This comment about the OpenClaw guy hits a little too close to home:
“Peter Steinberger is a great example of how AI is catnip very specifically for middle-aged tech guys. they spend their 20s and 30s writing code, burn out or do management stuff for a decade, then come back in their late 40s/50s and want to try to throw that fastball again. Claude Code makes them feel like they still got it.”
What an ageist quote. I am in my 40s and never stopped coding even as I've become the principal engineer. Claude just frees me from the mundane tasks I'd done a million times before and never wanted to do again if possible, which it now is.
I can still throw a fastball without AI, but why would I when I can throw it much faster, with much less effort now, while still enjoying what I am doing?
It's still coding. If you think it's not you probably think that letting the IDE auto-complete or apply refactorings is also not coding.
writing any git command, ever, writing any documentation, ever. writing comments in issue trackers, resolving issues in issue trackers, doing pretty much anything in the terminal, ever… basically every imaginable thing which takes time away from the actual job
Why not say “using a computer”. gcl (my alias for git clone) is way faster to use than any prompting. Any use case I found for LLMs, I noticed that a good script or a DSL (as an abstraction) would be way more useful.
A lot. I often study software I use (mostly OSS) to find how a feature is implemented.
If something is cumbersome and I find myself needing it often (or I think I will need it), I write an alias, a script, an emacs function, etc,... That's the magic of reducing lot of steps to a single button press (or a short command).
Yes. Older people do not become less inquisitive and eager to learn, they just become less open to hype as they've seen whatever younger folks think is the new hot idea several times before, just in different shapes and sizes.
However, with AI we're truly seeing something new that we had not seen before (the AI of the 80's, 90's and 2000's was interesting but it never managed to do anything truly generalist - it was mostly able to get good at a very narrow, specific activity, very different from today's LLMs), so I feel just as curious and eager to "learn it" as I was eager to learn, say, Functional Programming in my 20's and Neural Networks in my 30's.
Fluid intelligence peaks early (20s) but crystallized intelligence peaks much later (50s-60s), and it's not like you can't crystallize a desire to continue to learn, even if you're potentially less creative from a raw intelligence perspective.
Same but for me it's 25 years of accumulated personal backlog that I'm finally burning through. Like I've been a project hoarder and now I have a house elf to tidy up and do all that widget fobbering business. I just need to figure out what the rules of the house are.
This idea of LLMs a vehicle of midlife crisis is fascinating. I'm not sure if it's just about "throwing the fastball" though. Most of the usual midlife crisis things are a rejection of virtue. For example: buying a porsche, pickign up a frivolous hobby, or cheating on your wife, these are irresponsible uses of money, time, or attention that a smart, dedicated, family man wouldn't partake in.
In relation to LLM usage I think there's two interepretations. 1) This midlife crisis is a rejetion of empathy, understanding, and social obligation however minute. Writing a one-sentence update on an issue, understanding design decisions of another developer, reading documention are all boilerplate holding them back from their full potential in a perfectly objective experience. Of course, their personal satisfaction still relies on adoption of their products by customers (though decades of viewing customers through advertising surveillance has stripped away the customers' humanity from their perspective). Or 2) economic/political factors such as inflation, rising unemployment, supply chain issues, starvation of public services, and general instability means doing the usual midlife crisis activities are too expensive or risky, and LLMs present a local optmimum allowing them to reject societal virtues (eg. craftsmanship, collaboration, empathy) without endangering their financial position. Funny enough, I feel this latter point was also a factor of the NFT bubble (though, the finances were more clearly dubious).
Because they don't "got it". Asking the bot to program is the same as asking a junior engineer to write some code, and then claiming it as your own. It's not actually them programming. Just a misplaced sense of pride.
More gatekeeping, more no true Scotsman fallacies, more bitter cope.
You can absolutely take pride in having raised your own cows. But the guy down the street can also take pride in having cooked his own steak. In fact, the guy down the street might actually be a better chef than you, even though you know how to breed cattle.
You're wrong because you are making the wrong comparison.
In this analogy, The guy down the street didn't cook his own steak. He told someone else to cook the steak. And then claimed that he himself cooked it. Telling himself, "wow, I'm a great chef!". When In fact, he did not cook the steak.
Your greatness as a chef isn't measured by how well you manage restaurant kitchens. That would be a great manager. Your greatness as a chef is measured by actually cooking yourself. Claiming other chef's work as your own would be dishonest and self-deception.
If we want to stretch this analogy a bit - I believe all world-level chefs have a team of sous-chefs working for them. Doing things like chopping ingredients, prepping things, in fact probably doing a lot of th cooking. I think building with ai is pretty similar.
This is the exact analogy that Gene Kim and Steve Yegge used throughout their book Vibe Coding: Building Production-Grade Software With GenAI, Chat, Agents, and Beyond.
You have it completely backward, in fact in the culinary arts your greatness as a Chef is entirely dependent on being a manager of restaurant kitchens.
You get judged on the final end product, the full dining hospitality experience, as had by influential customers on a random night (like Michelin inspectors).
The food is just one factor of that experience, and the overwhelming majority of that food on any given night is not actually prepared by the chef with their name on the door, but by his/her staff (the AI Agents in this analogy).
I get it. Knowing good code and how to correctly build software that people actually want is experience that is consistently hampered by constantly having to learn yet another tech stack.
Using an LLM lets you quickly learn (or quickly avoid having to learn) yet another tech stack while you leverage your inherent software development knowledge.
This is it for me. I burned out on chasing the latest stack about 12 years into my career. I went into management and concerned myself with system design, product design, and process design. LLMs let me use that knowledge to build things I care about: features and products, without getting (too) bogged down in things I don't: super elegant code in the hot new framework that users will never see or pay for.
This describes me nearly perfectly. Though I didn’t exactly burn out of coding, I accidentally stumbled upon being an EM while I was coding well and enjoying. But being EM stuck so I got into managing team(s) at biggish companies which means doing everything except one that I enjoy the most which is coding.
However now that I run my own startup I’m back to enjoying coding immensely because Claude takes care of grunt work of writing code while allowing me to focus on architecture, orchestration etc. Immense fun.
Me too, only I'm "only" 42! Got my first job as a programmer at 18 and (in retrospect) burnt out at some point and thought going into managment was the fix.
“Peter Steinberger is a great example of how AI is catnip very specifically for middle-aged tech guys. they spend their 20s and 30s writing code, burn out or do management stuff for a decade, then come back in their late 40s/50s and want to try to throw that fastball again. Claude Code makes them feel like they still got it.”