Explaining Culture to Technology, Paul Ford | Compile 26

Wow. Good, everyone. Nice to see you. Thank you for having me.

While I am doing this, I will tell you, I am someone who now is a technologist, who vibe codes all day and does all the other stuff that you would expect. And I've been in software for a really long time.

But I've also been a magazine writer and editor for a really long time. And what happened is, Cursor called me up and they said, "Hey, you often explain technology to people in culture.

Would you come out and explain culture to people in technology?

And you will have 15 minutes." And I said, "Sure.

Why not? That actually sounds kind of fun." So

we're going to see how I do.

They also told me I was allowed to make it as weird as I wanted, so I did that, and we'll see if any of this makes sense, and it's okay if it doesn't.

I actually didn't need that one. All right.

So D-L-P-M.

I used to work at a bunch of different magazines, and I've written a bunch, and I thought I would just describe how a magazine works. Because another thing that happens, one of the ways I get called to

works. Because another thing that happens, one of the ways I get called to explain the world of culture to technology is sometimes people have amazing outcomes and IPOs and they're like, "I really want to buy a magazine."

And they get in touch with me as a magazine person, and I'm like: Don't do that.

It's not what you think. It's a really bad idea.

Sometimes they do it anyway. Anyway.

People think of a magazine a lot of times as its output.

They think of media as its output.

They think of it as a set of articles, a nice cover, some illustrations, things are really good, they're very carefully written. That

is not how I see it.

That's not how most people who work in magazines see it.

I see it as a weird network between a group of writers and freelancers and editors and their readership. It's kind of less than it is an artifact or a document.

It's a shared understanding. And so

it's kind of a distributed, weird network as opposed to something that gets published on a regular basis.

And I've worked for "Wired," I've worked for "The New York Times," I've worked for "Business Week," I've worked for all sorts of folks.

And I think everybody who works with them would pretty much agree.

But when I come out here, people see it in a different way.

They sort of see the magazine as a feed.

And I'll give you another idea, which is that I was editing my first piece as a national magazine editor.

This is way before I became a more serious technologist.

And I went to my editor, and it was a piece about Guantanamo Bay, and it was written by a lawyer, and I was like, "Look, I have organized all the facts.

Everything is fact-checked. Everything is beautiful, and they're all right here in the piece." And he read it and he said, "Wow, you got a lot of facts in here." And I said, "Yes, I do." And he said, "Good for you.

here." And I said, "Yes, I do." And he said, "Good for you.

You're a good boy.

But get rid of all of them." And I was like, "Well, that feels bad." And he's like, "No, we have a contract with the reader.

The reader understands who we are.

They understand what we stand for, and they trust us.

They don't have time for all that. Give them only the rhetoric.

Give them the rhetoric and assume that the facts will follow."

Rhetoric greater than facts.

So when I first started writing, I used to think, "Oh my God, I'm programming people's brains." I was very proud of myself. I'm like, "You're reading my stuff.

brains." I was very proud of myself. I'm like, "You're reading my stuff.

I'm in your head." But actually, over time, what I realized is nobody cares about me. They care about what I write as far as

about me. They care about what I write as far as they can use it to simulate and understand the world that they live in.

They like to use it to predict their own future.

And it was a blow to my ego, but I think it's also real that people don't let me live rent-free in their head.

They use me, and they use my ideas in order to figure out how they want to live their own lives. And I think that's true about most media.

I don't think we read fiction simply for the joy of reading fiction.

I think we read it to simulate and understand the world that we're in and think about how we might behave in certain circumstances.

And so that's magazines. What does that have to do with this conference?

that's magazines. What does that have to do with this conference?

Maybe we'll find out. Now I want to talk to culture.

Okay? Because that was the question. Can you explain culture to technology?

Culture lives inside of everybody's brain in a kind of shared way.

I don't think that's some shocking understanding.

But I was walking around on Mission a couple of days ago, and I went and the Databricks is having its enormous conference, and good for them.

It's not as pretty as this. And they were right across the street from the Catholic Church, and it's kind of a funny thing to think about right?

Because the Catholic Church is many things.

It's also a big publisher on AI. It's a good encyclical if you want to read it.

But the Catholic Church's basic function is a little bit of a predictive model.

You use this one book, you double down on it, and it kind of tells you how things are going to go, especially after you die.

And at some level, when you ask which is going to be around longer, right now, I think San Francisco is spending a lot more time thinking about Databricks than the Catholic Church.

But I do think the Catholic Church has a pretty good predictive model that people really lock into and that they spend their whole lives inside of.

So it's pretty good that way.

Let me go back. I'm going to throw out 14 ideas and not really connect them very well. No, I'm counting on you. You're a bunch of nerds.

very well. No, I'm counting on you. You're a bunch of nerds.

This is your whole thing.

I know you're going to connect them for me.

So where do we come from with all this stuff? What is this culture?

What is a predictive model? Why are we talking about this?

My favorite model of consciousness is actually Richard Leakey's, who's an anthropologist. He wrote about this in 1994,

anthropologist. He wrote about this in 1994, and the model is as follows: One monkey's looking at another monkey across the bay, across a pond, and the other monkey's got a banana.

And the first monkey is like, "Boy, I would like that banana.

