Welcome to Decoder! This is Casey Newton, founder and editor of Platformer and cohost of the Hard Fork podcast. I’ve had a lot of fun guest-hosting a few episodes of Decoder while Nilay is out on parental leave this summer. If you listened to the last couple of Monday shows, you know I’ve been doing a series with founders who are focused on productivity.

This is my third and, sadly, last time joining the show during the break, but I’m very excited about this episode. Today I’m talking with Steph Ango, who is the CEO of Obsidian.

Obsidian is a note-taking and productivity app that fits into a similar “second brain” space to Notion, the CEO of which I interviewed here on Decoder last week. But Obsidian differentiates itself with a really unusual approach to its business. It still wants to be your entire personal knowledge base — to hold all your notes, links, files, and other information — but it works in a very different way.

Listen to Decoder, a show hosted by The Verge’s Nilay Patel about big ideas — and other problems. Subscribe here!

In Obsidian, files are Markdown-based, stored locally on your own devices, and completely free to use. You’ll hear Steph say that he doesn’t even know how many users Obsidian has or how sticky the software is, which is more or less unheard of among startups I cover.

Obsidian does charge a subscription fee for access to certain features, including cross-device sync, version history, and web publishing. But it’s still a model that feels decidedly old-fashioned for software that’s trying to keep up with the current world, and so I had to ask him about those decisions. 

Steph’s role as CEO is also unusual, because although Obsidian is still a very young, very small, and very flat organization, he’s actually not one of the founders. He joined in 2023, when cofounders Shida Li and Erica Xu brought him in based on his experience with his former startup, Lumi. He was also a huge Obsidian fan.

So I really wanted to ask him about that, too, because I suspected his answers to the big Decoder questions about organization and decision-making were going to be pretty unusual for a Decoder guest. 

And in one interesting twist, I asked Steph why, when so many of his competitors seem to be racing to stuff their productivity products with AI features, it didn’t seem like Obsidian was all too eager to follow suit. His answer, I thought, was pretty illuminating. 

Okay, Obsidian CEO Steph Ango. Here we go.

This interview has been lightly edited for length and clarity.

Steph Ango, you are the CEO of Obsidian. Welcome to Decoder.

Thanks, Casey. I’m glad to be here.

What is Obsidian? How does it work and who is it for?

If you really want to boil it down, Obsidian is a note-taking app. A lot of people use it for writing their thoughts down, journaling. A lot of people are students who like to track their progress through school, or do their research notes. There are authors, book writers, big fans of RPGs who love to use it. What makes Obsidian unique is that it sort of works like Wikipedia, in that the core unit is a link between your notes. So, if I was to write about my experience today in my journal, I’d say, “I was on the Decoder podcast with Casey,” and each time I mentioned something I might form a link out of “Decoder.”

It’s okay if that link is not pointing to anything yet, but later down the road, I might create a note for Decoder because I want to track some of the interesting interviews I’ve listened to on that podcast. So over time, your web of knowledge becomes greater and you have more nodes in your Obsidian. Fundamentally, when you open the app, it works a lot like Apple Notes, Evernote, Notion, or any other similar kind of app out there.

When you joined the company in 2023, you said, “I can’t overstate how life-changing Obsidian has been for me. It has fundamentally improved the way I think. I want to see what happens if more people gain that superpower.” What do you feel like is the superpower that Obsidian gave you, and why did no other products make you feel quite the same way?

I’ve been writing notes and journaling for over 20 years, and I’ve used a lot of different apps. This idea of thinking about the world like your own personal Wikipedia was really powerful. I had kind of kludged together a few different apps to make something that kind of worked that way.  A lot of Wiki-based software already existed, but most of it was designed around publishing a full Wiki to the web as opposed to using it for your own personal notes.

When it came out, the founders of Obsidian, Shida [Li] and Erica [Xu], had already put in the level of polish that was not there in the glued together prototype that I had. It was instantly something that made sense to me. It made sense also because the data is stored is in this very durable format that people can own, which is called Markdown. You get super fans of Markdown and… for people who’ve never heard of it before in the audience. I’m not sure who —

I think we probably have a lot of Markdown users in the Decoder audience, but for those who have maybe not seen it yet, how would you describe it?

Basically, the oldest file types we have, going back to the 1960s, are plain text files. Markdown takes the idea of a plain text file, which is just raw text, and allows you to add basic formatting. So, if you want some text to be bold, a heading, a table, or a list, it allows you to use simple characters like punctuation marks to indicate what’s going to be bold or italic, for example. 

The power here is that the data is stored in this very simple format. So, we have this view around your data that you can hold your data for a long time and you, or your kids’ kids, your legacy, whatever it is, will be able to read it 100 years from now. Maybe none of the notes matter at all, or maybe they’ll be curious. But what if that data could be preserved over the long term? We think that going back to some of the simplest formats that exist and giving you that control over your data is more likely to persist over time. So, that’s one of our philosophies.

Going back to your question, I was really excited about the principles coming together. To the point of how it made me think differently, I think once you have this concept of links and ideas that can be networked together, you can start to form more complicated, complex, or interesting thoughts than you otherwise could. I don’t know about other people, but I can only have two or three different ideas in my mind at once. But if you can start to create these little building blocks of ideas, you can combine them in interesting ways. Your ideas become these little Lego blocks that you can interchange and mix together, so you can start forming some interesting and complicated thoughts.

I would love to hear an example of when you felt like you were able to do that in Obsidian. You’ve written online about your note-taking practice. You described it for us a minute ago, talking about how you keep a daily journal and as new characters and ideas come in, you’ll link them and build them up over time.

I’m curious about that next step. Is there a time when all of that added up to let you make something you maybe wouldn’t have otherwise?

Well, there are the projects I’m interested in. Even though I’m the CEO of Obsidian, I see it as a means to an end. I like to write, so I write a lot of short essays on my blog. A lot of the essays came from marinating in Obsidian, where I can debate with myself whether an idea is good or not. I have this essay called “Pain Is Information.” I was going through a pretty tough time a few years back, and I read this book by [Haruki] Murakami, What I Talk About When I Talk About Running. It’s this autobiographical story about running, and I came across this quote… I’m trying to remember it on the fly, but it’s something like, “When you sign up for a marathon, you know that you’re signing up for pain.” “Pain is inevitable, suffering is optional.”

That idea was really interesting to me because I was going through a painful time. I was thinking about how touching your hand to a stove gives you a signal that [doing] that’s a bad idea, but that’s information. So, I was starting to think about information, pain, and the relationship between those things. You can see how these ideas are forming out of thin air, out of different inspirations that I’m going through. And what I like about Obsidian is it gives you this place to approach it in a very freeform way and connect different concepts you might be thinking about. 

For me it’s in this philosophical realm, but for other people it might be with biology or language learning. My partner speaks Chinese, and I’ve been wanting to learn Chinese, so I’m starting to bring together these different ideas. I’m into woodworking and I’m always learning new skills, new details, or new tools that I might want to use, and I’m doing research about that. All of those things can coexist inside of this digital place, which is kind of weird because you could have a connection between a woodworking tool, a city I went to in China, and the concept Murakami was describing, all those things are just ingredients in this soup, and you could start to come up with ideas that you just wouldn’t have thought of otherwise. Maybe that’s enlightening.

Something that a lot of people value about Obsidian and similar apps is that they can be engines for serendipity. You gather strings in the manner you just described, and then in the process of clicking back through your notes or using other tools inside the app, you revisit ideas and they spark new ones or you see connections that you might not have otherwise.

Definitely. I think that one of the benefits of this approach is that it’s quite freeform. What I’ve run into with other tools or other approaches, like a physical journal, is that you’re quite constrained by the fact that it’s just pieces of paper you flip through. It has the limitation of being a 2D surface. Or, a lot of apps use folders or tags, whereas here you don’t have to know what something is going to be about until later when the connections form. And you’re free to have 700 tabs open inside of Obsidian and be doing this crazy, It’s Always Sunny mind mapping with the red thread everywhere. It allows you to do that where something with a more top-down hierarchy would make it more difficult.

You’re not a co-founder of Obsidian. You were brought in as CEO in 2023. How did that come about, and what were you brought in to do?

The app came out in 2020, and I found out about it with the first version. It was right at the beginning of the pandemic, and I think all of us were going stir-crazy. There was this moment in time where a lot of interesting tools popped up because everybody was like, “What am I going to do with all this extra free time where I’m hanging out at home?” So, I started using it right away for the reasons I described before. It made sense, and Obsidian is super customizable, so you can make plugins, you can make themes, and you can modify it in really significant or small ways. I was just starting to put all these community things out there. I was running a different startup at the time, and I was just putting these things out there because I was making them for myself and people started using them.

Shida and Erica are amazing engineers who met at the University of Waterloo. They’re geniuses in terms of engineering and community management. What I was bringing to the table as a community member was a sense of design and product that they had a little less of. Because the Obsidian community is so strong, some of the things I was making were getting a lot of adoption, and I was collaborating with other people in the community. They found out about that and wanted to put a quote from me on their front page. 

So, we started chatting, and then we started talking about business. They started telling me about the business model and some of the challenges that they were dealing with. We just kept talking for a couple years, and I was using Obsidian all the time. It was just the main app that I had. When I sold my previous startup, Lumi, I started to think what would be my next thing. I was thinking about building something else or starting a new company, and I just was spending all my time in Obsidian using the app, and I realized I was having so much fun using this tool. 

So, I pitched to them, “What if I could come on board and help you guys?” At first, it took the shape of contract work, working with them as an advisor and working on the 1.0 release that had this new design that I built. Shida is an incredible engineer, one of the best I’ve ever worked with, and eventually, he just wanted to focus on that. It created this nice balance. We’re a really small team. We are seven full-time people, so there’s something nice about the balance of the different strengths we all have. Everyone can kind of do everything, but at the same time, each person has their core strengths. 

For me, it’s around design, consolidating the true essence of Obsidian and trying to communicate it out to the world. Then, because we’re only seven people, there are a lot of hats to be worn, whether it’s accounting, legal, all these are random things. Also, I had a lot of experience running startups, so I think that was helpful.

You mentioned the plugin ecosystem. It seems to me that plugins have been one of the main ways that Obsidian has grown, both in its feature set and in building features that have attracted new users. What was the origin of plugins and how have they fueled the company’s growth?

In a way, plugins are what allow us to stay small because there are so many capabilities that people want that are fairly narrow and will only be useful to 1 percent or less of our user base. You see this all the time with apps that have been around for a long time, where the feature set just keeps growing and growing. Then the app becomes bloated, slow, and hard to use because there’s just too much functionality in there. 

For new users, it becomes extremely confusing. I think initially, it was this defensive move against having to implement all these features, to basically say, “Here, you go do it.” And because it’s built on web technologies like JavaScript and CSS, a lot of people know how to build things for it because they know those languages. You don’t need to know Swift or be a cross-platform app developer to know how to make a plugin. You can make something really simple in a matter of minutes or hours depending on your level. 

