Where have all the hackers gone?

When computers were small, dumb, and did very little on their own, creative integration of heterogeneous pieces was rewarded and rewarding. Computers are now large, complicated, distributed, and part of a whole ecosystem. The kind of "hacker-shaped" grit that makes someone do something neat gluing a telephone receiver to a TV tuner? Well that's a fun Adafruit demo, but you could have done it easier and faster on a Raspberry Pi, and hell, there's already the scragglypete/tv-phoner repo that you really should have used (if you'd done a bit of research before you whipped out the glue gun, you'd have found the StackOverflow post from three months ago where someone went "Hey, I want to glue my telephone receiver to my TV tuner" and they were directed to that project). And while we're on the subject, that's not nearly as useful as the telephone-receiver-glued-to-a-tv-tuner emulator in this Docker container, because it does exactly what your workbench solution does but you can deploy eighty of them in one minute with a few shell commands and some money thrown at AWS...

Hacking is less cowboy-shaped for the same reason there are fewer cowboys... Eventually, the pioneer towns give way to the skyscrapers and city streets because once the wilderness is tamed, the city is way more powerful, comfortable, and productive than the cowboy life.

But also the dude obviously is a great hacker. There are vanishingly few people who can bootstrap the infrastructure that will let a company grow to Facebook scale -- you need exceptional depth and breadth of computer knowledge, as well as great focus and an ability to prioritize and satisfice.

Anecdata:

I had a job recently that paid competitively, but the boss soon replaced interesting work with boring work (for me: writing boilerplate that's already been solved a thousand times over, or any FE work), so my passion petered out, which was quite obvious to everyone involved, so the relationship ended.

However, I'm with a client now, and we're building something really cool (using LLMs; I get access to H100 soon; etc) and I'm working every spare minute I can get away with, without being paid for overtime.

I feel like a "hacker" again with this role, learning new things, stretching the capabilities of the system given various constraints, and every second link I open in my browser is arxiv.

(That said, I grew up in South Africa, where cyber laws lagged behind the world's for a good few years, so we knew every trick in the book, yet were untouchable, haha!)

The few meetups I go to, I've also noticed that 90% of conversations stay at surface level, the folks who prefer to go deeper were always rare, its likely that in the past, meetups tended to attract that small percentage in the first place, so there's some selection bias.

In terms of language-level specifics, I will say that in 12 years of experience I have rarely ran into a problem where "soil"-level properties of a language actually made a difference. I'm not building systems-level stuff or libraries, I'm building apps and services for businesses, where IMO the surface or atmosphere-level issues are more important. It doesn't matter that Python is 1000x "slower" than some other language, up to a certain scale. It matters more that I have access to Python's ecosystem, or there are specifics in how a python app is structured that I just happen to prefer because I'm comfortable with it. This is just an example.

The other issue is that most applications we write are going to fall into a handful of categories. The needs of a locally-run stateful application are going to be different from the needs of a stateless web application. For web apps, your bottleneck is rarely in the language runtime, and more in your application design, database, infrastructure, etc.

Language-level enthusiasts (which is how I would describe the author) are incredibly valuable members of the community, but its not the only game in town. There are so many other dimensions to measure technology, and so many other fields to be a "hacker" in. So why do "innovation tokens" need to be spent on the language runtime?

b) Your argument that people solely use empathy as a kudgel to get their way is not reflected in reality. What you describe is emotional manipulation, and as far as I can tell, you're essentially arguing that all calls for empathy in conversation are emotional manipulation... which is ridiculous.

c) Simply speaking about people respectfully is actually important and doesn't change the concrete components of a conversation. Discussing whether or not there should be some concrete action in response to a bunch of people losing their livelihood doesn't benefit from "lol get a new career, idiot." If you don't understand that, you've probably got a serious problem understanding the effect your words have on other people.