If coding is your favorite part of software engineering, keep coding. AI doesn't stop you. But if coding was the only part you were good at, that's a different conversation. The job was always bigger than the code. #bransoncognac blog.bryanl.dev/posts/ai-sen...
It's like a war movie in that sense.
I was 16 when I came out and I was easily fooled by its rebellious anti-consumerism into viewing project mayhem as a valiant struggle.
Fully understanding it in my 20s and reckoning with my failure informed how carefully I navigate political spaces since.
I didn't expect "agentic AI" to usher in a resurgence of the Unix command line, but it makes sense in hindsight, as composition has more leverage now.
It's also resulted in a rush to poke holes in the walled gardens that have been closed since the death of web2 APIs.
github.com/googleworksp...
"but it might work for us" tobias funke meme where the first panel says "a beheading strike on the current leadership will result in a favorable regime change"
Caught this PBS segment on an organic Asian vegetable farm in upstate NY, and I need you to know that the farm is called Choy Division.
"Growing up, food felt like the one thing I really had in common with my family" is such a universal child-of-immigrants thing.
www.youtube.com/watch?v=mpOp...
The current terminology isn't great.
"Engineer" is too general and ambiguous and in some places reserved for a particular level of education.
"Hobbyist programmer" is even worse; it sounds pejorative or patronizing.
Maybe GenZ or GenA will recognize this dichotomy and help us out eventually.
The opposing force also causes problems.
A trend of "industrial programming" getting adopted by small shops and individuals for which it was unsuitable. "This is how we did it at Google."
In an earlier era, this was best described in the "Hammer FactoryFactory" story by Benji Smith:
This has all led to two unending issues.
The first is "code as craft" people and functional programming zealots constantly tutting (irrelevantly) that the industry is doing it wrong.
Those people are great programmers, but they're making a category error.
In my professional lifetime, we've tried to distinguish between "programming" and "engineering", but this is more about of all of the extra-programming activities and processes you need to build software at scale.
I'm talking about different types of programming.
Programming is a new and relatively undeveloped discipline. I've long felt that we're missing language to distinguish between different activities that we just call "programming."
For example, we have "carpentry" and "woodworking", but nothing clear like this for programming.
I don't begrudge anyone on the Go team adding this feature. They are incredibly exacting and conservative in what they add, more than most.
But languages always decide to become more complex, and it was nice for a time to have one that didn't. I'll miss that.
Can we get some goddamn enums now?
I've only written generic Go code once or twice since then. It's is not particularly useful most of the time, but on occasion it's vital.
There are a lot of these kinds of things in programming. A language can either decide to become complex, or to remain less/unsuitable to some domains.
The first draft for Go Generics was about 8 years ago. At the time I wrote a post that was linked in the usual places and pretty well received:
jmoiron.net/blog/notes-o...
This proposal was ultimately rejected and a different, better one was adopted.
Go generic methods have been accepted. Don't find this particularly readable:
func (*G[P]) m[Q any](x Q) { … }
Go's advantage has always been how transparent the code is. Maybe you hate for loops, but you knew what they were doing!
Even that's not so true anymore.
I keep seeing people repeat "AI is destroying the jobs market" uncritically, without any hard evidence, and it's crazy that this is accepted on its face despite the confounding factor of "a year of instability under an ever changing regime of spite-based tariffs."
Maybe the internet was compromised by a magic Nokia, but this train got me from Tokyo to Fukuoka (1090km) in 4h45m.
Some friends are taking Amtrack home to NYC from Toronto today because the airports here are still impacted by the storm, 750km.. 14 hours.
Yeah, I really hate this take that anything short of welcoming antisocial behavior in public spaces is like ordering a pogrom on the poor. It reinforces a conservative narrative that the only effective recourse is state violence, and since that's bad, you have to deny the problem.
To be honest I don't use JR much within Tokyo when I can avoid it, as Tokyo metro is almost always more convenient and easier to navigate.
Despite that, the Yamanote is legendary and has a special place in my heart and I don't mind taking it once in a while.
The same loop in NYC doesn't really fit but thanks to the G train, approximating it is at least possible.
Hong Kong's lack of a east/west New Territories connection makes such a loop impossible even though I'd say the MTR is generally much better than the NYC Subway.
Yamanote line overlayed on roughly the 63rd st tunnel from queens to manhattan, something like the 1/2/3 in manhattan, and then becoming an alternative to the G train that runs from Kensington through prospect park, conveniently avoiding major population areas.
This website displays a map with an overlay of the Yamanote line. You can move it to other cities and see how it fits. In Tokyo, this train takes ~60 minutes to do a loop and runs with a 90s headway.
www.dokodemo-yamanote.com?lat=40.71034...
I said this more or less but I'm not a liberal and also like, who am I anyway? But I don't think that private gun ownership is a useful way to resist state violence.
jmoiron.net/blog/abolish...
Bert Hubert wrote a great piece about how if you outsource enough of the production of something, innovation becomes impossible:
berthub.eu/articles/pos...
I can see this happening with software. With no humans writing software, who will have the knowledge to innovate that process?
As Cindy Sridharan has noted on the other site, despite the supposed productivity gains, there doesn't appear to be a bonanza of envelope pushing software in the post-AI era.
If anything, the state of the art for systems software was changing much faster in the 2010s.
Of course this is silly to do in a professional context, but up until now software engineers have often worked under a process of continuous skill development over their entire careers.
When the job market was good, people would often switch jobs when they felt they had stopped growing.
Back in the day, you could get coding lessons from magazines. Part of the lesson was typing the code in the magazine into an editor and saving it.
Many of these lessons stressed that this rote copying still had value. It familiarized you with the structure of your programs.
This concern isn't limited to AI. There are studies that show cognitive gaps between writing with a pen/pencil and typing: www.frontiersin.org/journals/psy...
But it's a safe bet that the cognitive gap between handwriting and typing is smaller than between writing and vibe-writing.
I believed this because of the observation in my quoted post above; generating code with AI doesn't feel cognitively similar to writing code.
As Lamport said, if you think without writing, you only think you're thinking. In my experience, this mirage is very convincing.
In my article on LLM environmental impact, I mentioned "erosion of human skill" as a potential downside to AI.
I've always assumed this was real based on personal experience, but there was a paper released recently from Anthropic backing this up, so I don't have to rely on vibes anymore.
Flying is something that has not really tracked inflation. The result is the experience of flying is garbage now unless you pay "extra" to get what used to be the standard experience.
Games are going the same way and it's probably bad.
A bunch of stuff has happened to offset costs since the 90s.. the market is much bigger, you don't need to produce costly media. But you also need a much larger team to produce an AAA title.
There's no reason to assume this balance will last in perpetuity.
