This blog post by Nolan Lawson is giving me a lot to think about, but maybe not in the way he intended:

We’ll miss the feeling of holding code in our hands and molding it like clay in the caress of a master sculptor. We’ll miss the sleepless wrangling of some odd bug that eventually relents to the debugger at 2 AM.

I don’t miss that. I do have fond memories of late nights hacking code, but it’s the nostalgia of being young. I’m getting older and those days are gone for me.

If I could trade anything to go back, I’d spend even more time with my kids and less time writing code.

I know there are a lot of programmers who love the craft more than the output. Daniel Jalkut and I used to talk on Core Intuition about this difference among programmers. That’s fine, and the more sympathetic you are to Nolan’s perspective, the harder the AI era will be for you.

Programming is perhaps unique among creative pursuits because the user never sees all the work that went into it. Sculpting, painting, writing novels… Those will require a more careful approach so that students don’t skip the hard part. The hard part is the art.

I’m not too worried that young programmers will miss out on the experience of debugging every line of code by hand. There are other things to tinker with. LLMs still need so much hand-holding, there will always be something to do. I’m more concerned about the very real problem of young people staying up late mindlessly scrolling TikTok, creating nothing. AI at its best is giving us time, not stealing it.

Wired reporting on a proposed bill to pause building data centers in New York for three years:

Two New York lawmakers on Friday announced that they are introducing a bill that would impose a three-year moratorium on data center development. The announcement makes New York at least the sixth state to introduce legislation putting a pause on data center development in the past few weeks—one of the latest signs of a growing and bipartisan backlash that is quickly finding traction in statehouses around the country.

I’m sure the AI backlash will intensify. I’ve written before about how AI will become a massive cultural divide. People feel very strongly about this and we shouldn’t brush that aside as if it doesn’t matter.

But three years is forever. I’d rather see new legislation that adds incentives for renewable energy or limits on how construction and noise affect communities. Data centers are going to be built. If New York blocks them, they will be built elsewhere. Texas has a lot of space for solar and wind farms.

The AI industry has an insane amount of money right now. It won’t always be like that. This might be a once-a-generation opportunity to funnel that money into clean energy and new infrastructure.

I’ve seen a few posts recently about how developers will increasingly not even read the code that AI generates for them. Most famously, a blog post from Peter Steinberger:

These days I don’t read much code anymore. I watch the stream and sometimes look at key parts, but I gotta be honest - most code I don’t read.

Peter also posted a photo on Twitter / X that I think is real, although there have been AI-generated joke spin-offs of it, adding more and more monitors. In the real photo, Peter has a ridiculous number of terminal windows open as he juggles many concurrent OpenAI Codex sessions. Extreme multitasking.

I read everything AI helps me with. Sometimes I edit it. It hardly takes any time to skim through code to understand what it’s doing. It would be challenging to design and keep improving an app without having the basic architecture and some of the details in your head.

In a Mastodon thread, Steve Troughton-Smith compared AI-generated code with compiler-generated code:

Much as you don’t generally go auditing the bytecode or intermediate representation generated by your compiler, I think the idea of manually reviewing LLM-written code will fall by the wayside too. Like it or not, these agents are the new compilers, and prompting them is the new programming.

I enjoyed Steve’s posts exploring Xcode 26.3. But compilers are deterministic. When you input C, you can be confident that the machine code it eventually spits out is correct. Decades ago developers might need to drop down to assembly language to optimize something. The compiler is just never going to go off in the weeds and make decisions about the design or structure of your app in the way that LLMs do.

This blog post from Matt Birchler is more my take:

If you don’t know what tech stack you’re using or why things work in the first place, then if things go wrong and the computer isn’t able to fix it for you, then you’re in a sticky situation.

And this is why I think that skilled developers, while they may not be writing every single line of code in the future, they will still bring value by understanding systems.

As I blogged last night, successful developers will now need to focus even more on other aspects of their business. Writing and reviewing code is still going to be part of the job for the foreseeable future.

I’ve been building apps for the Mac and the web for 30 years. In all that time, especially as an indie developer hacking away on side projects at night, success was largely about how much could be coded. Everyone had good ideas, but not everyone could put them into shipping software.

If you could invest more time, you could do more. If you could grow a team, you could move faster.

This year, those constraints of coding skill and time are going away. We’re already seeing it in many new AI-assisted apps. So much is happening it’s actually difficult to keep up with.

Here’s what I’ve been thinking. If everyone can ship software, what will distinguish the successful companies from the apps that are lost in the noise? It’s no longer enough to just spend more time coding, or to be the first with a good idea.

I think it’s going to become even more important to grow other aspects of running a software business:

• Marketing
• Customer support
• Documentation
• Building trust
• Servers (speed)

Of course I’ll be writing a lot of code too, adding features, fixing bugs. But that’s the bare minimum now.