying Twitter, beyond the decision on whether to whistle and pretend it isn’t happening. The last (checks notes) (sorry, last track) number of times someone (regardless of their politics, which varies) has said something anti-Semitic people have done a lot of whistling. There is kind of a history here.

Whereas now the people blaming Musk for things say Twitter is ‘ignoring anti-Semitism’ by letting it trend, or not Doing More about it, or something? I don’t even know which direction they object in here. My inclination is to make trending topics (if it must exist at all) a pure algorithmic decision. Barring that, I’d want to do the opposite of what those making choices have been doing for years, and avoid culture war and cesspool topics in general.

(Also, who clicks on the trending box? Don’t do it, Derek. It’s a trap!)

The timeline weak-tie follow change could be nothing. It could also be the same thing Eric is talking about, where the algorithm is being less aggressively interventionalist. I don’t know, because I don’t use the algorithm, why would anyone do that? Hello, Tweetdeck exists and also lists exist.

For similar reasons I can’t speak to ratio of ads or their quality. With everyone complaining about advertisers pulling their ads it would be odd for ads to be increasing in frequency. Can’t have it both ways.

The main character thing is legit. For the moment, Twitter is partly about Musk buying Twitter. Nothing to be done about that except wait. Should not be too long.

The weirdo replies thing I have not noticed. Could be nothing. Could be crypto bros testing boundaries. I doubt it is any change to the underlying system.

It does make sense that certain types ‘feel enabled’ in this new world. Some of that is good. Some of that will not be as good.

Others say they are seeing changes in follower flow already.

Yoel Roth is the head of safety and integrity, link goes to his feed. I quote him in a few other places. All of it seems to be standard ‘we are doing the same job we would have always done.’

Venkatesh Rao is deeply skeptical that Elon can succeed with Twitter. He has consulted for Tesla, and thinks that while its management style is effective for Tesla it is not a good fit for Twitter. He thinks the core issues are hard to solve, and he thinks a real exodus is plausible.

Yishan, who used to run Reddit, gives his perspective on the Twitter moderation problem. A lot of fascinating stuff here.

A central theme here is that moderation is mostly about behavior rather than content. He suggests that spam can be identified by its behavior patterns without any need to understand the messages or language, by its low complexity and repetitive content, and extends the principle to other areas including flame wars and protracted arguments. This insight seems important and mostly true.

He starts with spam. In his model, there is no principled reason to ban or restrict spam. We all agree to do it because in practice there is no choice. Failure to control spam ruins the platform.

I agree that it is behavior rather than topics being censored here.

However, I do think there is a principled reason to go after spam, which is that spam is asymmetrical theft of attention. I would disagree with the response that ‘that is simply agreeing that the outcomes are bad.’ In my model, rules when done right are largely about incentive design. If I am going to allow a platform and those on it to consume my attention, that will not work if it is too low cost to waste that attention.

More bluntly: ‘Don’t waste my time’ is not an unprincipled rule, nor is forcing those who seek my time to first costly signal in some way.

I agree that wasting my time is not illegal (whether or not it should be), and that this proves ‘all legal speech treated exactly equally’ is a non-viable answer without major adjustments to our systems.

I do wonder about whether there are viable filtering systems, that use other methods, that have the side benefit of containing spam. If a generous white-list system was used for notifications and DMs and replies I saw – say, anyone who is followed/listed by at least one person I have followed/listed, and is followed/listed by that group more often than muted/blocked, or that I follow/list, or that has some other costly signal of legitimacy – I am confused how someone would use Twitter to spam me enough to much bother me.

In section two Yishan moves on to flame wars and protracted arguments that saturate and ruin platforms.

I don’t think this is necessary for the thesis that you need to worry about such dynamics. My model is MIB-style, that an individual person is not so easy to get worked up unless you happen to get lucky, however people as a group are easy because you will get lucky somewhere and then it will build on itself.

In Yishan’s model, certain patterns are bad for platforms and if left unchecked spiral out of control, so when you see those patterns they need to be curtailed, and moderation has nothing to do with the information content itself.

Except, in practice, that’s not actually true? You face tons of pressure, both cultural and legal, to indeed moderate on the basis of content? And this clearly happens a lot?

That is assuming you yourself want to be in the Yishan ‘all that matters are the platform flow dynamics’ position and these annoying outsiders care about whether the discussion is about something they dislike. You, also, might care.

Yishan also recommends (and is an investor in) the app Block Party for Twitter, curious what people think of it and how best to use it if it’s worth using. So far, I haven’t encountered problems requiring such a service. Matt Yglesias prefers the seemingly similar MegaBlock. So far I have not felt the need. I have felt the temptation.

