There’s a weird paradox in computing. Computers are really, really fast. We love that about them. They are fast enough to show video, play games, and search emails—all at once. A single server can process millions of transactions per day, for pocket change.

But all the growth, excitement, and interest happens where things are slow.

The web started out really, really slow—web pages that took a full minute to load over slow modems. Millions of hours of effort across industries made it really fast. Chips sped up, which meant routers, servers, and home computers sped up, too. Software was optimized. Data was compressed. Trillions of dollars in revenue and/or market cap showed up as a result. Moving faster means more transactions, after all.

The history of internet technology is filled with things that were slow and ponderous, and then unbelievable amounts of effort made them fast. Search used to take a while. Video was low-resolution and buffered endlessly. Browsers used to plod along. 3D games had like nine polygons.

The other side of the paradox: Fast gets boring. We take fast for granted. So when slow things show up, I get interested. Blockchain is a good example. One of Bitcoin’s major features is that it takes a long, long time to make a computer make magic space nickels with your computer. This made the coins rare and valuable. This was so weird that, among other reasons, people decided it was the future of the economy.

Personally, I didn’t enjoy it. I believe that CPU cycles should…mean something. Computers should probably not sit there burning fossil fuels trying to guess magic numbers so that people can buy ketamine. Another way to put it: CPU cycles are a modern currency and the exchange rate on blockchain is terrible. Computers should help us do things! Learn skills, contact friends, build better support networks so that if things get bad we know who can loan us a spare canoe to paddle the hell out of here. Sometimes I feel like a complete goof believing this, given everything that is going on.

AI is the new slow thing that is getting faster, which means some people are out of their minds about it—they think it’s gonna be worth trillions. Maybe it will! A trillion isn’t what it used to be. And a lot of my fellow old-school web nerds—who sat out blockchain—are going, “Man, this is it, huh? This is the real thing.” They’re poking around and finding all kinds of ways to use this technology. One friend told me he loves it because he can use AI tools to rewrite his emails to sound like a white middle-class American (he’s not one of those, but he works with a lot of them), and that is amazing for his career.

I would bet that faster AI means it’ll be taken for granted, baked into everything, and more like cheap apps than a world-changing supertechnology. Because AI models are kind of…little. They fit in tens of gigabytes. You can run them locally. They keep speeding up. They’re…software. The magic will fade. Do we really, as a society, care that much if Ant Man VII: The Antening (2029) is made with AI-powered CGI? We may say we do, but I kind of doubt the government will intervene.

I try to remember that all real progress is slow. I’m learning piano and I’m now to the point where I can play the really simple Bach/Petzold Minuets, and I cannot believe how slow my fingers are. I am a fast-talking New Yorker but my fingers seem to be living in rural Illinois. What I would give for even Chicago fingers. Meanwhile, on YouTube, nine-year-olds absolutely crush Chopin, while I watch like a golden retriever trying to figure out where they’re hiding the ball. Nonetheless, I am getting better. In a couple of years, I will take over a bar and we will have a truly horrific night of the world’s worst interpretations of soft rock classics. You’re all invited.

Right now we’re roping together a new version of Aboard. Real soon now! REAL SOON. Software is…slow. Slow to develop. Slow to test and fix. But I do take comfort in the fact that, when things are slow, that can be the sign that you’re on the right track. Maybe! Things that are fast to build tend to be disposable (think of all those Ruby on Rails sites in the mid 2000s). I know there’s a lot to be said for doing things quickly and moving on to the next one. But in this fastest of all industries, the reality is that the stuff that is really slow, difficult, and complex to pull off has the most impact.

And of course, if it works, you’ll take us for granted. Like that guy at the piano bar singing for tips. You don’t think about him practicing for years. You just sing along with “New York State of Mind.” And he’s happy, too.