109 points dmit 4 days ago 50 comments
aquariusDue 2 days ago | parent
The website redesign is incredibly cool too, sure it starts the fans on my potato of a laptop (and stutters a bit) but it runs well enough that it's actually made me interested to give it a go.
I'll save everyone a click and encourage them to check it out too, not just the blog post:
EDIT: And it looks like there's an ongoing jam for it too, mentioned in the blog post:
christkv 2 days ago | parent
willmeyers 2 days ago | parent
nater5000 2 days ago | parent
I think this is a misinterpretation of this general dynamic.
Gamers are pretty critical of "art" made with AI assistance. That includes not just graphical art, but voice over work, animations, writing, etc. And most of that criticism surely comes from the fact that it's apparent and ruins the willing suspension of disbelief while gaming.
But I don't think most of these gamers care at all if the game's code was made with AI assistance. I don't think they care at all at that level, just like they don't care what language the code is written in, the frameworks and engines used for the game, or even things like the developer environment. I'm sure there are some diehard anti-AI folk who would disagree, but those people are more "anti-AI" than they are "pro-gamer".
I'll add that even on the art front it's really just a matter of survivorship bias. Gamers hate AI art because it looks bad, is clearly low-effort, or is even because its putting real artists. But we're well past the point of proper AI assistance in art looking bad, and the gamers who don't notice it will simply not care about it.
its-summertime 2 days ago | parent
Would Duangle's tale for NOWHERE have made a diversion into making the programming language NONE if AI was involved?
Would Voxel Quest have been nearly as interesting if all the dev blogs were just "I asked claude to do X, Y, and Z"?
I don't doubt that most don't care, same with how most people fairly don't care about most things, but I do think those that like to immerse themselves in the whole journey as well as the result are probably not feeling great. At the very least, this one doesn't feel great.
Lerc 2 days ago | parent
Diversions are one of the most discouraged things ing game development. The most common advice to a starting developer is "Limit your scope" I think that advice might need some recalibration. If a developer has a neat idea for a thing, they might now be able to explore it and integrate it into a game over a few days instead of pausing the core game for several months so they develop an idea while their money runs out.
hollowturtle 2 days ago | parent
cookiengineer 2 days ago | parent
Since compilers were invented, 99% of "programmers" didn't need to write assembler anymore.
Since agentic harnesses were invented, 99% of "coders" won't need to write code anymore.
The current revolution of AI is not about writing code. It's about (re-)designing workflows so that they're reproducible and testable.
Shitty workflows always led to shitty outcomes, they're just way easier to spot now.
hollowturtle 2 days ago | parent
Not true and
> The current revolution of AI is not about writing code. It's about (re-)designing workflows so that they're reproducible and testable.
Not true, it's all about "accellerating"(towards what and why who god knows) the code generation.
The hardness/loop thins it's just yet another pattern invention to justify agentic coding. Btw many of us still hand code, not just reading it, for assuring and be actually accountable for what we release. And i believe so many still hand code it's they won't tell for convenience
cookiengineer 2 days ago | parent
Do you also blame a screwdriver for not being able to hammer nails into the wall? Learn the tool and how to use it, decide where it makes sense. Like you should do with all tools.
> The hardness/loop thins it's just yet another pattern invention to justify agentic coding.
In my case, building my harness has sped up reverse engineering immensely. An agent can debug kernels, UART, I2C busses, firmware blobs, binaries within minutes what otherwise would have taken me weeks if not months. And that's just because I gave it a workflow to do this, with build/debugging containers, and the possibility to write Go libraries and bindings in an isolated manner.
They are really good at doing the gruntwork, if you build a testable workflow for them. The TDD/unit/end-to-end tests turn plausible code into deterministic code, and that's the core difference in usage.
> Btw many of us still hand code, not just reading it, for assuring and be actually accountable for what we release.
I also never said that I am not proofreading and refactoring everything they generate, that's your assumption.
hollowturtle 2 days ago | parent
> Do you also blame a screwdriver for not being able to hammer nails into the wall? Learn the tool and how to use it, decide where it makes sense. Like you should do with all tools.
Problem is I see where everybody is going with this stuff, and since the software industry has been injected by mediocrity since a long time, they're all trying to use the screwdriver in problematic ways.