As a counterpoint to Yishan’s ‘the content of the content does not matter’ position, here’s the person who pointed out Yishan’s thread to me first doing a ‘speed run of moderation,’ in which the content of the things very much does matter quite a lot. In particular, that various countries and especially the EU and Germany will tell you that whatever speech they do not like (they call this category ‘hate’ speech) is illegal and you always have to find it and take it down, and also this will open up other countries to say you can’t make fun of their prime ministers, also copyright and child porn and so on and so forth. Also spam.

I get that everyone is mocking Musk for his being in favor of free speech by pointing out that most of the world is against free speech, and for saying ‘free speech has consequences some people don’t like.’ The thing is, Musk is not starting this speed run from the first level. He is on at least level 19 already, because he bought the game machine for $44 billion dollars and got handed the controller.

One thing this points out is that it is inevitable that it will be impossible to comply with all the laws. This is already true of regular people trying to live normal lives, so it is not shocking that it is true of an international social network. It is more shocking that it hasn’t been more of an issue yet. Germany says ‘hate speech’ down quickly or else, Texas says ‘no viewpoint discrimination and no discrimination against Texas’ or else, and you still have people making fun of Moti’s beard. Now what?

Meanwhile, there are some content moderation decisions on Twitter that have absolutely very much been purely about the partisan or political or ‘scientific’ content of what was posted. I expect Musk’s Twitter to make very different decisions, on the margins, about those cases. I do not expect this to make Twitter fall apart, unless this causes blue tribe voices to successfully convince enough blue tribe people to flee on the basis of this change.

Also, I expect a lot of things that have nothing to do with Musk to now be loudly blamed on Musk, especially moderation decisions, and for it to be hilarious. This has started with the ‘fact checking’ Birdwatch operation, which is high variance. Sometimes it is super awesome, as described here. For example

The ones directly aiming at Biden are great too. Here’s a response to the White House bragging about Social Security having an unusually high cost of living adjustment (also known as ‘having high inflation’).

The problem, of course, is that when you give The People this opportunity to do what could be described as Wikipedia-style fact checking and edit wars on content notes on Tweets, some of them will not have truth as their top priority. The design always had this flaw, which will now of course be blamed on Musk. The idea (I think?) succeeds if good actors are sufficiently more willing to participate than bad actors, fails otherwise. One could argue that bad actors are emboldened right now, on all sides – bad actors who want to punish Musk/Twitter and make it look bad or at least no longer care about the rules, and bad actors who think they now have a moment due to their new savior, or something.

Perhaps the principle here is that civil society and the town square depend on there not being a lot of people determined to attempt to burn down the town square. So in some sense it is Musk’s fault, at least for now. Hopefully the process survives as something good.

I do think one can come up with ways to make the Birdwatch principle succeed despite enemy action, if one knows that is what one must do and the bulk of people are not hostile. If we can solve the Byzantine Generals problem we can solve this.

Are there also other ways to do filtering that could complement all these tactics?

Musk thinks there are and wants to focus on a different approach.

First He Came For the Blue Checks

Blue checks for the people? As in the people who are willing to pay? Would you pay?

The polling part of this is one of the things Musk should totally be doing. If you have a 100k+ vote poll on what people would pay for on your platform, you should notice, and you should give it positive reinforcement.

Assuming little or no cost to verification, and interpreting this naively despite the fact that This Is Not How Any of This Works (so please do not respond to inform me that This Is Not How Any of This Works because I know that already), charging $5/month would earn $1.10/voter/month, charging $10/month or $15/month would only earn $0.90/voter/month, so the sweet spot is likely under $10. Which Musk likely noticed.

Perhaps this would decline as blue checks became more common. Perhaps it would go up, because not having a blue check would be a sign you couldn’t be trusted. Perhaps you could give people the option ‘DMs are open if you have a blue check’ or ‘see all notifications from blue checks’ and now that is safe from spam. Whereas if you don’t care, This Website Is Still Free.

He’s actually going to do it at $8/month. Also, he’s threatening to fire the engineers if the feature isn’t ready within a week.

Musk started out saying he would be charging $20/month and including some extra features. Would ordinarily seem harsh, but if it comes with a promise to not fire them if they do get it ready, and lots of people are getting fired right about now, I’d rather have an engineering test like this than be subject to the whims of someone else. Then again, some of them got fired for cause anyway, in a true ‘if you do everything I say then maybe I will let you go’ moment. Then, in response to feedback, he cut to $8/month.

Those reactions were all over the place. Mostly very not positive. Here’s a take.

Here’s an even blunter take.


No, this is not the way to defeat bots & trolls unless the implementation is very different, and also I really, really wouldn’t want to be talking price with Steven King even if I didn’t have hundreds of billions of dollars.

