claude-code - 💡(How to fix) Fix Serious accessibility issue with "clever" synonyms for "Working..."

Preflight Checklist

• I have searched existing issues and this hasn't been reported yet
• This is a single bug report (please file separate reports for different bugs)
• I am using the latest version of Claude Code

What's Wrong?

Support people,

Thank you for creating Claude Code. I hope you're getting good feedback for the fine work you do, though support people need to have tough skins; I empathize. I write this to help--not vent!.. ; )

I just subscribed to it at the Pro level, but will probably have to cancel and caution anyone who asks about a serious problem for one kind of neurally atypical but super-high-functioning adults.

Claude Code has a behavior that spuriously activates many concepts and injects them into the human user's working memory. Leaving this "clever" trick in place without the ability for people to opt out is equivalent to creating an interface colorway of magenta text on cyan because some coder personally liked the combination. I'd hope you can make it optional. I refer to the seemingly-harmless (I assume many think) evocative alternatives to "Executing..." and "Done" flanking Claude's "thinking" periods.

One of the most severe (and in the long run sometimes valuable) characteristics of one kind of neural atypicality is that some minds follow every lead--any stimulus, even linguistic signs--for at least a little bit: instant and ineluctable concept activation. It's akin to hyperphantasia, but for concepts instead of images; I don't believe it's been independently studied/named yet. I believe "normies" find they can accommodate annoyances that the sensitive can be driven to distraction for; I would not be surprised if, like tinnitus, it can lead to depression. For these people (I am one of them) it is impossible to gate activation of the concepts associated with the signs (without a lot of alcohol). Add that to a well-developed vocabulary and these distractions can become a nightmare. The "clever" synonym words are almost always grotesquely incorrect--which distracts more--and activate several newly-useless concepts every time I press Enter.

Why is this "harmless" trick a problem fro some? Working memory has severe and tiny limits in humans--approximately four psycholinguistic "chunks" (not seven plus-or-minus two; tiny in itself). In fact, it is a serious accessibility issue not to allow it to be turned off, and I'm trying to report it as such. I believe though many can get past it, it still impairs them: the irrelevant concepts still may have some effect that jostles the usefully-activated ones out of working memory. It would be useful to do true cognitive-science tests on this effect to understand the extent of the damage, but I can verify right now that it has enough of an effect that an ethical interface design group should at least make it optional. (This has the useful effect of providing a post-hoc real-world test: if a lot of people turn it off, you can assume it was doing a lot of damage--an underestimate because many might not feel the damage or recognize what's causing it, or be able to find the off switch.)

Related, but at a perceptual rather than cognitive level of human mental processes, is the growing/shrinking asterisk-trick. Technically: it completely captures attention--for some--an impossible-to-control level, entraining and almost blanking their minds while they try to acclimate to tear their attention away. A heuristic I've developed and my team & hundreds of students find useful: any animation/transition shorter than about seven seconds halts people's minds until it finishes. If anyone on your team was alive to see the throbbing N similarly "clever" geeks put upper-left in the first Netscape? (Fortunately enough complained it got scuttled quickly.)

I am such a geek and understand! It's great fun to play with visual tricks and get people's attention. But these things never leave my lab out of respect for people's time. If deployed to millions, fractionally freezing or overflowing their mental processes without consent, it likely adds up to years of human effort lost. (A Steve Jobs argument about 1984 Mac boot time repurposed--not just me... ; )

For what it's worth I've taught interaction design at Columbia for decades, applying perceptual and cognitive understanding of how people engage with computers. Claude Code seriously breaks engagement and impairs thinking for a completely unnecessary quirk that seems impossibly to turn off. You may look at this picture-book of interface examples & the science behind them if you prefer real-world proof to the pseudo-credential of being an adjunct professor...

https://drive.google.com/file/d/1sZpkTxthiaFP6l4djns4Uh7bYEpjEtB5/view

Please contact me if you like--this is important and I'd like to help.

Kind but verrry frustrated regards, Brad

What Should Happen?

Claude should allow users to opt out of the clever synonyms. This should be a trivially-easy fix. Please do it?

Error Messages/Logs

Steps to Reproduce

Do anything at all in Claude Code. Press Enter.

Claude Model

Sonnet (default)

Is this a regression?

I don't know

Last Working Version

No response

Claude Code Version

2.1.150 (Claude Code)

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

Terminal.app (macOS)

Additional Information

Thank you for your attention, empathy, and care.