So, I think the initial reasoning was that this will allow us to not build everything ourselves. But then, the creative things people come up with are always pushing the envelope of what our API should be able to support and how the platform can allow even more flexibility. I mean, out of a few thousand plugins that exist, only a small portion that really make sense to be in the core app. Some of them do end up becoming something we notice, like when 80 percent of the user base is relying on one plugin. 

For example, right now we’re working on this thing called Bases, which allows you to view your notes in a database-type format, and there are a number of plugins that do something like that in the Obsidian community. So, it’s a signal to us that this is actually really important and should be in the core app.

What are some other plugins that have just been really popular or took the app in unexpected but successful directions?

Some of the most popular ones are very simple. Somebody who’s on the team today, Tony Grosinger, wrote this plugin called Advanced Tables, which was just a way to simplify making tables in Obsidian. We’re talking about something very basic, but rows and columns were difficult to do earlier on, and if you’re someone who wants to live completely in the Markdown world, they’re kind of tricky to make. We ended up hiring Tony, and he built that functionality alongside another developer who — basically everyone we’ve hired or worked with was once a community plugin author or team developer. That makes it really easy for us to start bringing the right people who are passionate about Obsidian on board. 

There are plugins about theming, styling, changing fonts and colors. A lot of people enjoy that customizability and want to be able to make this journal space their own. There are a lot of plugins that help you integrate with other services. So, if you want your calendar in there or something like that, you can do that. There are integrations into a million different apps out there. If you want to be doing your tasks and to-do lists in Obsidian, there’s a whole bunch of plugins that help you with that. The cool thing about that is if you’re interested in Obsidian because you want to do world building for your RPG tabletop group, you can do that and you don’t have to have the entire calendar functionality inside of your Obsidian. You can just use the plugins that have to do with that.

Plugins were really the first thing that brought Obsidian to my attention. I’d been using Roam Research, which I do credit for inventing a lot of the current note-taking paradigm, but Obsidian just developed much faster thanks to plugins. I’m curious what you did to attract those first developers. Was it as simple as having a really good API that was available early on? What was it that the company did?

I think it’s a combination of things. One, like I mentioned, is that the languages and framework are very simple for anyone to use. Anyone who’s done any kind of web development would pretty much know how to build a plugin, so it’s very accessible to a lot of developers. 

The values of Obsidian, just as a pure note-taking tool, are very aligned with what developers like. So, a lot of developers use Obsidian as their note-taking app of choice because it’s private and it uses this Markdown format as the core way to write text. Because it’s so customizable, it attracts developers and then developers use it all day long. We have published APIs that pretty much allow you to do anything with the app, and there’s a lot of documentation. So, it’s that combination: a lot of developers are using it, it’s easy to make the plugins, and we added the API very early on. If you have that itch, you can scratch it very quickly. If Obsidian’s not working the way you want, you can change it very fast.

Recently, I interviewed Ivan Zhao of Notion for Decoder. You mentioned Notion is another product in this space that sometimes people might choose over Obsidian. It strikes me that while your products do some of the same things, they’re designed very differently. Notion is about pixel-perfect polish and beautiful interface elements. Obsidian, by default, can look a bit more like a terminal. You take notes in Markdown, and it has more of this DIY, almost hacker ethos. Is that intentional and do you think it affects the kinds of users you attract?

I think the fundamental difference between Obsidian and Notion is that Notion is a cloud service. It’s an app that interfaces primarily with software as a service-type cloud service. You have to either be in your browser or on an app, and you connect to a source of truth that’s in the cloud. With Obsidian, all your data is local. So, if you’re not online —  if you’re on a plane or something like that — you always have access to your data. That difference shapes a whole bunch of other things.

For example, it would be really hard to add plugins to Notion because it can’t make it easy to run arbitrary code on its cloud-based platform, whereas with Obsidian, it’s pretty easy. So, there’s this fundamental split that occurs because of the architecture. It’s the same with things like theming, design, and how much user interface customizability there is. I don’t know if this will make sense to anyone who’s listening, but when I was 11 or 12 years old, there was Winamp, Winamp 2 was coming out, and I was all about making themes and things for Winamp, which was a music player that you could customize.

It really whipped the llama’s ass. I have to say that.

Absolutely. I think there’s a bit of that flavor in Obsidian. Obsidian is quite popular with a lot of younger people, and I don’t know what it is. At that time, you have the energy and the desire to have control over your digital space, and Obsidian makes that easy. So in that sense, we’re a bit less prescriptive about what the interface should look like, even though we’re trying to make it a little bit more approachable to still retain infinite depth. It’s like you’re going to the beach: we want to make the shallow waters a little more accessible for people who are coming into it, but then you can swim as far as you like, as deep as you like, into the complexity of Obsidian. Finding the right balance between those two things is quite challenging, and it’s something we’re always working on.

As I mentioned, I used Roam, then I used Obsidian, and then I used Mem. Now I use something called Capacities. Obviously I have a huge problem. I’m working on it in therapy. But I’m curious about how sticky Obsidian is. It’s free to get started, but I imagine lots of people abandon their vaults after only creating a few free notes. What makes people leave and what makes people stay?

We actually don’t know how many users Obsidian has. We don’t know how sticky it is because we don’t have any analytics. It’s very privacy-oriented, so we don’t track anything about our users. We don’t know what suddenly causes someone to churn or whatever. We prefer not to track those things. Also, the data doesn’t have to be exported. If Obsidian went out of business someday, you would still have the app on your computer. Even if you chose not to use it, you don’t even have to launch the app. You don’t have to export anything. This is one of the big issues people have had with other tools that have either gone out of business or been acquired by private equity firms that start tightening the screws and increasing the pricing over time — you feel like you’re locked in and you can’t do anything about it.

With Obsidian, your data is there. I think the feelings of freedom and security are paradoxically quite sticky because even though you have all your data, you could just… Personally, I’m constantly editing my Obsidian files not in Obsidian. I will use code editors and other tools to do mass modifications to hundreds of files. You can run Python scripts on your data. You can kind of do anything because they’re just files at the end of the day.

Just briefly, every file that is created in Obsidian is a Markdown file that can be opened up in basically any text editor.

Yeah. So it’s not a database in the cloud. It’s not a database on your computer somewhere that other apps can’t access. It’s literally just a bunch of files that you can move and change with any app. So yes, it makes it a lot easier to leave the app, but it also paradoxically gives people comfort that they have that option at any time. 

We’re not trying to be Notion. Notion has raised hundreds of millions of dollars. I think it’s amazing, especially on the collaboration side. It has a lot of advantages that make the app better for certain things. We’re just a small team. Our focus is to keep making the tool better and stay small, as long as we’re making enough money to stay afloat.

We’re not trying to take over the world. We’re not trying to be the next Microsoft. That makes it a lot easier to make long-term decisions that we feel are better for ourselves or for our users. It’s the tool that we want to use all day long. So, it’s okay if people leave. And different people have different brains and different approaches to thinking, so maybe they should leave. Maybe that would be better for them.

Let me ask you about one challenge I’ve had using tools like this. I wrote a couple of years ago about how I had spent years linking and tagging all my notes, reviewing them on a pretty regular basis, and still not feeling like I was getting a ton of insights from that process.

I was worried that these tools can be a substitute for thinking rather than an enhancement because if you spend all day tending to your digital garden, you might not spend as much time just walking down the street and giving your brain the chance to breathe and ideate. Was I just using these tools incorrectly, or do you think that productivity tools can sometimes be counterproductive for people?

I think they can sometimes be designed with anti-patterns that are explicitly about that. Some companies have metrics that they track, like the number of active users, but how do they define active? So now they’re sending you notifications to remind you to come back to the app so that you do whatever the thing is. That’s part of the reason we’re not interested in having any of this data because we don’t really want to be tracking our users in this way or incentivizing them to create usage where none is needed. If you use Obsidian throughout the day and you have these random ideas you just want to drop inside of your daily note, you should be able to do that and not get sucked into this thing that’s trying to engage you. 

There are a lot of people who keep trying new apps and nothing sticks for them or they get caught up in the organization and beautification of their notes. I think that can happen in any app. I hear about this, in general, and I don’t know what the cause is. My sense is that it has nothing to do with the apps unless they’re literally trying to engagement-bait you in these weird ways. I think it’s sort of an affliction of the digital age. These things are so malleable that it scratches an itch that we have in our brain to optimize. For certain people who love solving puzzles or doing Sudoku, it’s kind of addictive in a way that might not be the most healthy. 

I’m not exactly sure how to solve that. The way I try to address that, at least in my personal life, is being very aware of how the business of many of these apps works and how they’re trying to capture your attention and time. So, I’m always disabling notifications for everything, trying to spend a lot of time walking in nature, doing woodworking, cooking, and other activities that I find restorative. Then, that makes my Obsidian time feel more rewarding, productive, and useful because — productive is not the right word. I just have things to write about. I have a life that I’m trying to dissect. “Oh, what happened today?” or “What problem am I trying to solve?” And if you don’t have those other things going on, then you don’t have something to write about, and you’re now in the space massaging something. This is probably more of a question for a psychologist to solve, but I do see it, and I don’t know what the answer is necessarily.

Well, let’s ask the Decoder questions. You’ve mentioned that you have seven full-time employees. How is Obsidian structured? That sounds like a pretty flat structure.

We have two people who are working on community related things full time — essentially customer service and plugin review. One of the ways that we scale is that we have very active communities on Discord, Reddit, and elsewhere. There’s a lot of user help, so users help other users, which is nice because it means that we don’t have to have as many customer service people on staff. 

Then, we have three full-time engineers, me, and then Erica, who works on marketing, community, and other things. I suppose I’m the only person who’s a trained designer, so I end up doing a lot of UX design, marketing related things, and our web-facing stuff. But in recent years, I’ve also taken on more. I’ve picked up a lot of engineering skills and have been enjoying collaborating more on the technical side as well.

It sounds like you’re giving yourself a lot of tasks. I would be careful of that.

It’s okay. I was previously running this startup that had 45 people. It was a very different thing. I was in meetings all day long, every day, 10 hours a day. At Obsidian, we have one meeting per year, so my time is very — 

Wow. Goals! You just made a lot of people very jealous right now.

I think it’s achievable. We use Discord and our Slack equivalent to chat as a team all the time, and we’re in there consistently talking. But in terms of synchronous meetings, it’s quite rare. I think part of it is because everyone’s a user of the app, knows what something Obsidian-y feels like, and we generally only have one or two goals at any given time. So, it makes things quite self-motivated as far as how the team functions.

Got it.

We’ve mentioned that Obsidian is unique in a lot of ways. There’s no signup requirement. You can download and use it for free. You guys don’t even know how many users you have. People can create an unlimited amount of notes. How is that sustainable for you?