Here’s AOC’s take.

I got $8 of value from that exchange.

I will also note that if Elon Musk promised to read the tweets, let alone respond to them, $8 sounds like a bargain. Oh boy do we both have thoughts. So many thoughts.

AOC is now claiming that her account ‘got bricked’ for upsetting Elon.

Should have paid the $8.

Here’s equating willingness to pay for a service with being a cartoon villain, and think the same for anyone advocating for freedom of speech or not catering to elite ideas of acceptable and unacceptable voice.

Well, that perspective would be why everything is based on advertisements. Did you know people used to pay for the things they found useful, in exchange for using them or making them more useful?

We are about to get a very interesting clash of world views, then, between:

  1. Willingness to pay is a costly signal worth paying attention to, and also indicates that you find something of value. Allocate scarce resources via price.
  2. Willingness to pay means you’re running a scam (since nothing else is ever worth paying for) and also means you are evil (since only evil people pay for things.)

This second perspective believes incentives are bad, commerce and trade are always up to no good.

Even more than that, in this model: nothing commercial can hope to exist except scams.

Presumably this predicts that regular actual users won’t be willing to pay?

Here’s a different take.

Here’s an alternate ‘you raise? I reraise!’

I love the ambiguity. In order to get value when you pay $1k/month, you would either have to be trying to say ‘I am rich and I am blowing $1k/month on this gold check mark’ which some people would totally do, or you would want ambiguity, so people did not know if you were paying or got the mark for free. And if you’re not obviously a big enough celebrity, and are offered a free gold star, you have to worry people will think you bought it. Fascinating stuff, I want this to happen, prediction markets in who is paying for it await, love it.

Here’s both concern over equity and for approval of legible financial transactions.

New EA cause area just dropped, verifying worthwhile Twitter accounts!

Here’s one I think is clearly wrong:

Here’s another Matthew Yglesias take, where the NYT and media generally doing oppositional tech journalism while systematically being given free blue checks caused Musk to think that the journalists would pay for the blue check. That seems wrong to me in terms of causation and also because they will totally pay for the blue check. Journalists need the blue check to do their job, is there any world where it is not worth the charged subscription fee to them to get sources trusting they are who they say they are faster?

Twitter provides lots of value, and the blue checkmark also will provide value. Either the blue checkmark is worth a lot more than $20 to large accounts, or it is worth far less than $0 to those accounts. The question is, which is it?

Thus, if you wouldn’t pay $20 but unlike Steven King you would pay $8, what is the chance you are making good decisions?

Kelsey’s poll here suggests substantial willingness to pay even without goodies attached.

What if we threw in some goodies?

He also endorsed this description:

Interesting value propositions.

Half as many ads is translated by my brain to ‘twice as many ads for everyone else.’

Posting long video and audio is not an obvious advantage given the Twitter medium, especially given you can always link to YouTube. Forced brevity is kind of the point.

Priority in replies, mentions and search is where it gets interesting. Blue checks will get an algorithmic advantage.

Getting past paywalls seems great. I love this idea. Instead of giving subscriptions to lots of different newspapers, I pay Twitter once, and they divide some of this money among the various sites where I want to be able to follow links without relying on them as news sources.

Here’s another new value proposition, which may or may not get linked to the $8.

The whole point of Twitter is that brevity is the soul of wit, so ‘attach long-form texts’ is not obviously something that is better off being made easier or harder. It will always be possible because links and screenshots, and yes I find inability to copy/paste off them super annoying, yet I am torn on bug versus feature here. My guess is you do want this to be slightly better than it is now, but not too much better.

Given the stated goal is defeating spam, that would imply the goal is to get as many ‘real’ posting users as possible to blue check over and above maximizing revenue. In this model, you charge money to differentiate people from bots, not for the money.

This is the view that verification should be the default. The problem is the incentives of charging money are backwards. Verification and blue checks are things you want to see more of. Anonymity is something you want to see less of. There are indeed good reasons one would not want to ban anonymity, yet the question must be asked: As lord of the realm, which of these two things does it make sense to tax?

The issue is that you need the payment to serve as (and pay for) verification, to give the person something to lose or protect, and to guard against spam accounts. So even though you want more of it, you have to charge.

Which in turn means offering a bunch of features that humans value and bots don’t.

What are some other features that could be included?

One option is to make prioritizations and special rules opt-in. I could choose to say ‘DMs open, or you can reply to this (or I will see the reply to this) or what not, either if you are on one of my lists, I follow you or you have a check mark.’ If it’s opt-in all of that seems good to me.

I am not convinced Musk’s plan is correct. I do see it as eminently reasonable.

