jere
Joined 14 years ago
Joined 14 years ago
Yes, that’s an incomplete and tenuous belief, but that’s where the irony is. If you’re omniscient you don’t get to experience irony.
Seeing a gwern comment get downvoted is a first for me. An odd sight.
I feel like I might have a unique perspective here. I absolutely hate slop, but I've also experimented a lot this year with AI game dev. I was shocked at the negativity (being told snidely to "learn to code" despite having made games for decades).
This seems like a clash of an extremely AI pilled person coming into contact with the often irrational and overwhelming backlash against AI, which is common even on HN these days.
Slop is only an issue is when it's implied to be handwritten. Because then someone is asking me to waste my time reading something they didn't spend time writing. That's not the case here since you've called out the AI contribution, but I bet a lot of people are skimming over it.
The real problem I see here is that Claude often doesn't have a sense of what's interesting. In the 30,000 word write up, I don't see anything particularly new and interesting about the prototype.
I see: using turns as a score, Hamiltonian paths, and a reference to a "wide rake" and "no u-turns" which either I don't understand or doesn't show up until late into the game. If there is something interesting in there, Claude hasn't convinced me of it!
Where the audience starts getting rightly annoyed is an irreverence to the existing prior art and an overestimation of how unique/interesting this prototype is (presumably fueled by discussions with Claude), particularly these statements:
> I came up with the original idea entirely independently [an idea that is predated by 20 years]
> ZPG winds up not being as interesting, I think
Let me state: you've made something neat! I like the concept, I like the heightmap aesthetic. Fwiw, most of my vibe coding experiments were about digging in sand using voxels.
However, people are hypersensitive to anyone easily impressed by AI output while simultaneously downplaying the value of human creativity.
The scary thing is that most of these Claudisms can be tuned down easily if that's what Anthropic was trying to do (and may in the future). Then it would be nearly impossible to tell.
Sounds legit
You either have an absurd standard for "decent" or this is one of the weirdest comments I've ever read here.
I don't think everyone agree with your notion that the language and tools of mathematics aren't human inventions.