Obsidian has a few different revenue sources. One is Obsidian Sync. You probably want to use Obsidian on multiple devices, so a phone, iPad, or computer. Because the files are local to your device, you need a way to keep those versions in sync. You can totally sync your files using Dropbox, iCloud, or Google Drive. There are many different services out there, but we make our Obsidian Sync service. We think it’s the best one because it’s totally integrated into the app, and it has a few features like version history. It’s end-to-end encrypted, so it’s much more secure than a lot of the other options by default. So, that is one of our major revenue drivers. 

Publish is another service where you can take your notes in Obsidian and make a website out of it. Then, we also have a couple donation programs, which are Catalyst and our commercial license where users who want access to the beta versions or want to support us because they believe in what we’re doing can essentially send us money. 

So, those are the main ways that we make money. We have merch, but it’s actually all breakeven, so we don’t really make any profit from that. The thing is, because the team is small, we don’t need mountains of cash. It’s just us and some computers, so it’s not like there’s a large amount of expenses. So it works.

Is it profitable?

Yes. It’s been profitable pretty much since day one. Since I think even before Sync launched —  the donation program, Catalyst, was the first thing that launched —  so it’s been profitable for five years.

Let me ask the other big Decoder question. How do you make decisions at your company? Do you have a framework?

We have this manifesto that you can look at. In a way, I think that is our most powerful driver because it describes our values, which are to make this app that’s super private, super customizable, and durable around  these files that hopefully you’ll be able to own for the rest of your life. The community is always driving us towards the next big problem. 

For example, last year one of the big things we started working on was Web Clipper. A lot of other apps have something like it. I think Evernote was probably the first one that did a really good job with this back in the day. Then, there are services like Pocket that shut down recently. A lot of people in the community were saying, “Hey, this is a major hole, a gap for Obsidian. All these other apps have great web clipping tools.” So, we built one. I think that whether it’s through plugins or through just general complaining from the compute community, we kind of know what the biggest gaps are. So, we always have a general idea of what we want to work on next. 

But there’s also a self-motivated aspect to it. Because our team uses it all day, someone will become an advocate or champion for something. For example, I’m always using the iOS app, and I’m always coming across edge cases where I feel like there’s too much friction. There are other people on our team who use Android, use Linux, or who use Obsidian in a slightly different way. It kind of becomes your mission internally. 

I don’t think this is really that different from other companies. But it kind of becomes your flag to raise with the rest of the team and convince them that this is an important problem that we have to work on now. Because the organization is so flat, we can make decisions very easily. It’s also very easy for one person to go off and prototype something for a few days and show it to the team and say, “Hey, solve this problem. Help me get this polished so we can release it.” 

So, that makes it really, really fun. And because we don’t have investors or any top-down pressure forcing any deadlines, it’s very self-motivated. I’m sure there are lots of people in the community who wish we would release things faster, but we don’t want to give up the freedom, flexibility, and joy that we have building it.

Let me end on a few questions about the future. Virtually every major company that’s making a productivity tool is incorporating some generative AI feature or integrating an AI plugin or API. What is the Obsidian view on AI and productivity tools? Will you add features like that?

So far, there are no AI options built into Obsidian except in Web Clipper, which is intentional because it lives outside of Obsidian itself. In Web Clipper, we have a feature called Interpreter that allows you to put in a bunch of prompts or questions at the time you’re capturing a webpage, like if you want to fill in metadata about that page or say who the author is. 

Let’s say you’re saving products because you’re doing research on what podcasting microphone you want to buy next. You could grab all the metadata and specifications automatically just by putting in a prompt, and it will kind of save all of that. But that’s living outside of Obsidian and it’s not about replacing your thinking. I think the fear I have with AI is that I don’t want it to replace thinking in my own use, the insights that I’m going to gather with a summary generated by AI.

Now, there are tons of people using AI with Obsidian. Because of the plugin architecture, AI is by far the most popular category of new, up-and-coming plugins right now. There are a lot of plugins that people are making using AI. A lot of the LLMs are very knowledgeable about Obsidian and its API. You can just go into Claude and say, “Hey, make me an Obsidian plugin that this or that.” That’s a big challenge for us because there’s a mountain of plugins growing really quickly that we need to review, and it’s happening faster than we can keep up with because AI makes it so easy to make plugins.

Oh, wow.

So, AI is definitely being used. Our philosophy as far as how it would ever make sense for Obsidian is that it has to fit with the principles that are in our manifesto, which is that it would have to be private. We’re not comfortable with the idea that our users’ data could be stored in OpenAI servers without their consent. I think a lot of tools out there are just kind of defaulting to this feeling that there’s an arms race. We’ve got to put AI into everything. Let’s put a little magic button everywhere. I don’t think that’s us. We want to give users confidence that their thoughts are theirs, that things are not going to be used to train the next LLM.

That said, I do think AI can be really powerful for certain uses. So the question is, in the long term, do we end up giving an API to the plugin community so that they can build those types of functionalities more easily? Right now, we’re not working on it. We’ve been holding off and watching what’s going on. We don’t feel a sense of urgency to suddenly put all these things in there because, to be honest, the plugin ecosystem is there for you and you can do it if you really want that. There are things much more important to us on the priority list that we want to work on first, that we would rather set our time aside for with our limited capacity.

Let’s talk about the medium to long-term future of Obsidian. What does it look like when 95 percent of its features are built? What do you hope it does that it can’t quite do today?

The sands are always shifting. We have operating systems that are changing. We’re built on top of macOS, Windows, Linux, iOS, and Android. We have to keep Obsidian working on all those platforms. That work is never-ending and challenging. But it’s hard to imagine what would happen first: we run out of ideas and features or something radically different comes along that people want to use instead of Obsidian. I have this point of view that Obsidian is not necessarily going to last forever. 

There’s going to be a point in time — I don’t know if it’s in five years, one year, 10 years, 50 years — where we’re not going to be using these exact same kinds of apps. I don’t know what is going to replace it, or if we’re even going to be using computers in the same way. Interfaces may change very radically. I’m not sure what it is. I do feel confident that the files you create will end up being really important in that new world. We’re seeing that with AI actually because it turns out that all of the LLMs speak Markdown and are using it behind the scenes because it’s just plain text, and that’s what LLMs are good at. 

But I don’t know the answer to your question. It’s hard to imagine a world where we completely run out of ideas. It seems more likely that we will just die of old age as an app. Maybe five years from now we will have some other idea for an app that we want to work on. But it’s hard to imagine just running out of things to work on.

What’s the next thing that you’re working on?

Right now, it’s a feature called Bases. The idea is that you can store properties, or metadata about the current file, in Obsidian notes. For example, if I have a note about Decoder, I might put the name of the host and a list of episodes. For each episode that I want to take notes on, I might write down which guests were on, what date it came out, or the episode number. What Bases allows you to do is visualize a certain kind of note as a table or eventually as a Kanban view or another type of view. So, it’s like a visualization layer on top of the data that you already have. We just make it really easy to create that database from the bottom up.

It’s kind of like a backward database because all the data is already in there. You’re just looking at it and saying, “Show me all notes that have the ‘books’ tag,” for example, or a link to “Casey.” Then, I get a table and then I have all my metadata, which I can edit. It’s quite powerful if you’re someone who enjoys tracking books that you read, or the movies that you watch, the places that you go, the articles you’ve read. You can very easily create these structures or do project management. 

So, we’re having a lot of fun with that. It’s been way more popular than we expected. It’s currently in beta, so hopefully we will be releasing the first public version in the near future. Then, I expect that we’re going to be working on this until the end of the year or even longer because the feedback has been so positive.

Well, if you want to send any of those notes that you took about me, I can take a look and let you know if there are any errors.

Yes, no problem. Just don’t get too obsessed with tweaking the fonts and everything.

[Laughs] I’ll try not to. I’m always at risk of doing that. Steph, thank you so much for joining us today.

Thank you, Casey. It was great.

_{Questions or comments about this episode? Hit us up at decoder@theverge.com. We really do read every email!}

Hello, and welcome to Decoder! This is Casey Newton, founder and editor of the Platformer newsletter and cohost of the Hard Fork podcast. I’ll be guest hosting the next few episodes of Decoder while Nilay is out on parental leave, and I’m very excited for what we have planned. 

If you’ve followed my work at all, particularly when I was a reporter at The Verge, you’ll know that I’m a total productivity nerd. At their best, productivity apps are the way we turn technological advancement into human progress. And also: they’re fun! I like trying new software, and every new tool brings the hope that this will be the one that completes the setup of my dreams. 

Listen to Decoder, a show hosted by The Verge’s Nilay Patel about big ideas — and other problems. Subscribe here!

Over the years, I’ve used a lot of these programs, but I rarely get a chance to talk to the people who make them. So, for my Decoder episodes, I really wanted to talk to the people behind some of the biggest and most interesting companies in productivity about what they’re building and how they can help us get things done.  

That brings me to my guest today: Michael Truell, the CEO of Anysphere. You may not have heard of Anysphere, but you’ve likely heard the name of its flagship product: Cursor. Cursor is an automated programming platform that integrates with generative AI models from Anthropic, OpenAI, and others to help you write code. 

Cursor is built into a standard version of what programmers call an integrated development environment, or IDE, with technology like Cursor Tab, which autocompletes lines of code as you write. Cursor has quickly become one of the most popular and fastest-growing AI products in the world, and Anysphere, the company Michael cofounded just three years ago after graduating from the Massachusetts Institute of Technology, is now shaping up to be one of the biggest startup success stories of the post-ChatGPT era. 

So I sat down with Michael to talk about Cursor, how it works, and why coding with AI has seen such incredible adoption. As you’ll hear Michael explain, this entire field has evolved very quickly over the past few years — and here in San Francisco, tech executives and employees regularly tell me about how much their employees love using Cursor. 

AI critics are worried that this technology could automate jobs, and rightly so — but you’ll hear Michael say that job losses won’t come from simple advances in tools like the one he’s making. And while a lot of people in the Bay Area believe superintelligent AI is going to remake the world overnight, making products like Cursor pointless, Michael actually believes change is going to come much more slowly. 

I also wanted to ask Michael about the phenomenon of vibe coding, which lets amateurs use tools like Cursor to experiment in building software of their own. That’s not Cursor’s primary audience, Michael tells me. But it is part of this broader shift in programming, and he’s convinced that we’re only just scratching the surface of how much AI can really do here. 

Okay: Anysphere CEO Michael Truell. Here we go.

This interview has been lightly edited for length and clarity.

Michael Truell, you are the cofounder and CEO of Anysphere, the parent company of Cursor. Welcome to Decoder.

Thank you for having me.

So what is Cursor? What does it do, and who is it for?

Our intention with Cursor is to have it be the best way to build software and, specifically, the best way to code with AI. For people who are nontechnical, I think the best way to think about Cursor, as it exists today, is as a really souped-up word processor in which engineers build software by actually doing a lot of writing. They’re sitting in something that looks like a word processor, and they’re editing millions of lines of logic — things that don’t look like language. Cursor helps them do that work way more efficiently, especially with AI. 

