| ▲ | zarzavat 13 hours ago |
| I thought 'how bad can it be?', yet only three paragraphs in "a specifier now names a repository rather than choosing a transport" If you're going to serve us slop then at least serve us good slop, give it a pass through GPT 5.6 to clean up the English. |
|
| ▲ | serbuvlad 13 hours ago | parent | next [-] |
| I remember the most interesting things to do with 3.5 Turbo was to ask it to rewrite paragraphs in different styles. Why can modern LLMs basically not do that anymore? They each have their own very particular writing style? It it all just beaten out of them in post training? |
| |
| ▲ | zarzavat 11 hours ago | parent [-] | | The objective of RL changed circa 2024. The 3.5-4o era models were trained by RLHF primarily to write in a way that's pleasing to humans. Starting with o1 the focus switched to reasoning, coding and benchmarks. If you remember when GPT-5 came out it introduced a new "colder" writing style (which was partly intentional) which made a certain subsection of Reddit Very Upset that Sam Altman had killed their AI boyfriends and they forced him to reinstate the warmer 4o. The artificial default voice of modern models is optimized for reasoning efficiency rather than understandability. Unfortunately the artificialness has a tendency to leak out into the output tokens because reasoning tokens and output tokens are all still just tokens. You can prompt recent models to write in a more understandable style and they do a fairly good job of it, but it's not the default anymore, which means you will have to keep re-prompting it. I don't think it's an unfixable problem, Anthropic in particular are just not very motivated to fix it because Claude is much less popular for chat than ChatGPT is. | | |
| ▲ | serbuvlad 7 hours ago | parent [-] | | You can certainly nudge it directionally, but it has its own style. And verbal tics etc. So it simply cannot keep documentation or a codebase in the same style, and it cannot write <information> in <style>. With 3.5 Turbo I used to give it bulleted information and paragraphs of what I wanted it to sound like and it'd replicate that style, then I'd edit the factual details it got wrong. Now the situation has completely reversed! It can extract the data itself, but I have to write the stuff by hand. Humorously, it will do bulletpoints for me, and I write the thing. |
|
|
|
| ▲ | IanCal 13 hours ago | parent | prev | next [-] |
| What’s the problem with that? Seems to make sense to me with the example and it’s important to point out. I’d expect an llm to explain in more detail to be honest. |
| |
| ▲ | anon7000 13 hours ago | parent | next [-] | | It doesn’t quite make sense at face value by itself. It’s too condensed and takes some effort to parse out what it’s actually getting at. It’s not communicating why the change was made. It’s just a shiny way of saying the words without actually helping me easily understand what’s happening, what’s changing, why, etc. In plain English, maybe it’d be more like “Package specifiers for git repositories should no longer include the HTTPS/SSH protocol. Instead, reference the repository name, and pnpm will resolve it through HTTPS. If SSH is still needed, add an override in your global git settings” | | |
| ▲ | netniuq 13 hours ago | parent [-] | | yes! and this is sooo weird given that Opus 4.8 or 5 producing this are supposed to be "smart" models – how far into the benchmaxxing do you have to lean to get to this point? |
| |
| ▲ | zarzavat 13 hours ago | parent | prev | next [-] | | If you spend all day talking to Claude then its mannerisms begin to grate on you like fingernails on a chalkboard. There's nothing wrong with the English per se, it's grammatically correct, it just sounds so "Claude". Claude talks using pithy word soups that lack the niceties and structure of human writing. | |
| ▲ | fatata123 9 hours ago | parent | prev [-] | | [dead] |
|
|
| ▲ | deanc 13 hours ago | parent | prev [-] |
| Maybe the author is a non-native English speaker, and this enables him to communicate things he's not able to do himself. Should OSS maintainers start hiring professional translators? Are you going to fund that? We really should stop jumping to conclusions and shaming people for using these tools just because you don't like the language of the LLM. How is a non native speaker meant to determine what is English slop and what isn't? I should also note, I am a native speaker. There's absolutely nothing wrong with what's written in the part you chose to paraphrase, and it makes sense to me. |
| |
| ▲ | flohofwoe 13 hours ago | parent | next [-] | | A non-native speaker most likely writes more 'understandable' English than the weird convoluted Claude style. FWIW though, you can ask an LLM to write in 'Simplified Technical English' (see: https://www.asd-ste100.org/), which at times reads like its written for a child, but it is much better for inclusion of non-native speakers. | |
| ▲ | raincole 13 hours ago | parent | prev | next [-] | | > Should OSS maintainers start hiring professional translators? The world existed before 2022. | | | |
| ▲ | fredrikholm 13 hours ago | parent | prev | next [-] | | This is a recent degradation; they've been writing blog entries for years. As a non native English speaker I can also attest to this trade being extremely demanding of English fluency: > public static void DoTheThing(Configuration config) Isn't something you can not understand and still function as a software engineer. Yes, some schools and universities teach in other languages, but the entire ecosystem from code to documentation to articles is nearly almost always in English. | | |
| ▲ | deanc 13 hours ago | parent [-] | | It doesn't matter. I work with people who aren't native speakers (in a country where English-education is a very high standard) but are mostly fluent in day to day conversation and meetings and they are not comfortable writing long-form texts - and it shows. Your expectations are unrealistic. | | |
| |
| ▲ | veidr 13 hours ago | parent | prev [-] | | Ditto. I've started to wonder if the real problem with many of these complaints about LLM writing is actually reading comprehension. |
|
