I just finished reading this article and I'm really impressed. The author has learned a lot in a very short while. And is able to explain parts of it quite well.
This is the kind of stuff I always feel I should be able to do, yet it never happened. Seeing others just do it and share their learnings is such a joy.
As a graphics programmer this doesn't ring true to me.
Using more readable names definitely helps during development. I think the cause of this is twofold.
First, there's a lot of equations used in graphics programming where the canonical names of variables are single letters. If you know the formula a single letter is a good name and it is expected that others reading it also understand it - if you didn't you'd have to read up on the formula anyway.
But beyond that I also think it's a bit of misguided pride. Thinking it's cool to have as minimal inscrutable shader code as possible because that's trendy. It's very common for shaders to be developed with reasonable names and good layout then rewritten before publishing like it was an IOCCC entry.
> First, there's a lot of equations used in graphics programming where the canonical names of variables are single letters. If you know the formula a single letter is a good name and it is expected that others reading it also understand it - if you didn't you'd have to read up on the formula anyway.
That's what I was getting at in my comment. A lot of the shader code I encounter is pretty math-y.
The rest I'm not too sure about .. I don't come across a lot of shaders that are code-golf-y trying to optimize for least number of bytes, but, then again, the article did just that .. so .. :shrug:
It’s something I keep meaning to play. My wife got a decent headset for free through work about three or four years ago, and after we both played a collective 15 minutes of Beat Saber it’s been gathering dust on the top shelf of a closet ever since. We used to play tons of DDR and Wii Sports so we like moving-around games but the thing just didn’t interest either of us, at all.
Alyx just hasn’t been quite enough motivation for me to take on the project of figuring out how to get it hooked up and working, though I did at one point do the few minutes of googling to confirm it might work (I think it’s one of those Facebook ones, and as I recall it is supposed to work for Alyx but I’ll need some cable or other). Especially since it’s a fairly short side-story, it’s just not enough juice to be worth the squeeze.
The conclusion of the series (until it’s not the conclusion any more—yeah, I know how franchises work, lol) though? That’d do it.
If I already had it out and used it regularly I’m sure I’d have played Alyx by now, but with that being the only thing I have any interest in doing with it, just not enough to get me to set the thing up for that alone.
The more I hear about other developers' work, the more varied it seems. I've had a few different roles, from one programmer in a huge org to lead programmer in a small team, with a few stints of technical expert in-between. For each the kind of work I do most has varied a lot, but it's never been mostly about "clarifying requirements". As a grunt worker I mostly just wrote and tested code. As a lead I spent most time mentoring, reviewing code, or in meetings. These days I spend most of my time debugging issues and staring at graphics debugger captures.
Clarifying either business or technical requirements for newer or junior hires.
> reviewing code
See mentoring.
> or in meetings
So clarifying requirements from/for other teams, including scope, purely financial or technical concerns, etc.
Rephrase "clarifying requirements" to "human oriented aspects of software engineering".
Plus, based on the graphics debugger part of your comment, you're a game developer (or at least adjacent). That's a different world. Most software developers are line of business developers (pharmaceutical, healthcare, automotive, etc) or generalists in big tech companies that have to navigate very complex social environments. In both places, developers that are just heads down in code tend not to do well long term.
The irony is of course that humans in general and software professionals in particular (myself definitely included) notoriously struggle with communication, whereas RLHF is literally optimizing LLMs for clear communication. Why wouldn't you expect an AI that's both a superhuman coder and a superhuman communicator to be decent at translating between human requirements and code?
> Why wouldn't you expect an AI that's both a superhuman coder and a superhuman communicator to be decent at translating between human requirements and code?
At this point LLMs are a superhuman nothing, except in terms of volume, which is a standard computer thing ("To err is human, but to really foul things up you need a computer" - a quote from 60 years ago).
LLMs are fast, reasonably flexible, but at the moment they don't really raise the ceiling in terms of quality, which is what I would define as "superhuman".
They are comparatively cheaper than humans and volume matters ("quantity has a quality all its own" - speaking of quotes). But I'm fairly sure that superhuman to most people means "Superman", not 1 trillion ants :-)
I wrote that based on my experience comparing my prose writing and code to what I can get from ChatGPT or Claude Code, which I feel are on average significantly higher quality than what I can do on a single pass. The quality still improves when I critique its output and iterate with it, but from what I tried, the quality of the result of it doing the work and me critiquing it is better (and definitely faster) than what I get when I try to do it myself and have it critique my approach.