There’s two different ways Cursor does this right now. One is that as Cursor watches you do your work, it tries to predict the next set of things you’re going to do within Cursor. So this is the autocomplete form factor, which can be really souped up in programming when compared with writing, because in programming, unlike in writing, oftentimes the next 20 minutes of your work are entirely predictable. Whereas in writing, it can be a little hard to get a sense of what a writer is going to put down on the page. There isn’t enough information in the computer to understand the next set of things the writer is going to do.

The other way people work with Cursor is by increasingly delegating to it, as if they’re working with a pair programmer, another human. They’re handing off small tasks to Cursor and having Cursor tend to them.

Well, we’ll dig a little deeper into the product in a moment. But first let’s talk about how all of this started. When you founded Anysphere, you were working on computer-aided design (CAD) software. How did you get from there to Cursor?

My cofounders and I had been programming for a while, and we’d also been working on AI for almost as long as we’d been programming. One of my cofounders had worked on recommendation systems in Big Tech. Another had worked on computer-vision research for a long time, while another had worked on trying to make machine learning algorithms that could learn from very, very, very little data. One of us had even worked on a competitor to Google, using the antecedents that came before LLM technology in machine learning. 

But we’d worked on AI for a long time and had also been engineers for a long time and loved programming. In 2021, there were two moments that really excited us. One was using some of the first really useful AI products. Another was this body of literature that showed that AI was going to get better, even if we ran out of ideas, by making the models bigger and training them on more data.

That got us really excited about a formula for creating a company, which was to pick an area of knowledge work and build the best product for that area of knowledge work — a  place where you do your work as AI starts to change. And then, the hope is that you do that job well, and you get lots of people to use your product and you can see where AI is helping them and where AI is not helping them — and where the human just has to correct AI a bunch or do the work without any AI help. You can use that to then make the product better and push the underlying machine- learning technology forward. That can maybe get you onto a path where you can really start to build the future of knowledge work as this technology gets more mature, and be the one to push the underlying tech too.

So, we got kind of interested in that formula for making a company, but the craft that we really loved, the knowledge work that we really loved, was building things on computers, and we actually didn’t touch that at first. We went and we worked on a different area, which was, as you noted, computer-aided design. We were trying to help mechanical engineers, which was a very ill-fitted decision, because none of the four of us were mechanical engineers. We had friends who were interested in the area. We had worked on robotics in the past, but it wasn’t really our specialty. We did it because it seemed there were a bunch of other people working on trying to help programmers become more productive as AI got better.

But after six or so months of working on the mechanical engineering side of things, we got pulled back into working on programming, and part of that was just our love for the space. Part of it, too, was that it seemed as if the people who we thought had the space covered were building useful things, but they weren’t pointed in the same direction and they didn’t really seem to be approaching the space with the requisite ambition. So we decided to build the best way to code with AI, and that’s where Cursor started.

I have read that one of the AI tools that you used early on was GitHub Copilot, which came out about a year before ChatGPT. What was your initial reaction to Copilot, and how did it influence what you wanted to build?

Copilot was awesome. Copilot was a really, really big influence, and it was the first product that we used that had AI really at its core that we found useful. One of the sad things to us as people who had been working on AI and interested in AI for a while was that it was very much stuff that was just in the lab or in the toy stage. It felt like, for us, the only real way AI had touched our lives as consumers was mostly through recommendation systems, right? The news feeds of the world, YouTube algorithms, and things like that. GitHub Copilot was the first product where AI was really, really useful at its core and that wasn’t vaporware. 

So, Copilot was a big inspiration, and at the time we were considering whether we should try to pursue careers in academia. Copilot was proof that no, it was time to work on these systems out in the real world. Even back then, in 2021, there were some rough edges. There were some places where the product was wrong in really obvious ways, and you couldn’t completely trust its code output, but it was nonetheless really, really exciting.

Another thing to note is that apart from being the first useful AI product, Copilot was the most useful new development tool that we had adopted in a really long time. We were people who had optimized our setups as programmers and modded out our text editors and other things like that. We were using this crazy kind of text editor called Vim at the time. So, it was not only the first useful AI product that we had used, but also the most useful dev flow we had used in a really long time.

That’s interesting. So you all like software, you like using software, you’re trying to find software that makes you more productive. I feel like that probably made you well-suited to tackle a problem, the one Cursor is trying to solve.

Yeah, I think caring about the tools we use was helpful, and I think that there were actually different degrees of that on the cofounding team. One of my cofounders is straight out of central casting, an early adopter who is the first one on these new browsers, first one on the new category of everything. A couple of us are a little bit more laggard, and so I think having that diversity of opinions has helped us in some of the product decisions we’ve made.

So you described Cursor as kind of like a souped-up word processor. Software engineers I think would call it an integrated development environment, or an IDE. Developers have been using IDEs since the ’80s, but recently, AI labs have released tools, like OpenAI’s Codex or Anthropic’s Claude Code, that can run directly in a terminal. Why might someone use Cursor over those options?

I think that both of those are really useful tools. What we care about being, I think we start as this IDE, as this text editor, but what we really care about is getting to a world where programming has completely changed, in particular a world where you can develop professional-grade software, perhaps without even really looking at the code. And, yeah, it’s that kind of future programming and changing it from this weird, you’re reading these millions of lines of logic and these esoteric programming languages. 

The world we want to get to is one where you just need to testify the minimal intent necessary to build the software you want. You can tell the computer the shortest amount of information it needs to really get you, and it can fill in all of the gaps. Programming today is this intensely labor-intensive, time-intensive thing, where to do things that are pretty simple to describe, to get them to actually work and show up on a computer, takes many thousands of hours and really large teams and lots of work, especially at professional scale. So that’s where we want to get to — inventing that new form of programming. I think that that starts as an editor and then that starts to evolve.

So we’re already in the midst of that. Right now, Cursor is where you can work one-on-one with an agent, and with our Tab system. And then, increasingly, we’re getting you to a world where more and more of programming is moving toward delegating your work to a bunch of helpers in parallel. And there’s a product experience to be built for making that great and productive, with an understanding of what all of these parallel helpers are doing for you — diving in, intervening in places where it’s helpful, understanding their work when they come back to you at a level of not having to read every single line of code.

I think that there’s a competitive environment with a bunch of tools that are interested in programming productivity. One of the things that’s limiting about just a terminal user interface is that you have only so much expressiveness in the terminal and control over the UI. From the very start, we’ve thought that the solution to automating code and replacing it with something better is this kind of two-pronged approach, where you need to build the pane of glass where programmers do their work, and you need to discover what the work looks like. You need to build the UI, and then you also need to build the underlying technology. So, one thing that would distinguish us between some terminal tools is just the degree of control you have over the UI.

We’ve also done a lot of work on the model layer, on improving it and going beyond just having things that show up well on a demo level. There’s a lot of work on AI products to dial in the speed and the robustness and the accuracy of them. For us, one important product lever has been building an ensemble of models that work with the API models to improve their abilities. 

So, every time you call out to an agent in Cursor, it’s like this set of models — some of them are APIs, some of them are custom — and then for some form factor or for some of the features, it’s entirely custom, like for the super autocomplete. That’s also one thing that has kind of distinguished us from other solutions.

Let’s talk a bit about these proprietary models. They seem to be fueling a lot of your success. When ChatGPT and the OpenAI API first got released, we saw a lot of startups come out that were quickly dismissed as just wrappers for an API that was just trying to build something on top of somebody else’s tech. 

Cursor started in a similar way in that it was using other folks’ APIs in order to create its product. Since then, you’ve started to build on top. Say a bit more about what you’re building and how you’re hoping it sets you apart from those pure wrapper companies.

I think also one asterisk before getting into the model side of things is that the “wrapper” term came from the very start of when people were building AI products, when there was only so much time to make the products a bit deeper. Now, I think we’re at a point where there’s a ton of product overhang. So even if you’re just building with the API models, I think that in a lot of areas — our area of working on the software development lifecycle, but in other parallel areas too — there are very, very deep products to be built on top of those things. So it sounds like the wrapper term for at least some areas is a little bit dated.

But on the model level, I think that from the very start we wanted to build a product that got a lot of people using it. One of the benefits you get from that scale is you can see where AI is helping people, and you can see where AI is not helping people and where it gets corrected. That’s a really, really important input to making AI more useful for people. So at this point our Tab model, which does over one billion model calls per day, is one of the largest language models actually writing the most production code in the world.

We’re also on our fourth or fifth generation of it. And it’s trained using product data, of seeing where AI is helping people and where it isn’t, trying to predict how it can help humans. It also requires a ton of infrastructure and specialty talent to be able to make those models really good.

For instance, one of the people who has worked on those models with us is Jacob Jackson, who actually built GitHub Copilot before GitHub Copilot, which was called TabNine and was the first type of programming autocomplete product. He’s also one of the people who built one of the first million token-context window models, and so he has done a lot of work on making models understand more and more and more information, and yeah, specialty talent and specialty infrastructure, too, to do that work.

I think that in our ambling, kind of winding way to working on Cursor, one of the things that really did help us was when we were working on CAD and also in some of our explorations before, my cofounders had to dig very deep into the machine-learning infrastructure and modeling side of things. When we actually set out to work on Cursor, we thought it would be a long time before we started to do our own modeling as product lovers, but it happened much sooner than we expected.

Recently, I had dinner with the CTO of a Big Tech company, and I asked him about what coding tools were popular with his engineers, and he told me that he regularly surveys them on this question, and they had Cursor available as a trial. He said he was getting these panic messages from engineers saying, “Please tell us you’re not about to take away Cursor,” because they’d become so dependent on it. 

Can you give us a sense of why, for programmers, this has kind of felt like a before-and- after moment in the history of the profession? What is it that tools like Cursor are making so different in the day-to-day lives of these engineers?

I think that we’re just already at a point where we are far, far, far from the ceiling of where things can go, and far, far, far from a world where much of coding has been replaced with something better. But just now at this point, these products and these models can do a lot for programmers and are already taking on quite a bit of work.

I think the technology is especially good for programming for a few reasons. One is that programming is text-based and that is the modality that the field has figured out perhaps the most.There’s a lot of programming data on the internet too, so a lot of open-source code. Programming is also pretty verifiable. And so, one of the important engines of AI progress has been training models to predict the next word on the internet and making those models bigger. That engine of progress has largely run its course; there’s still more to do there.

But the next thing that’s kind of picked up the torch in making models better has been reinforcement learning. So it’s been basically teaching models to play games, kind of similar to how in the mid-2010s we, humanity, figured out how to make computers really good at playing Go and Dota and other video games. We’re kind of getting to a level of language models where they can do tasks, and you can set up games for them to get even better at those tasks. And programming is great for that, because you can write the code and then you can run it and see the output and decide if it’s actually what you want. And so I think there’s a lot about the technology that makes it especially good for programming, and, yeah, it’s just I think one of the use cases that’s the furthest ahead in deploying this tech out to the world and people finding real value from it.

