4 ms·

A while back, I noticed that using Claude all day left me feeling mentally tired. After thinking it over, I realised the problem was Claude’s complex language.

A while back, I noticed that using Claude all day left me feeling mentally tired. After thinking it over, I realised the problem was Claude’s complex language. I’ve used English daily for years, so that shouldn’t be an issue. To compare, I switched Claude to my native language and had the same problem the author described. The text still needed to be literally deciphered before I could understand it. After talking with Claude, I updated the settings to ask for simpler language and less dense information. Natural language works better with some entropy in the text. This change has made things easier.

jmartrican 2mo ago
I recently updated Claude to respond with shorter text and in bullet points, and other similar changes. It is an area of active improvements on my end. But early results are in. Now I feel like my mental capicity is not crowded out by the LLM's wall of text. There is now mental capicity to think about the problem at hand and come up with ideas.
- mstaoru 2mo ago
  [flagged]
- grimcompanion 2mo ago
  IMO this definitely helps, but only to a degree, because the more quickly I can read/digest output, the more quickly I've got more output to read/digest. The jargon/quirks aren't as dense per message but they still accumulate eventually, and so either way I hit a wall where I need to do something, anything, besides read another line from CC.
defgeneric 2mo ago
I've seen this sentiment so much in the past month that it's starting to make me worried that the labs will silently make this the default output to satisfy users. I do recognize the outputs can be so information-dense that it strains reading for many but they absolutely aren't nonsense or mere affectation. What people are asking for now amounts to asking for less intelligence per token only a few months after everyone was demanding more intelligence per token.
- remich 2mo ago
  Yeah I worry about this too. I have a remote agent that writes PRs for my job, and at the request of other devs I implemented a simplified writing skill that does the technical english thing mentioned elsewhere in these comments. It's definitely easier to read and the other devs seem to like it, so I'm okay with keeping it, but I haven't changed my personal claude settings because I never had a problem with the density of the output myself. Probably a side effect from having a multi-disciplinary background.

jmartrican 2mo ago

I recently updated Claude to respond with shorter text and in bullet points, and other similar changes. It is an area of active improvements on my end. But early results are in. Now I feel like my mental capicity is not crowded out by the LLM's wall of text. There is now mental capicity to think about the problem at hand and come up with ideas.

mstaoru 2mo ago
[flagged]
grimcompanion 2mo ago
IMO this definitely helps, but only to a degree, because the more quickly I can read/digest output, the more quickly I've got more output to read/digest. The jargon/quirks aren't as dense per message but they still accumulate eventually, and so either way I hit a wall where I need to do something, anything, besides read another line from CC.

defgeneric 2mo ago

I've seen this sentiment so much in the past month that it's starting to make me worried that the labs will silently make this the default output to satisfy users. I do recognize the outputs can be so information-dense that it strains reading for many but they absolutely aren't nonsense or mere affectation. What people are asking for now amounts to asking for less intelligence per token only a few months after everyone was demanding more intelligence per token.

remich 2mo ago
Yeah I worry about this too. I have a remote agent that writes PRs for my job, and at the request of other devs I implemented a simplified writing skill that does the technical english thing mentioned elsewhere in these comments. It's definitely easier to read and the other devs seem to like it, so I'm okay with keeping it, but I haven't changed my personal claude settings because I never had a problem with the density of the output myself. Probably a side effect from having a multi-disciplinary background.