In particular, the talk of ‘if you don’t pay Twitter goes to hell for you’ seems wrong.

The takes would not be complete without the take that this will End Democracy, so here you go via video clip from MSNBC.

The two nightmare scenarios laid out here are:

  1. Twitter’s website is a house of cards on top of a bunch of twigs and half its staff it will fall over and die when these oh-so-important elections happen.
  2. Musk rolls out his new $8 blue checks on Monday. Then anyone can buy one. Some people do. Then, on Tuesday, they (gasp) impersonate anyone they want. The only thing stopping them was the blue check, and now it is gone, and suddenly getting election information, or getting it out there, will be a nightmare.

I love the vibe of ‘the midterm American elections are the most important thing in the world and failure to have Twitter during them would be some huge disaster.’

Taking the very Democratic-leaning Twitter and news media and having them focus on Musk instead of the midterms does not seem likely to help their chances here.

Musk responded to these alarms by agreeing to postpone the changes until after the election, so we will never know what might have been. That seems wise, things could likely use a few more days of seasoning anyway.

So I do think it’s plausible that Twitter’s website is terrible and at risk of serious problems. To me that doesn’t exactly say you shouldn’t fire half the staff. Instead (huge if true and all that) it sounds like they were not doing their jobs all that well. It hasn’t been down very often, so my guess is that it is not that close to falling over, and I would hope that the ‘don’t let the site fall over’ group is not losing as many members right away. Elon is an engineer, he presumably knows ‘website stays up’ matters.

On the impersonations, yes the blue checks are helpful with that. Yes, people suddenly having blue checks that aren’t The Real McCoy could be confusing, especially that first two days. Let’s presume you can get it instantly if you pay.

There are a lot of other clues that one can use in these spots. Are you following them? Are people you follow following them? Twitter tells you these things without any need to click. How many followers do they have? It really, really isn’t that easy to importantly fool you into thinking I am the real Taylor Swift or the Arizona Secretary of State or what not.

I do not know exactly what everyone is worried about with these impersonations. Are Twitter DMs being used to coordinate official vote counts? Will people get the wrong idea and then try to Stop the Count? Might some people spend a few hours thinking things are going better or worse for their ingroup than they were indeed going?

I do, however, know what one particular person is worried about with these impersonations. Accounts only want one thing, and it’s hilarious?

If this rule only applies to blue checks then it seems… kind of reasonable? If a blue check is supposed to mean this person is verified to be who they say they are, and they change who they say they are to start saying they are someone different, it stands to reason they should need to be reverified. That is what verified means.

Also this solves the problem of ‘I impersonate an election official’ because yeah, that’s who people go around impersonating on social media to do social engineering black ops, sure.

If you don’t have a blue check and go around impersonating blue checks for comedic purposes, of course, that is totally fair play and anyone including Elon Musk who dares mess with this joins the Stanford Club, my new (just coined) name for those who Hate Fun.

Which Elon kind of joins anyway. I’m against the no-not-clearly-identified-parody-accounts rule because the alternative is funnier and I want comedy to be fully legal on Twitter. A true master of the art form would welcome such actions as long as they were parodies rather than phishing attempts. This is one more reason to put me in charge of Twitter. There would be tons of Zvi impersonation accounts and I would be totally here for it. I might have a rule saying that the Twitter handle in question has to be a hint (e.g. no @EolnMusk or @RealElonMusk or what not, but if you are using @ChrisWarcraft or @KathyGriffin or @SarahSilverman or @PopeHat then people seem like they should be on sufficient notice.)

If Elon actually permanently kicks out Sarah Silverman or other comedians over this, that would be a clear and important unforced error (and not the first one, which would be his reaction to the Paul Pelosi incident, he will need to learn to stop doing things like that).

It would make perfect sense, and fit the rest of his personality, for Elon Musk to not be able to take a joke, for him to be in favor of Free Speech except when someone is mean to him. He has, shall we say, a history, and even without that history I would suspect this anyway. The internet is not about to let him Get Away With This, and this seems like his point of greatest potential vulnerability.

If the real rule of Twitter was ‘free speech except when you get into a beef with Elon Musk at which point he might suspend or ban you, parody must be clearly labeled if and only if it is targeting Elon Musk’ and Elon keeps his beefs limited to himself (including Tesla and SpaceX but excluding anything political and so on) then how close is that to free speech? In principle it is deeply offensive. In practice it is still deeply offensive yet this is still mostly due to the principles involved. Counterpoint is that he kind of paid $44 billion dollars for the right to do that.

How Does This Play Out?

People are imagining a lot of strange outcomes of the blue check changes.