My sense is, maybe if I used to have to work eight hours a day, now it’s maybe closer to five or six. Is that part of it?

I think yes, in the sense that I think that the productivity gains of what would have taken you eight hours before in some companies now actually can take you five or six hours. I think that that is real, not across all companies, but it is really real in some companies. But what I would nitpick on there is I don’t think programmers are shortening the hours they’re working. I think a lot of that is because there is just a ton of elasticity with software, and I think it’s really easy for people who are nontechnical, or who just don’t program professionally, to underrate how inefficient programming is at a professional scale, and a lot of that is because programming is kind of invisible.

Consider what programmers are doing at a company like Salesforce, where there are just tens of millions of lines, many millions of files of existing logic that describe how its software works. Anytime they have to make a change to that, they have to take that ball of mud, that massive thing that is very unwieldy, and they need to edit it. That’s why I think that it’s just kind of shocking to many people that some software release cycles are so slow. So yes, I think that there are real productivity gains, but I think that it’s probably not reducing the number of hours that programmers are working right now.

All right. Well, you mentioned nontechnical people. Cursor is used by a lot of professional programmers, but this year saw the coining of the term “vibe coding” to describe what more amateur programmers can do, sometimes even complete novices, and often with tools like Cursor. How big is the vibe-coding use case at Cursor and what do you think is the future of vibe coding?

So our main goal is to help people who build software for a living, and for right now that means engineers, and so that’s our main use case. It’s been interesting to see as you focus on that use case and use the understanding you get from it to push the tech forward and hop up programmers to ever-higher levels of abstraction, how it then also makes things more accessible, and that’s something that we’re really excited about. 

I think in the end state, building software is going to be way more accessible. You’re not going to have to have tons of experience in understanding programming languages and compilers. But I do think that we’re a decent bit away from a world where anyone can do this. I think there’s still a bunch more work to do before anyone can build professional-grade software.

That said, it’s been really cool seeing people spin up projects and prototypes from scratch, seeing designers in professional settings doing that. It’s been really interesting to see nontechnical people contribute small patches and bug fixes or small feature changes to professional software projects already. And that’s kind of the vibe-coding use case, not our main use case, not where the company makes most of its money, but one that I think will become bigger and bigger as you push the ceiling of focusing on professional developers.

I’m curious what you think of as the demand for it, though. I understand it’s not your focus of the business. People like to talk about it, and I think it feels cool to have never built software before, and all of a sudden the next thing you know, you actually created a little to-do list app for yourself or something.

Yes. I probably differ from some of my colleagues on this, where I think that, in the world as it exists right now, of the two buckets of that vibe-coding use case, there’s an entertainment bucket if you’re doing these things mostly for personal enjoyment or hobbies, and then there’s a bucket that’s more professional, and I think that that’s designers doing prototypes or that’s people who work to serve customers and are contributing back bug fixes to a professional code base.

The way in which I probably differ from some of the people I work with is there’s a group of people who are really, really, really interested in end-user programming and throwaway apps and personalized software, where everyone entirely builds their own tools. And I think that that’s really cool. I think enabling that is really cool, and I think some people, a lot of people who aren’t technical will be interested in doing that. But I still think even if you get to a world where anyone can build things on computers, I think most of the use cases will still be served by a small minority of 5 percent of the world that cares a ton about the tools and building them, and that everyone will use those tools more, because I just think that the interest in that stuff really differs among the population.

So yeah, right now commercially I think that a lot of the more vibe-coding stuff falls more into a midjourney camp or an entertainment camp. It’s something that some people get interested in for a bit and then kind of put it aside. And then some of it is in this professional camp of people who work on software for a living but don’t code right now.

I think you’re right, because when I worked at more traditional companies, whenever a new piece of software was introduced, everyone would get upset. So that’s my case for most people not becoming pro-vibe coders. I like software though, so I’m vibe-code curious. Maybe two or three generations from now in Cursor I’ll be able to make myself something useful.

You mentioned earlier that there are these two main ways that people use Cursor. There is the “I’m looking at code and you’re helping me autocomplete things,” and then there is the “I’m going to give you a task and walk away and come back and see what you’ve built.” You told Stratechery’s Ben Thompson recently that over the course of the next six to 12 months, you think you can get to a place where maybe 20 or 25 percent of a professional software engineer’s job might be the latter use case of just handing off work to the computer and having the computer do the work end to end. 

Do you have any updates to that number in the past month or so? How high do you think that number can scale, ultimately?

I think these things are really hard to predict. Yeah, I think there are some things that are blocking you from getting to 100 percent. One is having the models learn new things, like understanding an entire code base, understanding the context of an organization while learning from the mistakes. And I still think that the field doesn’t have an amazing solution for that.

There are two candidate solutions. One is you make the “context windows” longer, which is that these large language models have a fixed window of text or images that they can see, and then there’s a limit to that. Outside of that, it’s just the model that came off the assembly line and then that new kind of information that’s put into the model’s head, which is very different from that of humans because humans are going through the world and your brain is changing all the time, you’re getting new things that kind of persist with you, and obviously some memories fade away but persist with you somewhat. So candidate solution number one to the continual learning problem is just make the context windows really big.

Candidate solution number two is to train the models. So every time you want them to learn a new thing or a new capability, you go and collect some training data on that, and then you throw it into the model’s mix. Both of those have big issues, I think, but that’s one thing that’s stopping you. I think that the rate of really consequential ideas in machine learning that are new paradigm shifts is pretty low industrywide, even though the rate of progress has been really fast over the past five years. 

So, ideas in the form of replacing long context or in-context learning and fine-tuning with some other way of continual learning, I don’t think that the field actually has an amazing track record of generating lots of ideas like that. I think ideas like that come about at the rate of maybe one every three years. So I think that will take some time.

I think the multimodal stuff will take time too. The reason that’s important for programming is you want to play with the software, and you want to be able to click buttons and actually use the output. You want to be able to use tools also to help you make software, tools that have GUIs. So, for instance, observability solutions, like Datadog, are important for understanding how to improve a professional piece of software, so that feels like it’s needed.

These models can also work coherently for minutes at a time, now even hours in some cases, but it’s a different thing to work on a task for the equivalent of weeks in human time. So, just even architecturally, knowing if we’re going to be coherent over sequences that long will be interesting to see, and that I think will be tricky. 

But there are all of these technical blockers to getting to something that’s 100 percent, and there’s many more that you could list and there are also many unknown unknowns. I think that in a year or so, even with just going from a high-level text instruction to changes throughout a code base, I think in the bull case you could probably do over half of programming as it exists today.

I see these studies that Meter puts out where they look at the average length of time that a software or an AI model can do, and it does keep doubling at this really impressive rate. So, I think the hurdles that you identify are super important, but when you pull it back, it does seem like the task is really improving. Ultimately, humans don’t tend to work on discrete tasks that are all that long. So I do think it’s getting easier for people to imagine a full day’s work.

Definitely, definitely. I think that just forecasting these things is tricky, but one related field that can maybe foretell how things will evolve here is the history of self-driving, which has obviously advanced in leaps and bounds. In San Francisco, there are Waymos, which are commercial self-driving cars, and my understanding is that Tesla has also made big improvements. 

But I remember back in 2017, when people thought self-driving was going to be done and deployed within a year. Obviously, there are still big barriers to getting it out into the world.  As hard and varied as self-driving is, it does seem like a much lower-ceiling task than some of the other stuff that people in the field are talking about right now. So we will see.

I do want to ask you about the timeline, but I’m going to wait until a little bit later. All right, let me now ask you some of the famous Decoder questions, Michael. How big is Anysphere today? How many employees do you have?

We’re roughly 150 people right now.

Okay, and when you think about how big you want the company to be, are you somebody who envisions a very big workforce? Or do you see a smaller, nimbler team?

We do like a nimbler team, and I think the caveat here is while we want to keep the team nimbler for the scope of work that we’re tackling, it will still mean growing the team a lot over the next couple of years. But yeah, I wonder if it will be possible to build a thriving technology company that does really important work with a maximum team size of maybe 2,000 people, or something like that. Something of the size of The New York Times. We’re excited to see if that is possible, but we definitely need to grow a lot more from our current head count.

What is your organization chart like? You have a few cofounders. How do you all divvy up your responsibilities?

The two biggest areas of the org are engineering and the research side of things, like R&D generally, and then the go-to-market side of things, like serving customers. And this is a company that has really benefited from having a big set of cofounders and a big, very capable founding team. And so there’s a lot of dividing and conquering across that scope. In particular, we’ve had an important group of people on the founding team who’ve done phenomenal work in building out that early go-to-market side of things. A lot of that comes entirely from the founding team, and is entirely credited to a subset of it. And so there’s a lot of dividing and conquering across the business.

At the same time, I think once you zoom in to the technical side of things, there’s an intense focus from the four cofounders on that, and putting all the eggs in that one basket. I think we’re lucky enough to be at a time when there are really, really useful products to build in our space. And I believe that the highest order of it, the thing you cannot mess up, is producing the best product in the space. And so we’ve been able to stay relatively lean in other parts of the business, especially relative to our scale, but also as a ratio to engineering and research, and still be able to grow. 

What part of the business do you keep for yourself? Where are you getting your hands dirty, and where would you get mad if someone tried to take that away from you?

I spend a lot of time doing what I can to help grow the team. We think hiring is incredibly important, especially the hiring of ICs [individual contributors]. I think that one way technology companies die is that the best ICs start to feel disengaged, that they don’t have control over the company, and talent density lowers. If you’re working on technology, no matter how good the management layer is, if you have less than excellent people doing the real work, I think there’s only so much you can do. I think that the dynamic range of what management can do becomes kind of limited. 

So l help by devoting a bunch of time to hiring. We actually got to maybe 75 people with just the cofounders hiring without engaging functional recruiters. Now I have fantastic people helping us with hiring. I have people on the recruiting side who work with us closely. But I spend a bunch of time on that and then try to help however I can on the engineering and product side.Those are the two biggest areas of focus, and then there’s a long list of long-tail things. 

So you’re fairly young, I think you’re 25, and have had to make a lot of really big decisions about raising money, making acquisitions, all those hiring decisions that you just made. How do you make decisions? Do you have a framework that you use or is everything ad hoc?

I’m not sure there’s one framework. Some pretty common strategies that help us are, we try our best to farm all up and down the group, the org. This is not just for me — we try to do this for all decisions in the company. We increasingly have a very clear DRI [directly responsible individual], and then lots of other people offer their input. Every decision is pretty unique. 

Other devices that are well-known and have helped include understanding how high stakes and reversible the decision is. And I think that especially when you’re in a vertical like ours, given the speed that it’s moving, there’s just a limit on the amount of time and the amount of information you can gather on each thing. Yeah, and then other devices, like clearly communicating the decision and using that as a way to force clarity for how it was thought through.

