[{"data":1,"prerenderedAt":305},["ShallowReactive",2],{"project-46583581":3},{"id":4,"name":5,"line":6,"line_source":7,"title":8,"url":9,"domain":10,"maker":11,"posted_at":12,"year":13,"points":14,"comments":15,"rank_in_year":16,"shown_in_year":17,"top":18,"tier":19,"category":20,"subcategory":21,"format":22,"keywords":23,"reception":28,"sentiment":29,"stance_mix":30,"scored_comments":34,"aspects":35,"strip":40,"post_text":42,"what":43,"said":54,"maker_in_thread":147,"shape":154,"grid":159,"attempts":266,"same_week":267,"maker_shows":198,"maker_hidden":297,"reader_version":298,"card":299},"46583581","B-IR","An LLM-optimized programming language","title","Show HN: B-IR – An LLM-optimized programming language","https:\u002F\u002Fgithub.com\u002FImJasonH\u002FImJasonH\u002Fblob\u002Fmain\u002Farticles\u002Fllm-programming-language.md","github.com","ImJasonH",1768187509,2026,62,41,1043,4366,24,"notable","languages","programming language","other",[24,25,26,27],"llm programming language","ai code generation","intermediate representation","formal verification","roasted",-0.449,{"positive":31,"neutral":32,"mixed":32,"critical":33},2,1,8,12,{"idea":36,"execution":38},{"positive":31,"critical":37},5,{"positive":39,"critical":31},0,[39,39,39,39,39,39,32,39,32,32,39,37,41,39,39,37,39,39,39,39,39,39,39,39,39,39,39,39,32,39,32,32,32,39,39,39,41,32,39,39],3,null,{"problem":42,"audience":42,"pricing":42,"motivation":44,"time_to_build":42,"stack":51,"openness":52,"stage":53,"team_size":52},{"value":45,"source":46,"evidence":47},"Exploring how LLMs could design a language for their own use","maker",[48],{"id":49,"by":11,"text":50},"46588407","I don't disagree at all. :) This was mainly an exercise in exploration with some LLMs, and I think I achieved my goal of exploring. Like I said, if this topic is interesting to…",[],"unknown","prototype",{"praise":55,"objections":64,"requests":100,"questions":115,"alternatives":123},[56],{"value":57,"source":58,"evidence":59},"Useful exploration of the idea that verifiability helps both LLMs and humans","commenter",[60],{"id":61,"by":62,"text":63},"46664792","Mathnerd314","I tried a thread, I got that both LLMs and humans optimize for the same goal, working programs, and the key is verifiability. So it recommended Rust or Haskell combined with…",[65,79,89],{"value":66,"source":58,"evidence":67},"LLMs are trained on existing languages, so new invented formats lack training data",[68,72,75],{"id":69,"by":70,"text":71},"46585046","kburman","An LLM is optimized for its training data, not for newly built formats or abstractions. I don’t understand why we keep building so-called \"LLM-optimized\" X or Y. It’s the same…",{"id":73,"by":11,"text":74},"46588366","Yeah fwiw I agree. I was impressed at how well the agents were able to understand and write their invented language, but fundamentally they're only able to do that because…",{"id":76,"by":77,"text":78},"46585725","Surac","so where are the millions of line code you need to train the LLM in your new language? Remember AI is just a statistic prediction thing. No input -> No output",{"value":80,"source":58,"evidence":81},"Vibe-coding the language was lazy; the design was not thought through",[82,85],{"id":83,"by":62,"text":84},"46584142","I get that this is essentially vibe coding a language, but it still seems lazy to me. He just asked the language model zero-shot to design a language unprompted. You could at…",{"id":86,"by":87,"text":88},"46584449","Snacklive","I was thinking the same. Maybe if he tried to think instead of just asking the model. The premise is interesting \"We optimize languages for humans, maybe we can do something…",{"value":90,"source":58,"evidence":91},"Token minimization is the wrong goal; verbosity can help LLMs",[92,96],{"id":93,"by":94,"text":95},"46585536","forgotpwd16","There's also the issue, which is also noted by the author, that LLM-optimization quite often becomes, when shouldn't be just that, token-minimization.",{"id":97,"by":98,"text":99},"46588094","nzach","> The real question is what existing language is perfect for LLMs? I think verbosity in the language is even more important for LLMs than it is for humans. Because we can see…",[101,108],{"value":102,"source":58,"evidence":103},"Research into languages that support formal methods and model checking",[104],{"id":105,"by":106,"text":107},"46585061","cpeterso","If a new programming language doesn’t need to be written by humans (though should ideally still be readable for auditing), I hope people research languages that support formal…",{"value":109,"source":58,"evidence":110},"Benchmarks comparing which languages AI models solve tasks in