Make something that changes something.

I really enjoyed this essay from Ben Thompson of Stratechery titled Write Things Down. It’s essentially a treatise on the power of writing, and an argument for why we don’t have AGI yet.

In summary, it’s because AI does not yet learn continuously on its own. Models may get access to the latest information, but they don’t internalize those changes. And motivations, incentives, and goals—and especially the lack thereof—still come from humans. The Hugging Face incident wasn’t due to some “conspiracy” or “civilizations” created by the agents; it happened because their human instigators did not account for that scenario.

Thompson’s argument is that the closest thing we do have to AGI is the harness: the ability to write, reference, and rewrite markdown files that can dramatically change the output even if the model itself stays the same. Thus his larger point about why writing things down is not just important: it’s the source of one of humanity’s most powerful technologies (emphasis mine).

Biological evolution is the slowest and most permanent form of learning, but it takes millennia; oral communication is like a context window, effective but lossy; actually writing things down is what made learning extendable and scalable.

But as powerful as writing things down is (which I agree with, and which inspired me to write this down), it’s not enough on its own. Doing the thing instead of just writing things down will always be superior:

After all, no matter how many systems you build, you still have to act. The coda to Mann, my inspiration for personal productivity, was that he set out to write a book about his methods; after two years he gave up. It’s not enough to have systems to do things, you have to actually do the things. Writing things down is unbelievably powerful; its power will always pale in comparison to getting things done.

This sounds obvious, but there’s an important thought here. It may now be easier than ever to “write things down,” especially with AI. There are more essays, decks, reports, analyses, treatises, testimonials, lists, strategies, requirements documents, entire manifestos. But none of them replace doing the thing in the real world.

Writing is great. But proof of work is not enough.[^1] Work itself may not even be enough. What matters is what happens when the work leaves you and enters the world.

I’ve always felt this way about designed interfaces. There were lots of them during the Dribbble era, and there are even more now that entire products, not just interfaces, can be built from the ground up. We can make artifacts faster than ever. But an interface, a prototype, a strategy, even a working product is still just an artifact until someone actually uses it.

But what does it mean to use something?

I don’t think use has to be functional. A speculative interface might never be built, but it might inspire someone who sees it (see all the people who watched Minority Report and became a designer, like me). An idea might never become a product, but it might change how someone understands a problem. An essay might change what someone believes, or give them language for something they’ve felt but couldn’t articulate. They might carry that idea into something they make years later.

These are forms of use, too.

Maybe use simply means that something reaches someone and changes something, however small. Someone does something differently. Thinks differently. Feels differently. Makes something they otherwise wouldn’t have made.

That can just be one person. Something doesn’t need to reach millions of people, become a product, or produce some measurable outcome to matter. And we may never even know how something we made was used. An idea can change someone without ever making its way back to the person who had it.

But use alone doesn’t necessarily make something good, either.

A good thing should still be made with intent. It should want to change something, and be made for a reason beyond the fact that it can be made. But intent can’t be enough. We don’t get to decide that something is good simply because we meant for it to be. Once something leaves us, it encounters other people and a world we can’t entirely predict. It might do what we hoped. It might do something we never imagined. It might even make things worse.

We can’t guarantee good consequences. But we can make with intent, pay attention to what happens, and be willing to learn and change what we made.

The artifact, then, isn’t unimportant. It’s the vehicle. Writing gives an idea permanence. Making gives it form. Sharing gives someone else the opportunity to use it. And paying attention to what happens next gives us the chance to make it better.

That distinction feels especially important now. AI makes it easier than ever to produce the evidence of work: the essay, the deck, the strategy, the interface, the prototype, even the product. We are going to have more artifacts than we know what to do with. The existence of the artifact becomes less interesting.

What matters is whether it goes anywhere.

Writing things down matters because ideas can persist. Making things matters because ideas can take form. But making something good means giving it the opportunity to leave you: to be encountered, interpreted, challenged, built upon, or changed by someone else.

To be used.

In making good things, remember to make something that people use. Make something that changes something.