I am continuing work on "Tidy First? An Exercise In Empirical Software Design" as a @substack . No paid tier yet, but that's where book chapters will be appearing. Sign up here: kentbeck.substack.com/.

"Daddy, what was vim?" For about 80 years, the motive force behind software was humans writing in formal languages, using their fingers and their brains. It is now obvious we are going to go on making software, and that it is going to be coding models doing the vast majority of writing, debugging, and increasingly designing that software. So what now? I am fortunate to have some great people to Talk Computers with in my life, and while we haven't found the answers yet, we at pebblebed have begun recording these discussions, under the working title "After Coding". Episode 1 is the inimitable Josh Wills. Do give it a listen, won't you dear catcher of cast pods? open.spotify.com/show/6bsC4D…

Keith is the best engineer I ever worked with and we got to hit on all of my favorite topics in this conversation: pure math, chess, engineering, and what all of this means for the kids.

Effective leaders are skilled business operators first, and kind people second. Charity Majors(@mipsytipsy), cofounder & CTO of Honeycomb: "Twitter had 16 years to figure it out and everyone could see that they were not figuring it out. You could argue that X has gotten better or worse, but you can't argue that Musk is running it with 20% as many people. And it's working. You could argue, and I think it would be true, that it's some of the work that those engineers did that allowed this. But this is the point. If we don't do it ourselves, meaning hold ourselves to a high standard, build with efficiency, constantly be like trying to get better, someone will come and do it to us. In the 2010s, there was so much money sloshing around in Silicon Valley and then times started to get tough and all of these companies canceled their DEI programs. They never believed in that. They were just trying to buy people off. I have taken a lot of lessons away from that: it's not enough to be a good person. I believe that people who are kind and care about people can, and usually do, do better than sociopaths in the same roles, but only if they are good at business."

So glad to be part of this. Building trust while we build code is the inner loop of our future.

All models, no matter how smart, will eventually build systems that they can no longer understand or maintain, if you let them. Fable 5 finally outbuilt itself, and flailed on me for a week. Fable 5.1 looks like it will fix it. For now. But you have to keep an iron grip on system size, or it'll run away from you.

I think Dwarkesh's language in this piece is a reflection of his understanding of the human experience.

Just @KentBeck saying how it is

My daughter Beth, on writing self-driving software in 2008: They told the car "if stuck, relax constraints until you're free." It hit a roadblock, thought a while, and relaxed "don't drive on sidewalks." So it drove up onto the sidewalk. "Exactly what I told it. Not what I wanted."

I asked my daughter Beth: how do we get the AI excited about writing tests? "First it needs an endocrine system." The model has no enthusiasm of its own. What exists is us plus the genie. The caring is still ours.

A bonus Still Burning with my daughter, Beth Andres-Beck. I asked when she first knew she was a geek. Her answer: her first pun, at the fridge, "but that one's cheddar." What she remembers is that I laughed. That was the start of all of it.

My daughter Beth went seven years without writing tests, then converted a whole team. The trick in code review: stop pointing out bugs, start pointing out the missing test. Every test caught a bug. Her line: it's your fingers on the keyboard. Write more tests out of defiance.

Org-design idea from my daughter Beth on Still Burning: the Guild of Guilds. Groups who do similar work compare notes. Then one more guild, of all the guild leaders, catches the company-wide problems no team can see. How you get recruited? Curiosity. It's a trap.

From my bonus Still Burning with my daughter, Beth Andres-Beck: Why do we keep building agents to run without us? Beth's answer: there's a nervousness about including people at all. We call things "automated" so no one has to be responsible.

If you are Exploring (in the 3X: Explore/Expand/Extract sense), you should be far more afraid of silence than failure. Silence, continued, is game over. Failure, addressed, is another shot at success.

On Still Burning with my daughter Beth: a driverless car makes an illegal U-turn, gets pulled over, and there's nobody in the car to ticket. Her reframe: it's not "the computer" driving. It's software a person wrote, under someone's incentives. The responsibility didn't vanish.

My daughter Beth on Still Burning: A kid learns to count past thirty and says "31." Another asks, "How do you know that?" The first kid is smart. The second one goes further, because she's asking how the solving is done. That's meta-intelligence, and you can learn it.

My daughter Beth's joke: make programmers "work to rule," so they're forced to refactor, test, integrate, collaborate. The problem as a strike is you'd get more productive. The part we stopped laughing at: teams did it, shipped great software, and got fired anyway.

I asked my daughter Beth what keeps her up at night. Not the robots doing whatever they want. The robots doing exactly what some people tell them to. We're building tech that could be a personal army, where you don't need anyone else's help to do great harm. #StillBurning #TechPodcasts

New bonus episode of Still Burning is out. My conversation with my daughter, Beth Andres-Beck. Engineer, ex-self-driving-car programmer, organizer, now running for Congress. On tests, AI agents, and who's accountable when nobody's in the car. Listen and watch: share.transistor.fm/s/54f009…

What would happen if we used the Genie not to program faster, but to master programming faster? Would we get more or less accomplished along the way?