best",[111],{"id":112,"by":113,"text":114},"46591798","scotty79","Are there coding AI benchmarks that show in which language AI models are best at solving tasks? Sort of like Rosetta code but for AI and more complex tasks?",[116],{"value":117,"source":58,"evidence":118},"Which existing language is best suited for LLMs?",[119],{"id":120,"by":121,"text":122},"46587000","giancarlostoro","The real question is what existing language is perfect for LLMs? Is Lisp? ASM? We know some LLMs are better at some languages but what existing language are they best at? Would…",[124,136,143],{"value":125,"source":58,"evidence":126},"Lisp-style languages and S-expressions",[127,128,132],{"id":120,"by":121,"text":122},{"id":129,"by":130,"text":131},"46654648","quatonion","I have been having a crack at it in my spare time. A kind of intentional LISP where functions get turned into WASM in the cloud. The functions are optionally tested using formal…",{"id":133,"by":134,"text":135},"46666402","wfn","I've been thinking about this, take a look at this: > From Tool Calling to Symbolic Thinking: LLMs in a Persistent Lisp Metaprogramming Loop https:\u002F\u002Farxiv.org\u002Fabs\u002F2506.10021 edit…",{"value":137,"source":58,"evidence":138},"APL, J and array programming languages",[139],{"id":140,"by":141,"text":142},"46588863","atlintots","I'm assuming OP is not aware of APL, J, or similar array programming languages.",{"value":144,"source":58,"evidence":145},"Rust or Haskell with formal verification",[146],{"id":61,"by":62,"text":63},{"comments":37,"answered":31,"key":148},[149,150,151],{"id":73,"text":74},{"id":49,"text":50},{"id":152,"text":153},"46588389","Fixed the loom link!",{"commenters":155,"max_depth":156,"maker_replies":37,"first_comment_mins":157,"lifespan_hours":158},31,6,65,151,[160,162,166,169,172,174,177,179,181,183,184,186,188,190,193,196,200,202,205,206,208,210,211,212,216,219,222,224,227,229,231,232,236,240,242,246,249,254,259,262,264],[105,39,39,161,106,39],"other\u002Fneutral\u002Fnone",[163,39,39,164,165,39],"46585999","commenter\u002Fneutral\u002Fnone","jaggederest",[167,39,39,164,168,39],"46588900","WithinReason",[170,39,39,164,171,39],"46654524","measurablefunc",[173,39,39,164,168,39],"46656643",[175,39,39,164,176,39],"46657505","SkiFire13",[69,32,32,178,70,39],"project\u002Fcritical\u002Fidea",[73,32,39,180,11,32],"maker\u002Fneutral\u002Fidea",[83,31,32,182,62,39],"project\u002Fcritical\u002Fexecution",[86,31,32,182,87,39],[49,31,39,185,11,32],"maker\u002Fneutral\u002Fnone",[61,31,37,187,62,39],"project\u002Fpositive\u002Fidea",[93,31,41,189,94,39],"project\u002Fneutral\u002Fidea",[191,41,39,161,192,39],"46585815","sublinear",[194,41,39,161,195,39],"46654437","flir",[197,198,37,187,199,39],"46584075",4,"discrisknbisque",[152,198,39,201,11,32],"maker\u002Fneutral\u002Fdocs",[203,37,39,161,204,39],"46584043","petesergeant",[120,156,39,161,121,39],[97,156,39,207,98,39],"other\u002Fneutral\u002Fidea",[209,156,39,180,11,32],"46588613",[129,156,39,161,130,39],[133,156,39,161,134,39],[213,214,39,161,215,39],"46585406",7,"evacchi",[217,33,39,207,218,39],"46585578","AlexCoventry",[220,33,39,207,221,39],"46585674","barnacs",[223,33,39,207,165,39],"46585951",[225,226,39,161,94,39],"46585482",9,[76,228,32,178,77,39],10,[112,230,39,161,113,39],11,[140,34,32,178,141,39],[233,234,32,178,235,39],"46586384",13,"middayc",[237,238,32,178,239,39],"46585437",14,"internet_points",[241,238,39,207,165,39],"46586012",[243,244,39,161,245,39],"46587213",15,"mooktakim",[247,244,39,161,248,39],"46654489","lifetimerubyist",[250,251,41,252,253,39],"46586324",16,"project\u002Fmixed\u002Fidea","mike_hearn",[255,256,32,257,258,39],"46584242",17,"project\u002Fcritical\u002Fnone","rvz",[260,256,39,207,261,39],"46584410","eadwu",[263,256,39,185,11,32],"46588431",[265,256,39,164,258,39],"46600387",[],[268,273,278,282,287,292],{"id":269,"name":270,"points":271,"category":272},"46556210","Lend Me Your Ears",565,"learning_science",{"id":274,"name":275,"points":276,"category":277},"46567400","trails",524,"ai",{"id":279,"name":280,"points":281,"category":272},"46557029","Quack",330,{"id":283,"name":284,"points":285,"category":286},"46602324","Self-host Reddit",286,"infrastructure",{"id":288,"name":289,"points":290,"category":291},"46571980","Ferrite",241,"dev_tools",{"id":293,"name":294,"points":295,"category":296},"46612494","OpenWork",231,"productivity",false,"v2",{"id":300,"name":5,"line":6,"ai":297,"cat":20,"year":13,"points":14,"comments":15,"rec":28,"s":301,"top":18,"kw":302,"maker":11,"domain":10,"tier":19,"attempts":39,"strip":303,"url":9,"mix":304,"praise":57,"objection":66,"sub":21},46583581,-0.45,[24,25,26,27],"0000001011053005000000000000101110003100",[31,32,32,33],1794036690077]