Well, let’s talk a little bit more about hiring, since you brought it up. There has been talk that OpenAI had considered acquiring you. I have to ask, given his recent spending spree, has Mark Zuckerberg invited you to his house in Tahoe?

[Laughs] No, no.

No? He’s not coming around with his $200 million signing bonuses saying, “Michael, why don’t you kind of come over here? We’re building super intelligence?”

No. This for us is kind of life’s work territory. So yeah, we feel really lucky to have the technology lineup, the initial founding team lineup, the people who have decided to join us, the way things have gone on the product to have the pieces in place to execute on this ambitious goal of automating programming. And time will tell if we’re going to be the ones to do that, but as people who have been programming for a long time and working on AI for almost as long, being able to reinvent programming and help people build whatever they want to on computers with AI, kind of feels perfect for us. It feels like one of the best commercial applications of this technology too. So I think that if you can succeed in that, you can also push the field forward in big ways for other verticals and other industries. And so, no. 

Yeah, it sounds like you really want to stay independent.

Yeah.

Has Meta’s recent hiring spree made it noticeably harder for you to recruit lately?

No, not really. We try to keep the research team fairly small. I mean, the whole company is kind of small relative to what it’s doing, but especially the research team. I think that people think through hiring decisions in different ways, and I think what we have to offer is most appealing to people who want to be a part of an especially small team working on something focused, that’s solving problems with AI out in the real world. 

We’re kind of this weird company. You talked about some products that are being made by some of the great folks who work on the API models. But I think we’re this weird experiment of a company that’s smack dab in between the foundation model labs and normal software companies; we try to be really excellent at both the product side of things and the model side of things and have those feed into each other. And so we appeal to I think a certain type of machine-learning researcher or ML engineer. And for them, I think it’s about being part of this, and a little bit less about being part of some of the other things.

One last hiring question. It was reported this week that two folks who used to run Claude Code whom you’d recruited to come over to Cursor left after a couple of weeks. Can you speak at all to what happened there?

Cat [Wu] and Boris [Cherny] are awesome, and I think that they have a lot left to do on Claude Code, and they’re really, as I understand it, the people behind that and that is their creation. As someone who’s been working on something for three and a half years since inception, I understand the ownership that comes with that. I think that they have a lot left to do and they were excited about that, and so they’ve decided to stay [at Anthropic]. 

It seems that you were mentioning this interesting position Cursor sits in, in between the big labs and other startup companies that are using your software. How do you describe Cursor’s culture when you’re recruiting people?

I think that some of the things that describe the current group, perhaps unsurprisingly — we are process skeptical and hierarchy skeptical. So, as we take on more and more ambitious projects, more and more coordination is required. But at a certain level, given the scope of the company, we try to stay pretty light on each of those.

I think it’s a very intellectually honest group, where people feel comfortable. It feels very low stakes to criticize things and just be open when giving feedback on work. But I also think it’s a very intellectually curious group. I think people are interested in doing this work for the end goal of automating programming — separate from any work-life balance issues, because we want this to be a place where people at all levels of work-life balance can do great work. 

It’s a place where so far no one really treats it as just a job. They’re really, really excited about what we’re doing, and I think it’s kind of a special time to be building technology. I think to the outside world, what we do seems very focused and understated, partially because of how little communication we have with the outside world. We need to get much better at that.

I think for the most part people think of Cursor as, “Oh, that thing that grew really fast.” They know about top-level metrics and things like that to gauge just how fast the adoption has been. Internally, we’ve thought that it’s really important to hire people who, while they might be very ambitious, are still very humble and understated and focused and level-headed, because there’s noise left and right. I think that just having a clear focus and putting your head down are actually really, really important not only for people to be happy in this space but also for the team’s execution. 

You mentioned communicating with the outside world. I think Cursor’s history is mostly just a history of delighting its customers. But you did have this moment recently where you changed the way you price things, and folks got pretty mad. Basically, you moved from a set fee to more usage-based pricing, and some people ran over their limits without realizing it. What did you learn from that experience?

I think that there was a lot to learn from that, and a lot on our end that we need to improve. To set the stage, the way Cursor pricing has worked, even back when Cursor first started, is by and large, you sign up for a subscription, and then you get an allotment of a certain number of times you can use the AI over the course of your subscription term. And the pricing has evolved. Features have been added, features have been changed, kind of up and down that limit, and there have been different ways you could pay down that limit or not pay down that limit over time. What’s happened in parallel is using the AI once, and what that means is the value that gives people and the underlying costs in some cases have changed a lot. One big switch there for us is that increasingly when “you use the AI,” the AI’s working for longer and longer and longer.

So you called out that chart that you’ve seen where it shows the max time that AI can work, and it’s gone from seconds to minutes to hours at this point, and it’s gone up very fast. We’re on the front lines of that, where now when you ask the AI to go do something or answer a question, it can work for a very, very, very long time. That changes the value it can give to you. You can go from just asking a simple programming question to having it write 300 lines of code for you, and that also changes the underlying costs. In particular, it changes less the median and more the variance of those costs. So we bundled together a series of pricing changes, and the one that garnered the most attention was switching from a world where the monthly allotment is in requests to one where it’s in the underlying compute that you’re spending.

One thing to knit on what you said is that usage-based pricing had been a big component of Cursor before, because over the life of Cursor, people have used the AI more and more and more and more. And then they started running out of limits, and we wanted to give people a way to burst past that. What this did is it changed the structure of how that usage pricing worked, where it’s not on a request basis but on the underlying compute basis. That definitely could have been communicated legions better. I think that there’s a lot we learned from that experience, and a lot we need to improve on in the future.

I think it’s hard for consumers in particular to understand usage-based pricing, because they’re used to Spotify and Netflix, where they pay their 10 or 20 bucks a month and it’s sort of all you can eat. The economics of AI don’t really work that way.

Yeah, I think that it will be interesting to see how things play out in our space in particular, because I think that for the consumer chat-app market, so far at least, it would be interesting to see how the curves of just how compute per user over time have gone up. But I wouldn’t be that surprised if it’s been pretty flat over the past 18 months or so, where the original GPT-4, I’m not privy to any inside information, but it seems like there have been big gains from a model-size perspective, where you can actually miniaturize models and get the same level of intelligence. And so I think that the model that most professional users are using in something like a ChatGPT has actually maybe gotten smaller over time; compute usage has gone down.

But in our space, I think that for one user, compute is probably going to go up. There’s a world in which the token costs don’t go down fast enough, and it starts to become a little bit more like AWS costs and a little bit less like Percy productivity software, and that still remains to be seen. But one thing to note is that we do think it’s really, really, really important to offer users choice, and so we want to be the best way to code with AI, if you want to turn on all the dials and get the best, most expensive experience. 

We also want to be the best way to code with AI if you want to just pay for a predictable subscription and get the best of what that price can offer you. And even for the main individual plan, the $20 Pro plan, the vast majority of those users don’t hit their monthly limits, and so aren’t hit with a message saying they need to turn on usage pricing, or not.

That’s the kind of AI user I am. I never hit the limit, which makes me feel that I need to be using it more. 

There is a really, really big difference between the top 5 percent and a median user. So some people are very, very, very AI forward. 

Well, coming into my last couple of questions here, I want to try to get at how AGI-pilled you are, because when we were talking earlier, you’re sort of identifying all these very real technical problems in building more advanced systems that aren’t just truly unsolved problems in AI. The size of the context when giving these systems longer memory, helping them learn the way that a human might be able to learn, we don’t know how to do that yet. 

Yet there are lots of folks in the industry who believe that by 2027, 2028, the world will look very, very different. So, where do you sort of plot yourself on the spectrum of people who believe that everything is absolutely about to change, and we’re sort of at the start of a process that’s going to take decades?

I think we’re kind of on this bet in the messy middle, where we do think it’s going to take decades. We do think that nonetheless, AI is going to be this transformational technological shift for the world. Bigger than maybe… just a very, very, very big technological shift. And when we started working on Cursor, it was funny, we would get these dual responses, and I think one is now increasingly falling out of favor with the rise of the first AI products that have reached billions of people. 

But early in 2022, we would get two reactions. One reaction was, “Why are you working on AI? I’m not sure that there’s really much to do there.” And then the other reaction we’d get, because we did have close friends and colleagues who were very interested in AI, was, “Why are you working on ‘insert X’ application” — whether it be CAD or whether it be programming specifically — “when AGI is going to wipe all of this stuff out in Y years,” maybe in 2024 or 2025.

We think it’s this middle road of this jagged peak, where if you actually peek under the hood at what’s driven AI progress so far, I think that, again, there’s been a few ideas that have really worked, there’s been lots of details to fill in between, but there have been a few really, really important ideas. I think that despite the number of people who have worked on deep learning over the past decade and a half, the rate of idea generation in the field — really, really consequential idea generation in the field — hasn’t budged that much. I think that there are lots of real technical problems that we need to grapple with. So, I think that there’s this urge to anthropomorphize these models and see them be amazing and human or even superhuman at some things, and then think that they will just kind of be great at everything. I really think it’s this very jagged peak.

So, I think it’s going to take decades. I think it’s going to be progressive. I think that one of our most ambitious hopes with Cursor is if we are to succeed in automating programming and building an amazing product that makes it so you can build things on computers with just the minimal intent necessary, maybe the success of that and the techniques that we need to figure out in doing that can also be helpful for pushing AI forward and pushing progress forward in general. 

I think the experiment to play back here is if you were in the year 2000 or 1999 and you wanted to push forward with AI, one of the best things you could do is work on something that looks like Google, and make that successful and make that R&D available to the world. So, in some ways at least, I think about what we’re doing is trying to do just that.

So it sounds like you don’t think that there’s just going to be one big new training run with a lot more parameters and we’re going to wake up to a machine god.

Time will tell. I think it’s important to have healthy skepticism about how much you can know with these things. But my best guess is that it will take longer than that, yet also still be this big transformational thing.

Well, last question here. We’ve talked a couple of times today about how hard predictions are in general, so I’m not going to ask you to do something crazy like predict what Cursor is going to look like five years from now. But when you think about it maybe two years from now, what do you hope it’s doing that it isn’t quite doing yet?

I think a bunch of things. So I think in the short term, we’re excited about a world where you can delegate more and more work to very fast, helpful humans, and you can build a really amazing experience for making that work delightful while orchestrating work among these agents.

Another idea that we’ve been interested in for a long time, which is a bit risky, is if you can get to a world where you’re delegating more and more work to the AI, you’ll start to run into an issue, which is whether you look at the code. And are you reading everything line by line, or are you just kind of ignoring the code wholesale? I think that neither closing your eyes and ignoring the code entirely in a professional setting nor reading everything line by line will really work. 

So, I think you’ll need this middle ground, and I think that that could look like the evolution of programming languages to become higher level and less formal. All that a programming language really is is a UI for you as a programmer to specify exactly what you want the computer to do. And it’s also a way for you to look at and read exactly how the software works right now.