My presumption is that this reinforces the existing hierarchy of content sources when Twitter is debating what to show you, either through Tweets, replies, DMs or searches, which is something like:

  1. Accounts you follow.
  2. Accounts endorsed by your followers.
  3. Blue checks.
  4. Everyone else.

My main objection to this is that Twitter is leaning too much on following versus listing, and otherwise forcing you to see someone’s Tweets (or mute them) in order to get what you want in terms of sorting here.

Right now, the blue check is limited to a few hundred thousand accounts. By expanding that a lot, the theory goes, we can discriminate harder against and be more suspicious about those without check marks. Spam bots will have a much harder time. It is unclear what happens to the ‘more human’ versions of spam that are not trivially obviously pure spam, usually crypto schemes at least in my experience. They can and will pay for the check mark if allowed to do so, so the system would need to be willing to take that check mark away quickly if they misbehave, and not let them get a new one so easily. At which point, it seems highly useful there. My friends may not want to pay monthly, but if you are running ‘legit’ crypto business on Twitter then pay up.

None of this is a complete solution. Still seems like it would help. What is the issue?

This class of response is where I get confused. Why will those who don’t pay get flooded with spam? Is the theory that spammers will pay for blue check marks, so they can then spam everyone without blue check marks? How does that even work?

If I have no checkmark, my experience is worse in that I have lower priority. I will get interacted with less often, because more people do have check marks than before and the check mark is being upweighted as a metric. That sucks for me.

However, in terms of what information is displayed to me, my lack of check mark seems irrelevant aside from the line about half as many ads.

If all I am doing is consuming content, what is the problem? What (again, aside from ads which seem mostly harmless) changes? This is not different than if YouTube started charging people for blue checks on channels, if you don’t have a channel and are watching content, there is no reason to much worry.

One other potential issue is that people with protected Tweets might not approve your follow request if they don’t know you, I guess? Seems minor.

If you are producing content (which includes ‘talking to people’) then yes you take a priority hit, and your DMs and replies will be less visible. Which, again, sucks to be you. Doesn’t seem like a hellscape. If you are talking to your friends and people who follow you, very little should change and you should have modestly better protections. I anticipate continuing to at least read every DM and reply that I get unless volume goes up quite a lot.

What Comes Next?

Twitter was in trouble before Musk got there.

Advertisers for now are pulling out, because a story is being told about how Twitter is no longer a safe association for brands. Because there is a clear effort by various elites to go after Twitter, whether or not all coordination is involved is implicit, both in general and in particular via targeting its advertising revenue.

The language here could not be clearer. Musk is Bad Monkey, do not let us see you associating with Bad Monkey or you will also be Bad Monkey. This is not because of anything in particular Twitter did or anything like that, other than the lack of Musk committing to doing things. It started before Musk had any opportunity to Do Things. It is because Musk is in charge now, they don’t control Musk, they worry about what Musk might do in the future. And also they hate Musk. Have for a long time.

Mike is exactly right. You cannot appease activists. This is not a thing. They do not exist to be appeased, they exist to create more activists and for you to try to appease them and for them to then say no so they can keep on doing more activism. When you try to appease activists, and accept the jurisdiction of the court, all they do is smell blood in the water. Washington Post has them citing the layoffs in and of themselves. Here are specific claims about specific well-funded activists making the demand that advertisers pull out if Twitter ‘if [Musk] follows through on his plans to undermine brand safety and community standards including gutting content moderation.’ Which I am rather confident is intended to mean ‘lets people like us determine safety and community standards including content moderation’ plus a demand to not fire anyone anywhere at the company, or at least in the departments in question. Claims that Musk ‘has changed nothing’ or that the remaining team can handle things are irrelevant here, whether or not they are true. The point is that if these people tell Musk to jump and he says ‘how high?’ you can be confident he is making a weed joke.

Presumably Musk mostly knew this, and he claims to know it now. He didn’t actually ‘do everything we could to appease the activists.’ What he did was to not change the underlying moderation policies right away. Surely he is not so foolish as to ever think that was what this was about.

It is not only activists. For example, the Anti-Defamation League has called on advertisers to leave.

Given I don’t see reports of ‘whoa look at all these ads we are not seeing’ I presume this means prices for Twitter ads have dropped. Or will. Market, do your thing.

Here’s that word ‘appease’ again:

You can appease advertisers. I’m sure he did make reasonable attempts to appease advertisers, except for his first move being to try and shift away from an advertising to a subscription model. And also except for not doing these things until after he bought Twitter, because he spent all the time before that trying to avoid buying Twitter rather than building it up. That was indeed a mistake.