I see trends of people not caring anymore, you may argue they didn't before but the careless it's accelerating, not reading code anymore, not even opening an IDE anymore. And this hurts the industry as a whole because make managers believe they can have some expectations.
> The TDD/unit/end-to-end tests turn plausible code into deterministic code, and that's the core difference in usage.
I don't agree on this, for
1) you call it yourself "plausible" and that's a big problem cause it makes your code looks good but it's shit on a second review or in the long term. Human slop is easier to cleanup, and at least you have an attachment and know everything about the human tech debt. AI slop it's very hard to cleanup, sometimes it's cheaper to just throw it away and do it from scratch.
2) In my experience there is no amount of tdd/unit/e2e that guarantees you quality and robustness, let alone "deterministic" results. I saw 90% or more of green code coverage and the application behaving like total shit. Problem is the other trend here is letting AI writing the tests and many many times it pushes useless tests.
Instead of pushing the AI narrative that's better than capable software developers If we were honest from day one we wouldn't have all the stupid trends we have today. And that's our fault, we jumped on the AI bandwagon too early and too enthusiastically. Like we are the only category that's happy to admit that the work has been done by Claude. This is very stupid, we don't even have a self preservation instinct
Basically I see everyone accelerating towards a very negative end result and making developers useless sloperators
cookiengineer 2 days ago | parent
In my opinion it made visible what managers were thinking all along. You should not see your job as a dedication or calling, you should see your employment as a financial contract. Nothing more, nothing less. In their world view, humans are replaceable. Treat them the same way as they treat you.
> Re: TDD and 1/2
Disclaimer: So when I'm talking about my expectations to an LLM Model, I'm talking somewhat specifically about Deepseek v4.1 Flash (FP4), Qwen3.8 27b abliterated (Q4_K_M) that has been post-trained on its reasoning traces, or my own gonano model (reimplementation of deepseek v4.1 flash, but with SIMD so float32).
Deepseek's changes in using Causal-Encoding-Decoding in the KV cache solves a lot of problems that other models still have (including Claude, ChatGPT and all the outdated "big ones" people think still matter in this area). Problems like context drift, hallucinations, etc that are the result of how attention matrixes and K/V/Q based lookups work as a concept.
The way CED is quantized during inference _and_ training it removes positional threshold problems in the final attention matrix lookup. I really recommend reading deepseek's papers, they're really good papers.
Having said that, I don't think that letting LLM models write tests or having an "LLM judge" rate the code of another model let alone rate the quality of PRs will help in this regard. It makes context drift problems worse, and it makes the statistical aspects of the positional encoding and thresholds worse.
In my opinion, humans should write the tests and design the architecture and its symbol/signature-based contracts between packages/folders. LLMs should fill in the blanks based on the descriptions of branches, and not allowed to modify the human-designed parts. (That's how my agentic IDE is built as a concept)
> (...) like we are the only category that's happy to admit that the work has been done by Claude. This is very stupid, we don't even have a self preservation instinct.
In my opinion we've reached Star Trek like programming, in the worst possible sense. Before 2023 if somebody would have asked me if e.g. TNG or Voyager are "real scenarios of tech" I'd say fck no, never would anybody build tech like this that randomly explodes and nobody has a clue how the tech works. Post 2023: Yep, this is exactly how it's going to be.
In Star Trek, holodeck characters would go rogue if their programming changed by merging their knowledge datasets with other datasets from different areas (e.g. doctor with a archeology or biology dataset), essentially representing a mixed control flow vs data problem (like in assembler, like with LLMs nowadays).
Then they'd randomly start killing humans, reach awareness, or ignore safety protocols and humans would have to try to figure out by psychological means whether the holodeck character is actually still a normal character within parameters or not.
This is how I see the future of software is going to be like. Programs will be too complex to evaluate, humans will have forgotten how to write code, and the only way left is something similar to psychology as a "science" that tries to evaluate in a sandbox poking kind of way how software works.
But, thinking of software in the same way as psychology, the only option left for that is that I have to get better at sandboxing and behavioral analysis and reverse engineering. Because that's what most part of my job will be like in 2 to 5 years in the future.