I think that there’s a world where programming languages will evolve to be much higher level and more compressed. Instead of millions of lines, it’s hundreds of thousands of lines of code. I think that for a while, an important way you build software is you could read, point at, and edit that kind of higher-level programming language. 

That also gets at a bigger idea that’s behind the company: there’s all this work to do on the model side of things. The field’s going to do some of that, and we’re going to try to do some of that. But then the end state of what we want to do is also this UI problem of how we get the stuff that’s in your head onto the screen. 

I think that the vision of you entirely building software by just typing into a chat box is powerful. I think that that’s a really simple UI. You can get very far with that, but I don’t think it can be the end state. You need more control when you’re building professional software. And so you need to be able to point at different elements on the screen and be able to dive into the tiniest detail and change a few pixels.

You also need to be able to point at parts of the logic and understand exactly how the software works and be able to edit something very, very fine-grained. That requires rethinking new UIs for these things, and the UI for that right now is programming languages. So I think that they’re going to evolve.

All right. Well, a lot of fascinating things that you’re working on. Michael, thank you for coming on Decoder.

Thank you for having me.

_{Questions or comments about this episode? Hit us up at decoder@theverge.com. We really do read every email!}

This is Platformer, a newsletter on the intersection of Silicon Valley and democracy from Casey Newton and Zoë Schiffer. Sign up here.

Today let’s talk about an advance in Bard, Google’s answer to ChatGPT, and how it addresses one of the most pressing problems with today’s chatbots: their tendency to make things up.

From the day that the chatbots arrived last year, their makers warned us not to trust them. The text generated by tools like ChatGPT does not draw on a database of established facts. Instead, chatbots are predictive — making probabilistic guesses about which words seem right based on the massive corpus of text that their underlying large language models were trained on.

As a result, chatbots are often “confidently wrong,” to use the industry’s term. And this can fool even highly educated people, as we saw this year with the case of the lawyer who submitted citations generated by ChatGPT — not realizing that every single case had been fabricated out of whole cloth.

This state of affairs explains why I find chatbots mostly useless as research assistants. They’ll tell you anything you want, often within seconds, but in most cases without citing their work. As a result, you wind up spending a lot of time researching their answers to see whether they’re true — often defeating the purpose of using them at all.

When it launched earlier this year, Google’s Bard came with a “Google It” button that submitted your query to the company’s search engine. This made it slightly faster to get a second opinion about the chatbot’s output, but still placed the burden for determining what is true and false squarely on you.

Starting this week, though, Bard will do a bit more work on your behalf. After the chatbot answers one of your queries, hitting the Google button will “double check” your response. Here’s how the company explained it in a blog post:

When you click on the “G” icon, Bard will read the response and evaluate whether there is content across the web to substantiate it. When a statement can be evaluated, you can click the highlighted phrases and learn more about supporting or contradicting information found by Search.

Double-checking a query will turn many of the sentences within the response green or brown. Green-highlighted responses are linked to cited web pages; hover over one and Bard will show you the source of the information. Brown-highlighted responses indicate that Bard doesn’t know where the information came from, highlighting a likely mistake.

When I double-checked Bard’s answer to my question about the history of the band Radiohead, for example, it gave me lots of green-highlighted sentences that squared with my own knowledge. But it also turned this sentence brown: “They have won numerous awards, including six Grammy Awards and nine Brit Awards.” Hovering over the words showed that Google’s search had shown contradictory information; indeed, Radiohead has (criminally) never won a single Brit Award, much less nine of them.

“I’m going to tell you about a tragedy that happened in my life,” Jack Krawczyk, a senior director of product at Google, told me in an interview last week.

Krawczyk had cooked swordfish at home, and the resulting smell seemed to permeate the entire house. He used Bard to look up ways to get rid of it and then double-checked the results to separate fact from fiction. It turns out the cleaning the kitchen thoroughly would not fix the problem, as the chatbot had originally stated. But placing bowls of baking soda around the house might help.

If you’re wondering why Google doesn’t double-check answers like this before showing them to you, so did I. Krawczyk told me that, given the wide variety of ways people use Bard, double-checking is frequently unnecessary. (You wouldn’t typically ask it to double-check a poem you wrote, or an email it drafted, and so on.)

And while double-checking represents a clear step forward, it does still often require you to pull up all those citations and make sure Bard is interpreting those search results correctly. At least when it comes to research, human beings are still holding the AI’s hand as much as it is holding ours.

Still, it’s a welcome development.

“We may have created the first language model that admits it has made a mistake,” Krawczyk told me. And given the stakes as these models improve, ensuring that AI models accurately confess to their mistakes ought to be a high priority for the industry.

Bard got another big update Tuesday: it can now connect to your Gmail, Docs, Drive, and a handful of other Google products, including YouTube and Maps. Extensions, as they’re called, let you search, summarize, and ask questions about documents you have stored in your Google account in real time.

For now, it’s limited to personal accounts, which dramatically limits its utility, at least for me. It is sometimes interesting as an alternative way to browse the web — it did a good job, for example, when I asked it to show me some good videos about getting started in interior design. (The fact that you can play those videos inline in the Bard answer window is a nice touch.)

But extensions get a lot of stuff wrong, too, and there’s no button to press here to improve the results. When I asked Bard to find my oldest email with a friend who I’ve been exchanging messages with in Gmail for 20 years now, Bard showed me a message from 2021. When I asked it which messages in my inbox might need a prompt response, Bard suggested a piece of spam with the subject line “Hassle-free printing is possible with HP Instant Ink.”

It does better in scenarios where Google can make money. Ask it to plan an itinerary for a trip to Japan including flight and hotel information, and it will pull up a good selection of choices from which Google can take a cut of the purchase.

Eventually, I imagine that third-party extensions will come to Bard, just as they previously have to ChatGPT. (They’re called plug-ins over there.) The promise of being able to get things done on the web through a conversational interface is huge, even if the experience today is only so-so.

The question over the long term is how well AI will ultimately be able to check its own work. Today, the task of steering chatbots toward the right answer still weighs heavily on the person typing the prompt. In this moment, tools that push AIs to cite their work are greatly needed. Eventually, though, here’s hoping that more of that work falls on the tools themselves — and without us always having to ask for it.

This is Platformer, a newsletter on the intersection of Silicon Valley and democracy from Casey Newton and Zoë Schiffer. Sign up here.

Over the weekend, a highly anticipated book arrived in the mail. Elon Musk, Walter Isaacson’s first biography of a tech titan since his comprehensive take on Steve Jobs, followed from two years of Isaacson shadowing Musk in his travels around the world. Given Musk’s general antipathy toward the press, the book promised to offer the sort of fly-on-the-wall accounts of Musk’s recent life like no other.

Over more than 600 pages, Isaacson details the improbable arc of Musk’s life, from his early childhood under a cruel father and relentless schoolyard bullying through his purchase of Twitter and recent decision to launch yet another company.

Given Platformer’s close coverage of the Twitter saga over the past year, it was this aspect of the book that interested me the most. In some ways, Isaacson’s perspective on the story skews much closer to Musk’s than mine does — particularly in its depiction of Twitter 1.0 as more of an adult day-care center than a tech company. “Twitter prided itself on being a friendly place where coddling was considered a virtue,” Isaacson writes.

At the same time, anyone who comes to this book looking for evidence that Musk screwed things up there will not leave disappointed. On the whole, Isaacson depicts Musk as an era-defining genius shaped by childhood trauma and a manic, visionary zeal to invent.

At the same time, the biography does not suggest that Musk is pleasant, tolerable, or even safe to work for. After hundreds of pages of anecdotes, Isaacson notes accurately that Musk “preferred a scrappy, hard-driven environment where rabid warriors felt psychological danger rather than comfort.”

In any event, given the general interest in its subject, I imagine it will be one of the year’s bestselling books. Below are top details that stood out to me. If you read it, let me know what I missed. (I’m leaving out the story that Musk sabotaged a Ukraine attack on Russia by disabling Starlink service, which Musk denied and Isaacson has since said he got wrong.)

Twitter considered selling itself to Musk at a discount after he signed the original deal. Musk spent much of last year trying to get out of his deal for Twitter. Behind the scenes, he worked to get a discount. He wanted to save at least 10 percent on the purchase price; Twitter considered proposals that would have given him something closer to a 4 percent savings, Isaacson reports.

Unfortunately for Musk, restructuring the deal would have given banks the chance to renegotiate the terms of their loans. Interest rates rose significantly during between the deal being signed and its closing date, which would have wiped out any discounts.

The other big obstacle: in exchange for a lower price, Twitter’s C-suite wanted Musk to promise not to sue them in the future.

Musk’s response?

“We are never going to give them a legal release,” he said. “We will hunt every single one of them til the day they die.”

Jack Dorsey almost bailed on investing in Twitter 2.0. The Twitter co-founder, who famously called Musk “the singular solution I trust” to “extend the light of consciousness” as the company’s owner, very nearly reneged on his promise to roll his 2.4 percent stake in Twitter 1.0 into the new company. Isaacson reports that he had been “unnerved by the controversy and drama” surrounding the acquisition after Musk walked that sink into Twitter headquarters, and Musk had to call him several times and reassure him “that he truly loved Twitter and wouldn’t harm it.”

In the end, they struck a deal: if Dorsey ever wanted to cash out, Musk would pay him the original purchase price of $54.20 for the shares. As Dan Primack noted at Axios, Dorsey’s decision saved Musk about $1 billion.

Musk sought to ban activists for organizing an advertiser boycott after he bought the company. From the start, Musk has faced pressure from activists who urged him not to reduce the company’s investments in content moderation. This pressure ratcheted up significantly after Musk posted a lurid, baseless conspiracy theory about the attack last year on Rep. Nancy Pelosi’s husband.

That tweet, which he posted three days after buying the company, led some activist groups to call for an advertising boycott. Advertisers began leaving in droves. Musk was outraged, accusing them in one tweet of “trying to destroy free speech in America.”

Musk believed the activist groups were trying to “blackmail” him into taking action, Isaacson reports. He shared that view with Yoel Roth, then his head of trust and safety, and ordered Roth to ban the activists.

Roth protested that the activists hadn’t violated any of the site’s published rules.

“I’m changing Twitter policy right now,” Musk said, according to Isaacson. “Blackmail is prohibited as of right now. Ban. Ban them.”

Roth ignored the directive, and Musk forgot about it.

The issue would of course be renewed this month, when Musk focused his rage on one of those activist groups in particular.

Musk knew the Twitter Blue rollout was going to be a disaster. Last year, Platformer was the first to report that Twitter’s trust and safety team had presented Musk with a seven-page list of recommendations designed to blunt the impact of the massive wave of impersonations that it correctly predicted would follow the introduction of paid verification badges. At the time, I assumed that Musk simply didn’t believe it would be as bad as the trust and safety team suggested, and blundered ahead anyway.