One explanation is ‘Twitter could not give any assurances about “safety,” because of Musk, so no one knew if it would be a bad look, so no one was willing to buy anything up front, which is very bad for Twitter, and it’s a little late for that now.’ It’s true that he was very much Not Helping matters on this front yet I don’t know what exactly he should have said instead that would have made that much difference.

The thing is, the advertisers do not care about any of the things Musk can offer them, because what they care about is the perception that Twitter Bad Now, or Twitter Risky, or Twitter Gets Me Targeted If I Advertise There.

Which again is Because Musk not because of particular actions, so assurances about his actions are not going to fix it. Or at least, nothing compatible with being Musk.

Thus, the list of companies pausing advertising on Twitter contains some big names.

(I doubt anyone reading this needs to to hear this, but the whole boycott thing is dumb. If someone previously bought advertising and now isn’t buying it, I am not going to worry that I bought a box of Oreo cookies, which incidentally are the Correct Order out of all the things listed above. Stop it.)

(I was curious, so I loaded my actual Twitter timeline (pro tip: never do this) to see what ads it would show me. None. Same with when I look at lists. None. And now that I think of it, I look at lists and posts on the site often and can’t remember the last time I’ve seen an ad. Huh.)

Note that the watchword here is ‘pause.’

It makes perfect sense to pause advertisements on Twitter while it is the Current Thing and activists might get mad at you and for all you know the place suddenly really is a hellhole or will successfully be framed as one. The advertising opportunities are in plenty of places and this one will still be there in a month.

Then, quietly, you can perhaps resume, who even cares. Can the heat be sustained over time. Maybe it can, maybe it can’t. A good place to advertise is a good place to advertise. Are real people going to be angry that a company is buying ads on Twitter? Is this actually going to matter? I mean, come on, no, of course not.

Musk is out there asking people for their ideas and getting into conversations acting like he has no idea what to do. Seems good to me. Mock this as ‘he has no idea what to do’ all you want. How do you think people figure out what to do?

If your answer is ‘lol he paid $44 billion without a plan and is asking us for one’ then, all right, sure, it’s not not funny. It still seems like the right approach. If there is One Weird Trick then this makes it more likely to be found. If there isn’t, this makes it more likely that an array of things will be found.

He then ignores those telling him what to do when he thinks they’re wrong. Love that.

Paul Graham’s theory is that little will change and that we can’t know what will happen.

As with all other reactions and predictions (including mine!), this tells us a lot about how to model the source, in this case Paul Graham. The idea that we can’t know what is going to happen, the idea that underlying stuff matters, the story that what people say and what elites decide to approve of and fund is not worth worrying about? That’s his story, more generally, and he’s sticking to it.

Know Your Business and Your Budget

Twitter at core has two distinct problems. You need to build a great product that people use, and you need to pay for it. Musk’s purchase made the second problem much worse and much more urgent. Also general economic conditions have created a vibe within tech (and elsewhere but especially tech) that makes fundraising extremely difficult and makes people unwilling to lose money short term like they used to. Talib here is one of many expecting persistent unemployment and lower salaries and such across the board for software engineers. I predict that the demise of abundant high tech salaries for competent engineers, especially the 10x ones, is greatly exaggerated. If I was an incompetent engineer I would however be worried.

Musk is acting like he needs to solve both these problems simultaneously. That Twitter is bleeding money and in deep trouble, that can be solved only by some combination of (1) drastically cutting expenses and (2) finding new revenue sources, shoring up advertising or both.

Some of this is doubtless an excuse to run a tight ship and put people under pressure to get things done. It is standard VC/SV procedure to use the threat of running out of money to light fires under everyone and cause things to happen. People who say ‘oh there was no rush why is Musk moving so fast’ are the ones who think that is fine to take seven years to paint the curbs for a new bus lane because it is important to have the proper public comment periods.

One still must ask why Twitter is bleeding money. A lot of the answer is ‘Twitter is not making much off advertising.’ Yet the bulk (perhaps all) of the money Twitter is bleeding comes from the debt Musk took on to buy the place, combined with the fall off in advertising due to pressure placed upon them, either explicit and direct or implicit and automatic, combined with the uncertainty of what the platform will look like.

Before, it wasn’t a super business, but it wasn’t doing a ton of bleeding. When you compare the amount it is losing to how much Elon has, or how much he lost buying it, the bleeding is not all that big. Elon didn’t have to take on a bunch of debt here. The interest rates are high and the payments are a serious problem for the business. He chose to do so anyway.

Also, that debt would trade right now at a very steep discount due to changing conditions, such that the banks are not even going to try to sell it. If I was Elon Musk, I would strongly consider buying the debt back from the banks at market value. I consider the move ‘issue debt, then when the debt falls in value a lot buy the debt back’ a vastly underused move.

