If you want to do less with less it's also pretty easy to set up an environment to develop for the Game Boy, NES, or DOS, or ??? In addition to the simplicity, there's also the fact that the architecture is set in stone. If your application runs on the original hardware, you did it right.

This is like listening to Dave Grohl teaching programming

Patches on patches ... Old code + new code + libraries. Nobody writes from scratch.

On the seatbelt stuff: I highly recommend the book "killed by a traffic engineer" It talks about how road designs that "make roads safer" like wider lanes and such actually cause people to drive faster and crash more often.

Doesn't COMPLEXITY become manageable, when each of the INTERACTING COMPONENTS, are SIMPLE & ELEGANT?

No where near as powerful as the Sharp X68000 which also looked far sexier too.

I like to think of software complexity as entropy. Entropy is always increasing but we can control the rate of growth by being deliberate.

This video has far too many ads and I'm not sure the content really is that great, in not really learning anything surprising and honestly some of the "revelations" are quite useless and dumb. Imo.

I've had enough of authoritative software opinions/commentary and minimal falsifiable statements and evidence. But at least we can agree on a vision of simpler software

You can't make simple software because modern programmers are afraid of code that is allowed to break.

"Vigilance is not a strategy" Oh, it very much is. Ask anyone who works with anything physically dangerous for a living (heavy machinery, firearms, etc.) and there is one thing you will hear repeated throughout all of these industries: You don't blindly trust safety. Even when you are 100% sure the gun doesn't have any bullets in it, you still don't point it at anyone until that barrel is physically removed from the gun. Likewise, you don't stick your hand in a piece of heavy machinery until the power has been physically disconnected and the appropriate lockout/tagout procedure has been carried out. When it comes to coding, never just trust that the library will be perfectly secure or even intuitively coded. You really should audit the source code (or at least the documentation) of the library to know exactly how it handles edge cases and code accordingly. Just the other day I found out that I couldn't even assume that std::stoi() in a try/catch would ensure that the input string is a valid integer because std::stoi("123 hello") will return 123.

Andrew Kelley rules.

dude, if you are gonna talk about anything, using terms that are broad and can be ambiguous, it is always better to define them before hand.

This talk answered many existential questions I had for decades... Especially the part "better tools won't solve the problem". Thanks 🙏!!

Absolute nonsense. An insult to the field.

40 minutes to say kind of nothing

33:28 This is what Terry A. Davis di with Temple OS. He reduced possibilities and set hard constraints.

The dude made his entire point by how he introduced himself.

5:30 refers to "Shotgun parsers in the cross-hairs" from BruCON Security Conference

As information processing system grows, it tends to a state where it uses infinite resources to do absolutely nothing.

Most people are already either suffering or muffed to silence on the current status quo of IT janitoring. The problem with inventing new languages is ITU (Inventing The Unnecessary). It is akin how Dr Nobel Price invent the talky stick on Sesame Street.

quote goes after the period in the U.S. !!!

embrace complexity

30:00 wow they actually wrote an emulator to overcome raw testing on hardware without OS. ❤❤❤❤❤

Great talk, I specifically enjoyed the idea of reducing scope to just try building with what you know. I struggle with that a lot, and think AI will help with that... if we can figure out how that would work in scope 😂

this is gold. love ink and switch

You can beat the complexity by applying "reboot" by AI code transformations

Stupid YT blocks final slide with related videos, maybe Handmade Cities in the future can make sure to put some filler seconds at the end with their logo so presentation is not ruined.

What an outstanding presentation! Your eloquence, sense of humor, captivating theme, seamless coherence between the slides and your narration, and, above all, your incredible charisma made it truly remarkable. Thank you!

Because there are LOTS (the majority of) developers who are amateurs. I have spent a month immersed in the heinous world of poorly documented poorly written "opensource" software that the hobbiest developers change and tweak at whatever liesurely whim strikes them.

He's got the exact same voice as Dave Grohl. Yes, that's my contribution to this comment section.

skill issue

"many articulation points in a distributed system with more free variables , more places where things can go fast or slow, or arrive out of order". I love the term "articulation points". If you think of a system as a structure with components under compression and tension, more points of articulation makes you see why complexity is an issue.

We need to talk more about complexity in the software industry :) Nice talk, I really enjoyed it. There's a minor "mistake", though. Accidental and Essential complexity concepts were introduced by Frederick Brooks in his brilliant paper: No Silver Bullet. Moseley and Marks did an outstanding job by elaborating further these concepts, though. Great talk Mr Van Hardenberg.

Until...

Smart money is on Rust.

22:15 The truth hurts

Because you are bad at programming.

Yo I loved teenage zombies as a kid, def gunna watch the rest of the talk now lol

this was quite enlightening about why and where zig’s endless shortcomings come from

Er… about the seat belts: I'm sorry, but no. If you're going to do some unconventional, supposedly though provoking take about a _widely studied_ topic, at least check the Wikipedia article about it, or provide some actual data.

C++ - I looked through the book ~ 1989, and tried to visualize how to write a parser for it. I have avoided this hairball ever since.

Regarding the seatbelt thing, is there data to suggest that people drive riskier now than they did prior to the mandate?

This is the whole fun part of Factorio, constantly growing the complexity as you grow the factory and your goals.

Strong talk

Nothing beats pure C. I can understand people don't like C++. When I use C++ I only use a subset, classes, and very little more.

I think one of the important things that doesn't really get much time in the talk is that moving complexity off to libraries and services means we can write software that does many more things with smaller budgets. Webapps are popular not only because they're the only truly cross-platform environment, but because they ship with a very flexible rendering engine and huge ecosystem support for many frameworks. I think the real complexity is hidden in state management. External (outside your computer) dependencies add more state to track. User interaction adds more state. File access adds state. Multi-step UI flows add state. Updating third-party dependencies (often for potential security issues) is state (on the dev's side) that changes and can mess up your code. All of these state-heavy systems can fail at various points, requiring other parts of your applications' state to update in response. It's covered somewhat in the NxM complexity section but I think it could use more focus.

this talk reminds me of most seinfeld episodes

I think I'll be rewatching this a few times and nodding in agreement each time.

kinda kinda kinda kinda