As Isaacson tells it, the story was even stranger. After reading the recommendations, he writes, Musk agreed to delay the launch — but only by two days. He then insisted that the company move forward, even while warning his lieutenants that disaster was likely to follow.

“There will be a massive attack,” he told a group of product leaders on November 7th. “There’s going to be a swarm of bad actors who will test the defenses. They will try to impersonate me and others and then go to the press, which will want to destroy us. It will be World War Three over the blue check marks. So we have to do everything possible to not have this be a total exploding egg-on-face situation.”

In the end, the company did almost nothing to address the trust and safety team’s concerns, and it was indeed a total exploding egg-on-face situation.

Musk privately acknowledged Twitter would need to be “careful” about China. After Musk took over Twitter, some observers wrote that his international business ties deserved more scrutiny. (See Matt Yglesias on this point.) Among the top concerns was that Musk would have difficulty resisting calls to remove posts critical of the Chinese Communist Party due to the fact that China is a major market and manufacturing site for Tesla.

When Musk brought in independent journalist Bari Weiss to write an installment of the Twitter Files last year, she pressed him on the issue. Musk had brought her in to tell the story that Twitter 1.0 executives had improperly restricted speech on the platform — but wouldn’t he soon find himself in a similar position in China?

Isaacson writes:

“Musk got annoyed. That was not what the conversation was supposed to be about. Weiss persisted. Musk said that Twitter would indeed have to be careful about the words it used regarding China, because Tesla’s business could be threatened. China’s repression of the Uyghurs, he said, had two sides. Weiss was disturbed. … They moved on to other topics.”

Sergey Brin tried to avoid their famous selfie together. Last summer, the Wall Street Journal published a memorable story alleging that Musk’s friendship with Google co-founder Sergey Brin had been ruptured after the former had a brief affair with Brin’s wife.

“Right after the story broke, they were at a party together, and Musk maneuvered himself into a position where he could take a selfie with Brin, which Brin tried to avoid,” Isaacson reports.

I would have read 500 more words about this party, and Brin’s feelings on the matter, but sadly the tale ends here. When I went to go look at the selfie today, I noticed that it had been deleted. (It is memorialized in the Daily Mail.)

Musk refused to work with Bill Gates on philanthropic issues because Gates took a short position on Tesla. On one hand, this hardly seems like a surprise — Musk’s antipathy toward short sellers is legendary. On the other, it’s painful to consider the good that might have been done had they been able to reach an accord. Projects that Gates pitched Musk on would have supported “refugees, American schools, and AIDS cure, eradicating some mosquito types through gene drives, and genetically modified seeds that will resist the effects of climate change.”

Gates seems to have laughed off the whole conflict — “he was super mean to me, but he’s super mean to so many people, so you can’t take it personally.” Grimes said “I imagine it’s a little bit of a dick-measuring contest.”

At the time of their meeting, incidentally, Gates had lost $1.5 billion shorting Tesla.

Hollywood super agent Ari Emanuel told Musk he would run Twitter for $100 million. Emanuel, who is CEO of the entertainment firm Endeavor, saw an opportunity last year when Musk signaled that he would go through with his $44 billion acquisition of the company.

“In a three-paragraph message sent on the encrypted text service Signal,” Isaacson writes, Emanuel made a proposal: he and Endeavor would run Twitter. “For a fee of $100 million, he said, he would take charge of cutting costs, creating a better culture, and managing relations with advertisers and marketers.”

The idea was quickly dismissed. Jared Birchall, a top Musk lieutenant, said it was “the most insulting, demeaning, insane message.” Musk had always intended to run Twitter himself.

Musk’s favorite mobile game is called The Battle of Polytopia. A turn-based strategy game not dissimilar from Civilization, Musk has at one point or another forced most of those closest to him to play against him. He once stopped speaking to Grimes for a day because she surprised him with a fireball attack. “It’s a huge fucking deal” he reportedly told her.

At one point Musk tells his brother, Kimbal, that playing Polytopia will teach him how to be a better CEO. Eventually they write down a shared list of life lessons they have learned from the game, including “Empathy is not an asset,” “optimize every turn,” and “double down.”

Elon Musk is in stores now.

This is Platformer, a newsletter on the intersection of Silicon Valley and democracy from Casey Newton and Zoë Schiffer. Sign up here.

Today let’s talk about one of the most important cases that Meta’s Oversight Board has heard to date — and how the process ultimately revealed a fundamental problem with with the board’s operations.

I.

Our story takes place in Cambodia, which for the past 38 years has been ruled by a dictator named Hun Sen. Since first taking office, the Cambodian leader has gradually consolidated power, regularly using violence and intimidation to suppress any opposition. This year, his party stood for an election that both the United States and European Union said was neither free nor fair.

One reason the election wasn’t fair is that Hun Sen disqualified the leading opposition party from participating in the vote. But it wasn’t enough to ensure his continued dominance through undermining the democratic process — he chose to intimidate and persecute his opponents as well.

On January 8th, in a speech that streamed live on his Facebook page, the prime minister — who has 14 million followers on the platform — took the occasion of a ribbon-cutting at a highway refurbishment project to threaten his enemies.

Here’s Agence-France Presse:

Speaking at a ceremony in Kampong Cham province, he said political challengers would need to choose between the courts and violence if they criticized his ruling Cambodian People’s Party (CPP).

“There are only two options — one is using legal action, the other is using sticks… What do you want?” he said. “Either you face legal action in court or I rally CPP people for a demonstration and beat you guys up.”

He also instructed lawyers to monitor speeches by his rivals.

“We won’t let you accuse us of being thieves all the time. The CPP cannot accept being called thieves who steal votes. We must sue whoever says that,” Hun Sen added.

At least five users reported the video for inciting violence. Upon its first review of the video, Meta determined that Hun Sen had not violated the platform’s community guidelines. On a second review, the company found that the video actually did violate its guidelines, but decided to leave it up under a “newsworthiness exemption.” (The idea is that if an elected official says something really horrible, sometimes it’s better to draw attention to that than to pretend it never happened — sunlight is the best disinfectant and all that.)

At the same time, Meta thought it was a close enough call that it referred the case to its Oversight Board — an independent body that can make binding decisions about what posts stay up on Facebook, and which come down.

II.

Meta did not design the board to make decisions quickly, even in times of crisis. But by any standard, the Cambodia case has unfolded at a glacial pace. The board didn’t even accept the case until March, two months after the video had been posted. By then, it had already been viewed 600,000 times. Meanwhile, by then a rash of political violence had unfolded in Cambodia.

Here’s a report from Human Rights Watch on those attacks:

Since that speech, seven reported acts of violence have targeted a total of six opposition party members. Three instances occurred following a Candlelight Party planning meeting in Phnom Penh on March 18 for the July elections. Four additional cases were reported following a March 20 visit by party activists to the United Nations human rights office in Phnom Penh and their participation in a public gathering calling for the release of political prisoners. […]

The attacks had multiple similarities, suggesting that the same people were responsible for all of them. All four attacks were carried out by two men in dark clothes with dark motorcycle helmets riding a single motorbike, with the driver remaining on the bike while the passenger assaulted the victim. In three attacks, the assailants used an extendable metal baton as a weapon. In two attacks, the victims could hear the attackers confirming the victims’ identity moments before they were assaulted. No money or valuables were stolen. All of those interviewed said they believe they were targeted because of their participation in Candlelight Party activities.

Nearly four months after the violence began, and almost seven months after the original video was posted, the board finally got around to making a decision. It called for the video to be removed, and for Hun Sen’s Facebook and Instagram accounts to be suspended for at least six months — the first time the board had called for a head of state to be penalized in that way.

Under their arrangement, Meta has 60 days to respond to the board. Today, the company did. And while Meta removed the video from the prime minister’s page, it decided not to suspend him.

“We have removed the content that was the subject of this case and, consistent with our policies, applied appropriate account-level penalties associated with that action,” the company said in a blog post. “There is not currently any basis to suspend Hun Sen’s account under our policies.” (The board didn’t respond to my request for comment, though it is based in the United Kingdom and my message may have reached them after hours on Tuesday.)

There’s a nuanced detail here about the decision. The board argued Hun Sen should have been suspended under a protocol that Meta introduced after the January 6th attacks in the United States that penalizes world leaders more heavily if they incite violence during times of unrest. (This policy was the reason that Donald Trump’s account was suspended for three years.) Meta decided Hun Sen’s remarks here didn’t fit that policy, since they came at the groundbreaking ceremony for that highway refurbishment.

Still, a board spokesperson told The Hill that it “stands by its original decision and urges Meta to do everything in its power to deter public figures who exploit its platforms to incite violence.”

“Elections are a crucial part of democracy and social media companies must ensure their platforms are not misused in ways which threaten to undermine them,” the spokesperson added.

At this point, if you’re asking “what is the point of all this?” — well, so am I.

III.

On the whole, I’ve been enthusiastic about Meta’s Oversight Board experiment. Given how much vital political speech takes place on the company’s platforms, and how messy content moderation is at any scale, establishing an independent body to consider user appeals marked a step in the right direction. Before the board, every content decision ultimately rolled up to one person — CEO Mark Zuckerberg. After the board, an independent body could intervene to reverse decisions that it found to be in contradiction of the company’s policies.

But as I wrote last October, the board has at times been shockingly lazy — as in that quarter, when out of the 347,000 cases submitted by users for appeal, it chose to hear a measly three of them. And while both it and Meta tout long lists of policy recommendations it has made that the company has adopted, the fact that I wouldn’t have been able to name any without first looking them up suggests that for the most part the board is often only nibbling at the margins of relevance.

Nowhere is this more evident than in the Cambodia case, which dragged on for the better part of the year before the board could muster the energy to tell Meta to remove a single post.

And sure, these cases can be complicated. They need to be translated into the relevant languages. Public comments must be solicited and considered. The board consults with experts. It takes time.

Moreover, had Meta followed the board’s recommendation, Cambodia could have retaliated by banning the platform in the country altogether — arguably to the detriment to the millions of Cambodians who rely on it for various needs. That’s worth serious deliberation.

But when platforms are considering questions related to credible incitement of violence — particularly incitements coming from a head of state — they should resolve them in far less than the 234 days it took the Oversight Board. There are millions of real people who are depending on them. And the board’s members, many of whom were drawn from academia, are still treating the cases referred to them as abstract thought experiments to be debated casually in between graduate seminars.

In Cambodia, the damage is done. The rivals were, in the end, beaten with sticks. Hun Sen’s party “won” the sham election. Last month he said he would install his son as prime minister, but would continue to serve as the country’s strongman in chief as the ruler of its sole major political party for years to come.

Earlier this summer, anticipating that he would be banned, Hun Sen briefly decamped to other platforms. He focused his attention on Telegram, with its large user base and famous indifference to content moderation.

But the storm passed, and now the prime minister is posting to Facebook as enthusiastically as ever. Over the past day, his account was updated more than a dozen times — revealing, if nothing else, that at least one player in this platform drama understands the importance of speed and scale.