The point of all this is that Elon could be on tilt or be in ‘I am not willing to lose any more money on this’ mode, or he could be using that to get things done, or some combination thereof, but all of that is a choice. Twitter is only in financial danger in the medium term if he decides Twitter is in danger. If it could make much better decisions by not worrying about profitability that much for a bit, it seems like that would generate a lot more enterprise value than it would cost in losses.

Elon has happily gone around losing money and almost going bankrupt or needing bailouts at the last minute quite a lot. He is very familiar with the ancient art of bleeding money for years and years. I have to suspect this focus on revenue is not about the money, it is about using that to justify charging money as a way to fight spam and improve the platform experience via favoring paid users, and as a way to exert pressure to create a worthy value proposition.

Do I think favoring those who pay will improve the Twitter experience? Yes, as long as it is not too aggressive and has proper safeguards. Too aggressive would be things like ‘paid people have priority over the people you follow.’ Safeguards means that those who abuse this priority rapidly lose it.

The revenue is nice as well, as will be the increased pressure to find new ways to enhance Twitter.

The danger is if Musk himself buys too much into the urgency of cash flow, and goes way too far in prioritizing running a profitable (or non-losing) business over building enterprise value. Or of course if he simply moves too fast and breaks too many or the wrong things.

Or if those aligned against him succeed in generating enough FUD and momentum to break Twitter’s network effects before it can all fade away (which would otherwise inevitably happen through a combination of time and things actually improving.) That would be the obvious failure mode.


My prediction in terms of the tech and UI and what can be done on the site is that the non-paid experience will mostly not get worse, it will get better for most users who chat with their friends and are otherwise passive consumers of content. This will be both because having some people pay will enhance the website’s ability to usefully filter content, and because improvements will be made some of which will be made available to all. I’m maybe 75% confident in this.

I predict the website stays functional. My market has substantial down time within a year at 16% and I would take the no side of that.

The possible exceptions will be those whose goals are things like ‘go viral’ or ‘interact meaningfully with big accounts that don’t know who you are.’ In which case, that will be $8. In terms of features, I predict that many new features will get restricted to those who pay, but few if any existing features beyond the blue check will not get hidden behind the paywall, and that the new features would otherwise have not existed at all – Twitter basically was in ‘never change, never ship anything’ mode for many years.

Of course Twitter mostly is not about the tech and UI, it is about the people and what they choose to say. That is what will determine whether Twitter gets better or worse. The tech matters insofar as it impacts who is there and it changes what people choose to say or get to see.

People are not going to go, en masse, to Mastodon in anything like its current incarnation. Mastodon is not shovel ready. Kai reports that the experts (medical and otherwise) he knows are largely creating accounts there as backups. The reports of regular people attempting to use it, and I do mean attempting, are hilarious. The people attempting to explain why using it is easy are even funnier. Also sad. I wish we lived in the world where such people were right. The other existing social networks are complements to Twitter rather than substitutes, and are not threats. I would put the probability of Mastodon mattering at less than 10%.

As for whether Musk will blow the whole place up or cause a mass exodus? I do think it is possible, there is a lot of optimization pressure trying to make that happen and he is on at least some amount of tilt and taking big risks. Some would say Musk has been on tilt since childhood, whether or not it has been working. I’d say something like 25% chance of a major decline in Twitter use in general over 2023.

Also, good luck with this one, also I sincerely hope this is not The Mission.

Good luck with that.

In the sense that Elon means this, I put the probability at epsilon. In the sense of when Twitter is properly used by someone who puts a lot of work into it and studies such problems very hard it can be the best and most accurate information source now. For regular people? This is not going to be a thing. I think there is about a 70% chance that I continue to think Twitter is, for me, the best source of information two years from now when the election rolls around, with about half of that time it getting actively better at it.

New Comment
12 comments, sorted by Click to highlight new comments since: Today at 8:03 PM

My best guess is that in Musk's model, purchasing Twitter only makes him the legal owner. On a practical level, he has to figure out how to take power at Twitter. He wants to be able to radically change Twitter, but Twitter's former staff mostly wanted to retain the status quo.

So regardless of the effect on short-term profitability, the right move for Musk was to create chaos. His power to reshape Twitter increases as whatever power networks that might have been capable of resisting his influence get thrashed by massive and unpredictable firings. He is creating an environment where he is in his element: small enough to understand, weak enough that employees know he's willing to suffer sharp short-term social and monetary costs in order to keep control, malleable enough to reshape at his will.

So read it all as a costly signal: "I am in charge, I can torch this company, I will burn your job arbitrarily just to show you I'm not afraid to do it, there is nothing you can do to humiliate me that will make me change course. Comply or be fired, and have no doubt that I will gleefully fire you."

