I really don’t use Substack at all, but your joy and curiosity posts show up every Sunday in my mail. What used to be “scanning before delete” has turned into “make coffee and read joy & curiosity”.
Grateful to have intersected on an interest in interpreters and compilers with you, because it lead to getting great perspectives on the software world every week! You are great at finding the fun (joy) in what we all do. Thank you!
Hey Thorsten. I am a bit sad to see that every post here starts with "everything is changing" fear-mongering. I get that you need to market Amp, but isn't it too much?
And before you say that you are simply doing your duty as a good Samaritan by letting people see the true reality, here's an honest question. How long do you think it would take someone who has solid programming experience but has never used agents before to get the hang of a tool like Claude Code?
You got it backwards. I'm not doing this to market Amp. I'm working on Amp (literally created it with Quinn!) because I think it's a good idea.
If someone believes that solar energy is the future, and they create a company to harness the power of the sun, and then they write a personal newsletter — don't you think they'd talk about solar power?
If it comes across as fear-mongering, I'm sorry. My goal is to try to help people understand what's going on and in the past (I wrote two books) my explanations of how I see the world and how I understand things have resonated with others.
Here's what I wrote on Twitter on that:
> What I see as the tiger is AI. What I see is people on here saying "code was never the bottleneck, haha, this is all old news" and "AI can't do what John Carmack can do" and implying or sometimes even explicitly saying that this won't change software development as an industry.
> And I'm saying: you can't judge what AI is going to do to this industry by looking at extreme outliers. AI will change software, the industry, fundamentally, because it's better at writing code than the average developer. It's changing a very fundamental value in a big equation.
> And what I aim to do and have been trying to do for years now is to help my peers clearly see what's happening and what's coming. Because I think that's actually more kind and helpful than keeping your eyes closed and pretending like the average (!) value of software and software engineers [isn't] changing tremendously.
---
As for: how long would it take someone who has solid programming experience but has never used agent to get a hang of them? I'd say it correlates with their ability to think about software products, business, problem solving, and their communication skills.
I've seen many engineers struggle with agents and others just fly from the get-go.
> If someone believes that solar energy is the future, and they create a company to harness the power of the sun, and then they write a personal newsletter — don't you think they'd talk about solar power?
Of course, but I would like them to focus on solid graphs that show savings and efficiency, rather than saying that people who still use gas are blind.
And I doubt that your books said that if you don't study compilers and interpreters, the world will eat you alive, right? :-)
Perhaps I am misinterpreting some things, but these are my impressions from your recent media presence.
Where am I saying that people are blind? I'm not. This is a free newsletter that everyone can choose to read or not. I'm sharing my perspective here, that's it.
I'd say, yeah, you're misinterpreting or misunderstanding a lot of things.
Many, many, many times have I given advice here in this very newsletter about how to see what AI can do and what it can't do. I have videos online in which I program using AI. I ship a product many times a day that's built using AI. Don't ask me for "solid graphs."
> Where am I saying that people are blind? I'm not.
In #72 you end with (emphasis mine): "It’s time to see for yourself, **with open eyes**, what these models can and can’t do, and you won’t get a good look if you don’t push them hard enough in all directions."
That's a strange interpretation. Saying "look!" is not the same as saying "you're blind and don't see anything." There's quite a few people I personally know who have used agents half a year ago only, vastly underestimated the progress they've made and following from that vastly underestimated the progress they will make, and then were very surprised after giving them another spin in the last few weeks. That's why I'm writing this: I don't want people to be surprised.
I really don’t use Substack at all, but your joy and curiosity posts show up every Sunday in my mail. What used to be “scanning before delete” has turned into “make coffee and read joy & curiosity”.
Grateful to have intersected on an interest in interpreters and compilers with you, because it lead to getting great perspectives on the software world every week! You are great at finding the fun (joy) in what we all do. Thank you!
Very happy to see Geese infiltrate the software space
Hey Thorsten. I am a bit sad to see that every post here starts with "everything is changing" fear-mongering. I get that you need to market Amp, but isn't it too much?
And before you say that you are simply doing your duty as a good Samaritan by letting people see the true reality, here's an honest question. How long do you think it would take someone who has solid programming experience but has never used agents before to get the hang of a tool like Claude Code?
You got it backwards. I'm not doing this to market Amp. I'm working on Amp (literally created it with Quinn!) because I think it's a good idea.
If someone believes that solar energy is the future, and they create a company to harness the power of the sun, and then they write a personal newsletter — don't you think they'd talk about solar power?
If it comes across as fear-mongering, I'm sorry. My goal is to try to help people understand what's going on and in the past (I wrote two books) my explanations of how I see the world and how I understand things have resonated with others.
Here's what I wrote on Twitter on that:
> What I see as the tiger is AI. What I see is people on here saying "code was never the bottleneck, haha, this is all old news" and "AI can't do what John Carmack can do" and implying or sometimes even explicitly saying that this won't change software development as an industry.
> And I'm saying: you can't judge what AI is going to do to this industry by looking at extreme outliers. AI will change software, the industry, fundamentally, because it's better at writing code than the average developer. It's changing a very fundamental value in a big equation.
> And what I aim to do and have been trying to do for years now is to help my peers clearly see what's happening and what's coming. Because I think that's actually more kind and helpful than keeping your eyes closed and pretending like the average (!) value of software and software engineers [isn't] changing tremendously.
---
As for: how long would it take someone who has solid programming experience but has never used agent to get a hang of them? I'd say it correlates with their ability to think about software products, business, problem solving, and their communication skills.
I've seen many engineers struggle with agents and others just fly from the get-go.
> If someone believes that solar energy is the future, and they create a company to harness the power of the sun, and then they write a personal newsletter — don't you think they'd talk about solar power?
Of course, but I would like them to focus on solid graphs that show savings and efficiency, rather than saying that people who still use gas are blind.
And I doubt that your books said that if you don't study compilers and interpreters, the world will eat you alive, right? :-)
Perhaps I am misinterpreting some things, but these are my impressions from your recent media presence.
Where am I saying that people are blind? I'm not. This is a free newsletter that everyone can choose to read or not. I'm sharing my perspective here, that's it.
I'd say, yeah, you're misinterpreting or misunderstanding a lot of things.
Many, many, many times have I given advice here in this very newsletter about how to see what AI can do and what it can't do. I have videos online in which I program using AI. I ship a product many times a day that's built using AI. Don't ask me for "solid graphs."
> Where am I saying that people are blind? I'm not.
In #72 you end with (emphasis mine): "It’s time to see for yourself, **with open eyes**, what these models can and can’t do, and you won’t get a good look if you don’t push them hard enough in all directions."
And just for the record, I use agents daily.
That's a strange interpretation. Saying "look!" is not the same as saying "you're blind and don't see anything." There's quite a few people I personally know who have used agents half a year ago only, vastly underestimated the progress they've made and following from that vastly underestimated the progress they will make, and then were very surprised after giving them another spin in the last few weeks. That's why I'm writing this: I don't want people to be surprised.