But maybe it's just because I personally am not as good as others, so let me try to offer some examples of tasks where the quality of AI output is empirically better than the human baseline:
1. Chess (and other games) - Stockfish has an ELO of 3644[0], compared to Magnus Carlsen at 2882
2. Natural Language understanding - AIs surpassed the human expert baseline on SuperGlue a while ago [1]
3. General image classification - On Imagenet top-5, facebook's convnext is at 98.55 [2], while humans are at about 94.9% [3]. Humans are still better at poor lighting conditions, but with additional training data, AIs are catching up quickly.
4. Cancer diagnosis - on lymph-node whole slide images, the best human pathologist in the study got an AUC of 0.884, while the best AI classifier was at 0.994 [4]
5. Competition math - AI is at the level of the best competitors, achieving gold level at the IMO this year [5]. It's not clearly superhuman yet, but I expect it will be very soon.
6. Competition coding - Here too AI is head to head with the best competitors, successfully solving all problems at this year's ICPC [6]. Similarly, at the AtCoder World Tour Finals 2025 Heuristic contest, only one human managed to beat the OpenAI submission [7].
So summing this up, I'll say that even if AI isn't better at all of these tasks than the best prepared humans, it's extremely unlikely that I'll get one of those humans to do tasks for me. So while AI is still very flawed, I already quite often prefer to rely on it rather to delegate to another human, and this is as bad as it ever will be.
P.S.
While not a benchmark, there's a small study from last year that looked at the quality of AI-generated code documentation in comparison to the actual human-written documentation in a variety of code bases and found "results indicate that all LLMs (except StarChat) consistently outperform the original documentation generated by humans." [8]
Brother, you are not going to convince people who dedicated their lives to learning a language, knowledge that bankrolls a pretty cushy life, that that language is likely to soon be readily accessible to everyone with access to a machine translator.
At least to start. Microsoft strongly encourages all Xbox games to also come out on PC, though they sometimes release later. I cannot find any game developed originally for Xbox Series X|S where this hasn't happened eventually (and the developers definitively aren't still working on the PC version).
There's been quite a few games in recent years where I notice some system and think "ugh, do I really need to bother with this, too?". Especially crafting or skill point systems which feel slapped on. Some games make them a fun and integral part of the gameplay, some seem to include them because it's trendy and it just adds friction and mental load with little payoff.
I don't mind complexity, some of my favorite games are ridiculously complex (Dwarf Fortress), but the complexity needs to pay for itself.
I’ve had similar thoughts too: the older I get, the less “extra features” translate to value if I’m expected to stretch my concentration across all of them to have fun.
I’m not as sophisticated as the average Dwarf Fortress player, but an emergent quality of that game that I’ve admired from afar has been how you can ignore various mechanics and you’re rewarded with an interesting ride.
It’s dynamic enough that by pulling various gameplay “levers” you can get wildly different outcomes (and thus value through replayability), but things will sort of run themselves (for better or worse) if you forget about them. So you’re half writing your own story, half discovering it as it writes itself.
My cynical take is that crafting systems are probably the most attractive on the ratio of "amount of dev effort required to implement" relative to "amount of play time added." They're also trivially tunable. You can add (or subtract) hours of play time just by changing the numbers required to craft things.
Unless they're an integral feature of the game (like in Minecraft), they always feel slapped on to me.
I really hope they release something as well, because I loved their research papers on analyzing how Claude thinks[0] and how they analyzed it[1] and I'm eager for more.
I always get confused around this and it seems to depend on specific keyboard and software. It seems to mostly be Alt, sometimes Alt Gr, and sometimes the Windows key. But I do remember using Meta to mean Alt when I was setting things up in Ubuntu maybe ten years ago.
I think Super and Meta are the vendor independent terms.
Ctrl - Ctrl - Ctrl
Super - Windows - Command
Left/Right Super - ?? - Open/Close Apple
Left Meta - Alt - Option
Right Meta - Alt Gr - Right Option?
Anybody able to fill the gaps?
Approaching an Apple keyboard for the first time, I naively thought Command would be Ctrl and was quite confused, since there also is Ctrl. But once you start a terminal it became quite clear that this was not the case. This is quite neat, it directly solves the SIGINT,SIGSTOP / Copy, Ctrl-D? confusion. Also having these operation as OS command means that all programs support them.
This is the kind of stuff I always feel I should be able to do, yet it never happened. Seeing others just do it and share their learnings is such a joy.
reply