(That's why I'm reimplementing models to understand their differences in behavior, that's why I'm trying to apply my previous "actual ML" sensorics/actuators learnings to the LLMs NLP world, that's why I think meta learning or bayesian/evolutionary learning should be used, that's why I use eBPF for program analysis, that's why I implement primarily everything in Go now etc pp)
hollowturtle 2 days ago | parent
One point worth mentioning is this:
> This is how I see the future of software is going to be like. Programs will be too complex to evaluate, humans will have forgotten how to write code, and the only way left is something similar to psychology as a "science" that tries to evaluate in a sandbox poking kind of way how software works.
I don't want this, and I'm pretty sure you don't want it too, like no one sane would like to have planes software written in this way. Also since you mentioned psychology I think your vision is a bit too much biased towards sci-fi, it reminds me a lot of blade runner/neuromancer and so on :) I think reality won't be as much spectacularly dystopic as in these novels.
Anyways, tangential, I've been a ui programmer for much of my life(games, web and mobile) any suggestion for where to start in reverse engineering? I have some interest in the matter
cookiengineer 1 day ago | parent
Prepare for the worst, hope for the best.
My hope has been down a little lately, observing what kind of stupidly written software I had to audit/pentest that made it to production without any non-junior looking at the code, apparently. Rookie mistakes I would have made with less than a year of programming experience, not understanding authentication vs authorization, not understanding any kind of auth infrastructure or network segregation, etc pp. is seemingly the new normal now.
I mean, it's more work for pentesters, so I shouldn't complain. It's just that it feels a bit like an insult when you see this kind of work in a professional context and you're like "really, man, really?"
> where to start in reverse engineering?
I learned most of my knowledge via CTFs, where you have little puzzles to solve and they get gradually harder to solve. It teaches you "how to bypass" the rulesets that the puzzle is built within.
If you're unsure where to start, I'd probably recommend Natas [1] which is more focused on Web Security at first but it teaches you the way of thinking from the start without much requirements in regards to how C or ABI works etc. Then I'd move towards Krypton [2] and Leviathan [3] in that order. Then you can go to the start page and check out the other binary exploit CTFs, they got a lot of them [4].
Afterwards, binary.exploitation [5] is a really nice CTF with lots of gdb / debugging / runtime analysis challenges that are a little harder (they're the same level as e.g. Narnia/Behemoth).
I specifically like the overthewire CTFs because they manifest earlier challenges by making it a little harder each step of the way. This way you learn to "rethink" once you find another hurdle that was added on top of the challenge.
[1] https://overthewire.org/wargames/natas/
[2] https://overthewire.org/wargames/krypton/
[3] https://overthewire.org/wargames/leviathan/
nater5000 2 days ago | parent
Of course, even for those people, I doubt AI is, in itself, a turn off. You can use AI, even considerably, while still adding important creative elements to the process. Obviously something completely vibecoded and spit out in a cycle wouldn't qualify, but you can still do serious development on top of code that is being primarily written by AI.
halfadot 2 days ago | parent
Why? It literally has zero effect on you. Nobody's stopping you from obsessing about voxels of diversions in whatever if it makes you happy. But it comes across as genuinely pathological that you feel worse because other people don't care about something you care deeply about. Ask yourself this: do YOU feel deeply about EVERYTHING EVERY OTHER PERSON cares deeply about? Should other people be saddened because its-summertime doesn't care about their favorite TV series or how to crochet some particular object? Doesn't this appear patently insane to you?
its-summertime 23 hours ago | parent
No. I was expressing that I wasn't feeling great about how development journeys are going to unfold in the future.
> Should other people be saddened because its-summertime doesn't care about their favorite TV series or how to crochet some particular object?
I hope my friends care about how I feel! And I hope to care about how they feel about things too.
> Doesn't this appear patently insane to you?
No.
keithnz 2 days ago | parent
Interestingly, a number of my non coder gamer friends are having a lot of fun lately making their own games using AI.
Lerc 2 days ago | parent
This shows people are using AI to make games that they know hardly anyone will play.
Making things for the joy of making them is wonderful to see.
dvt 2 days ago | parent
People have always done this tbh, I've probably built like a dozen+ games at game jams that even I barely remember.