I don't really know about this or about this type of stuff in general, but my instinct is to disagree with your guess due to Occam's Razor reasons.

Half as many ads.

And if $8 is too much, you're free to remain as the product. opposed to remaining as half the product, I guess?

I put a very low value on reducing ads by 50%, compared to reducing ads by 100%. The annoyance doesn't scale linearly.

Isn't the most accurate source of information about the world Wikipedia?

And it works fairly well?

Anyways, great post. Watching the pitchforks come out against this whole ordeal is kind of comical, and I am excited to see how this Culture Battle over Musk-owned private Twitter plays out. 

Something that people often miss is that the big 5 tech companies (FB, Google, Amazon, msft, Apple) have an extremely enviable position in not just the tech world, but the entire economy as a whole. Tons of rich people, of all stripes (not just equity billionaires like Musk) all want to join their ranks in some way or another.

Musk is one of the few people who actually can (at least, something more than getting hired as an executive). Musk has deep natsec government contracts via Spacex, AI from OpenAI, and user data and big industry from Tesla (which is also the world's biggest robotics company, which spices things up a little). Taking control of twitter to some sufficient extent would always have been difficult, since the platform might be dominated by murky botnets and unknown hackers due to the company's alleged failure to develop the same sort of cyber-sovereignty that the big 5 tech companies probably have.

So it's not about what Twitter is. It's about what Twitter's not, and what it could be.

I have a theory that this is about corporate tax rates - that Musk was motivated to assemble his coalition of investors, and that the American ones pitched in, because they were worried that the Democrats would eventually raise corporate tax rates significantly. Certainly, moving Twitter out of the sphere of Democrat-aligned institutions is also a blow against progressive curation of public communication channels, thereby impacting many other issues and trends; but one would expect economic issues to be decisive, among investors. 

I have another theory, that AI, and AI analysis of Twitter users, would be a big part of Musk's plans for Twitter - through this acquisition, he has also acquired a database of user profiles to rival anything that Google or Facebook possesses. Perhaps he will make Twitter into a rival of Facebook? A comparison with China's Sina Weibo might be in order here. 

What percentage of the money that Musk paid for Twitter do you think comes from other American investors?

Somehow I don't think that the opinion on other people were very driving for Musks plans.

through this acquisition, he has also acquired a database of user profiles to rival anything that Google or Facebook possesses

Given that Twitter doesn't manage target advertising well, I don't think that's the case. 

Musk's stated plan is to do what he wanted to do with back in the day. 

A comparison with China's Sina Weibo might be in order here. 

If you take Chinese examples Musk wants to follow in WeChat's footsteps. If Tesla manages to produce driverless rented cars, they will be likely booked via Twitter.

The media does have its biases but their reaction seems perfectly reasonable to me. Occam's razor suggests this is not only unorthodox, but shows extremely poor judgment. This demonstrates that (a) either Elon is actually NOT as smart he has been hyped to be, or (b) there's some ulterior motive, but these are long-tailed.

Typically when one joins a company, you don't do anything for X number of months and get the lay of the land. I'm inclined to believe this is not just a local minimum, but typically close to the optimal strategy for a human being (but not a superintelligence playing 5D chess). It's unlikely the case that he bought the company only months from bankruptcy. Everywhere in big tech is doing layoffs but not to this magnitude. Also, coming into an office and demanding people work twice as hard and completely change their schedules around, would not work in any company. No employee with a family would be able to switch that quickly. No sane employee would be willing to pivot like this. Also, why should they? They have leverage.

None of the methods described above are actually reasonable in a real company, like blanket layoffs by LoC. Yes we can discuss above the motivations, how maybe to first order (probably not even that) it gives an approximation, but it's not like it's that hard to do it better and more accurate than this. No, at the end of the day, he's either an idiot, or deliberately trying to destroy it either out of some kind of revenge, or maybe somehow in the view that this buys time for AI alignment :)

Here are my predictions:
* He will have trouble staffing the company and complain loudly about it with the tired "no one wants to work anymore"
* He will move the company to TX and hire from there at 1/2 the salary or so.
* The site will stabilize, though not improve in any meaningful way, but he will be lauded as a hero in red states.
* The move to TX will be intended to signal a shift away from Silicon Valley and have a small but measurable effect, but CA will remain the dominant hub.

For similar reasons I can’t speak to ratio of ads or their quality. With everyone complaining about advertisers pulling their ads it would be odd for ads to be increasing in frequency. Can’t have it both ways.

Seems perfectly possible for a market to simultaneously experience decreased demand and increased supply if exogenous forces make it so.

[+][comment deleted]1y10