That looks amazing. I'm going to wait until he turns, and then I'll grab that banana." And that particular moment is when consciousness starts to show up,

banana." And that particular moment is when consciousness starts to show up, because that monkey is simulating the other monkey and saying, "Hey, I'm not just going to grab the banana. I'm going to wait for a preexisting condition based on that other monkey's internal state.

I get that banana, my reproductive fitness increases." This

is a very smart monkey. He gets to have more children that are good simulators.

monkey. He gets to have more children that are good simulators.

Eventually, the simulation turns inward and consciousness arises. That's fun.

That's one minute out of my 12. I'll just let you guys think about that.

But when- Putting all that together, which is your job, not mine, we've got culture as a kind of operating system where there are shared states in the brain. It's very lossy. It's a very lossy system.

No one has all of culture in their brain.

And media becomes almost the file system for culture. When you want to experience meeting media,

for culture. When you want to experience meeting media, what you're doing, you watch a movie, you read a book, you load it into your cultural brain, you experience it, and then you tend to put it away or throw it away.

And that is a way that you get to participate in the world without a lot of risk. It lets you explore your own anxiety,

risk. It lets you explore your own anxiety, where your relationship is, what your relationship is to status, kind of who you are.

And so that's what I think culture is.

And, oh, I forgot to put in these words, distributed lossy prediction model.

Woof.

All right. Prediction model. That's what that says there.

Don't worry about that. I'm not used to working with chalk.

Now we come to the tech industry, which we all know and love.

Let me narrate where we were before this year, okay?

I always felt that technology is an industrial process.

It's a risk reduction process, and when you work with high-level people, all they talk about is risk because software doesn't want to ship.

It hates shipping. It's like the one thing, when software is born, it never wants to be produced and actually put into the world.

And so we've spent the last 50 years building Agile and building different methodologies and just begging engineers and all the things that we have to do in order to get software to cross the line.

So you have this very formal culture based around risk, where the whole function was for the software industry to make the computer, which is impossible to work with, just impossible, make the computer accessible to the culture, figure out word processing, figure out the web, and it was a priesthood.

It was a very sort of high level, very specific thing that very few people could do. I was very proud of it. I liked having skills and craft, but it

could do. I was very proud of it. I liked having skills and craft, but it changed. Okay? And now today, we're in this very funny zone where

changed. Okay? And now today, we're in this very funny zone where I find-- I vibe code as much as anybody in this room.

I'm doing it all the time. I do it on the plane, which is one of the places where I feel really bad about my carbon footprint.

I ate a hamburger on the plane while vibe coding, and I felt that that was probably one of the lowest moments of my life.

Then I'm Googling offsets on the plane, and I'm like, "This isn't good either."

Anyway, I'm finding very much as a magazine writer that my technology production, which always used to be the form-- I was good at my stand-ups, I was good at Agile, I was good at shipping, I was good as a programmer, I was good as a manager.

I'm finding that my technology experience is a lot more like prose production.

It's a lot more like what I used to do over here, and that's spooky and confusing because prose is a teeny-tiny industry, and programming is trillions and trillions of dollars.

So I don't know how that all works out, but it'll work out great.

The other thing is that everybody worries about slop and inaccuracy, but to me, those feel like no big deal because that's actually what an editorial process is for. Every first draft is an atrocity.

is for. Every first draft is an atrocity.

Everything a writer creates is an embarrassing, moronic disaster, and that includes me. When I write for The Times now,

includes me. When I write for The Times now, I've been doing it for 500 years, they still rip everything to shreds and somebody younger than me, and I'm just regularly humiliated, and that's the process.

But slop is really easy to get rid of, if you have a good editorial set of tools or frameworks for getting rid of slop. Slop is actually omnipresent.

slop. Slop is actually omnipresent.

It wasn't invented by AI, and that's because AI, even though we keep talking about LLMs in terms of consciousness and the goal was to create consciousness, if you think about it, they're actually just all this stuff, all the media that is produced by the culture. It's the file system for our culture.

It is being compressed into an LLM, which we'll just call it culture.zip.

And we're in this very funny place where this very formerly formal culture, which was very driven by process, is now being driven by individuals using a culture simulator to produce artifacts that are using methods and approaches that are actually a lot like media. At least that's how it feels to me.

And so that's sort of my provocation for the room, and I'm going to just close with a little story. When I was a kid, my father was an English professor, but he was also a big nerd. Go figure.

And he sat me down, and he was like: Check this out.

And he showed me a basic program. I was about 10. And he said, "This is weird.

It's like a poem in that you are compressing as much information as you can into this one little space in order to get it to really accomplish something.

And poems are like that, too." And that stuck in my head for 40-some years, and it always felt like something that was very, very hard to articulate, but I think it actually is starting to make sense now.

I think you work with culture in the same way that writers and artists work with culture as a programmer going forward.

And I think that the processes that we're using, which are more industrial, are probably no longer as relevant and that a lot of the processes we're going to be using in the future are going to be more about creating media.

And I am out of time, but that is something I just want to leave you with.

That maybe we're a little more like this side of things and a little less like a big formal industrial process.

And I think that's exciting and interesting, and I am really grateful for your time.