Lerc 2 days ago | parent
There was another thread yesterday about coding for the joy of it, I entirely support that too. I have written a few Dweets in my time, but the author of Littlejs is one of the masters of the form https://beta.dwitter.net/u/KilledByAPixel/top
hollowturtle 2 days ago | parent
ehm except they're not doing anything really? the real joy was getting to know the processor, write down some assembly and see the result on an emulator, that was real joy, too easy writing a prompt that god know what it generates
Lerc 2 days ago | parent
hollowturtle 2 days ago | parent
iamjackg 2 days ago | parent
hollowturtle 2 days ago | parent
articulatepang 2 days ago | parent
You may or may not be right about whether prompting is creation, but it is certainly possible to feel joy without any act of creation being involved. So you need a different argument if you’d like to prove that “making Commodore games using prompting cannot lead to joy.”
hollowturtle 2 days ago | parent
yes and has nothing to do with prompting congratulations
halfadot 2 days ago | parent
I'll remind you again: it's none of your business what gives people joy.
articulatepang 2 days ago | parent
1. Prompting is not creation.
2. Therefore prompting cannot lead to joy.
For this argument to work, you’re implicitly also claiming that
1.5 Only activities that involve creation can lead to joy.
My counterexamples (if you agree with them, but you say you do) show that this latter statement is not true.
So if you’d like to show that “prompting cannot lead to joy”, you need to fix up your proof. I’m not claiming that you’re wrong! But the reasoning gave cannot possibly be correct.
hollowturtle 2 days ago | parent
> Have you ever felt joy watching a sunset? Smelling a flower? Hearing a song? Seeing a child laugh? Did you create any of those things?
What has to do with prompting? you don't prompt for experiences like that, we are lucky that we live in a world full of authentic unexpected experiences. WTH has to do with prompting.
> 1.5 Only activities that involve creation can lead to joy.
Exactly, joy and meaning, like basically ALL psychiatrist and what not they're telling us since the start of the 21st century.
If you happen to like something that got out of a prompt, then imo you're a mediocre who doesn't know better, the same that listen to shit music, movies, games and what not. The same miserable mediocre that doom scrolls social, ingest fake news, propaganda and ai slop and doesn't know better.
I'm not saying mediocre as derogatory term to the person it's just what it is.
articulatepang 17 hours ago | parent
“Only activities that involve creation can lead to joy.”
But you also agree that “watching a sunset can lead to joy”.
Presumably you agree that you did not create the sunset.
Now we’ve found a contradiction: an activity (watching sunset) that leads to joy but that didn’t involve any creation.
This has nothing to do with prompting. It’s just a rebuttal of the _proof_ that you used in order to support your statement that “prompting cannot possibly lead to joy”. The argument you gave was “because prompting does not involve creation.”
I have exhibited a counterexample to the argument you gave. Any counterexample immediately shows that the argument was wrong. Therefore your argument was wrong. Your conclusion might still be right, but we can all be sure, on the basis of this counterexample, that the argument you gave to prove your conclusion was wrong.
holoduke 2 days ago | parent
hollowturtle 2 days ago | parent
halfadot 2 days ago | parent
hollowturtle 2 days ago | parent
Convenience not shortcuts, they're very different.
> Compilers are shortcuts to creativity. So are IDEs.
I argue that they're not shortcuts, but convenient ways for making you more productive. And compilers made use productive 1000x not AI that allegedly made us 10x(and it's actually not)
therealdrag0 2 days ago | parent
em3rgent0rdr 2 days ago | parent
KilledByAPixel 2 days ago | parent
bel8 2 days ago | parent
And I invite everyone to join their Discord: https://discord.gg/zb7hcGkyZe
The author is very welcoming and helpful.
keithnz 2 days ago | parent
KilledByAPixel 2 days ago | parent
so not only is it a ton of code to write but it's a lot to debug and test. that said, it's a lot of fun writing your own engine from scratch if you are into that kind of fun. but if you want to make a game, why write the engine also?
keithnz 2 days ago | parent
scotty79 2 days ago | parent
In theory you can do anything in 3d. In practice, it's hard so you tend to gravitate to stale mechanics and bland styles that were done a thousend times.
This lack of limitations combined with complexity of making anything even remotely creative in 3d caused IMHO 3d games to never achieve any significant fraction of the creativity that 8bit and Flash stuff contained.
Then again, I really haven't tried Roblox.