Please stop asking me to provide feedback #8036
github.com
github.com
It's kind of shocking how some people just don't get how insanely insulting it is for an application to constantly ask for feedback.
If it was just one app every now and then. But instead it's (nearly) everything you buy, every restaurant you go to, every app you use, every doctor you see, every hotel you stay at, etc.
It's like when a guy on the street asks you for money. Like you haven't already been asked by everyone else on the block, including the guy standing right next to him.
When I get done with a Teams call, I'm often asked to rate the call. If I pick 2 stars, I'm asked for written feedback on why it's low. If I pick 4 stars, I'm not.
Given how often they ask this, many will select 4-5 stars to dismiss the window.
But I do submit 2 star with the feedback window being the reason because it comes up after every call.
Well that and the chat is terrible, and the calls aren't better than anyone else, and if you do the small group calls it does a chat call rather than a meeting so it is silent to the recipient.
I mean, who the fuck thought *that* was a good idea? You normally have to call and put "call here" in the chat to make people aware.
"will you quit asking that?"
"but many of my dates like being asked!"
Vscodium, and Claude replacement. And its abliterated to boot, so no censorship and garbage.
We actually don't train on this survey data. It's just for vibes so we can make sure people are having a good experience.
See https://code.claude.com/docs/en/data-usage#session-quality-s...
We asked for Claude coz we really wanted it (as devs). Our security and legal guys evaluated it and the data usage agreements and such, which took ages and they stated that we can use it as long as we never never ever give any kind of bug report or feedback because that might retain data. This was all prior to this new feedback prompt being added to Claude Code.
We were happy. We could use Claude Code, finally!
Then suddenly, about once per day we are getting these prompts and there is NO option to disable them. The prompt just randomly appears. I might be typing part of my next prompt, which can very definitely include numbers from 1 upwards, e.g. because I'm making a todo list and boom I've given feedback. I've inadvertently done the thing we were never never ever supposed to do.
Do you really think our first thought is to double check if maybe the data usage agreement has been changed / hopefully says that this particular random new survey thing is somehow not included in "feedback"?
No, we panic and are mad at Claude Code/Anthropic. Heck if you noticed that you inadvertently gave feedback you might make a disclosure to the security team that will now go and evaluate how bad this was. Hopefully they'd then find your updated data usage agreement.
It would have been so easy to just include an option to opt out, which just set the equivalent of the env var you now provide into the settings itself. In fact, I see the first comment a few days after the bug was filed already asked for that and was thumbs'd up by many people. And you ignored it and all we got was a manual env var or settings entry that we also need to first find out even exists.
So you are happy to use Claude.
> Then suddenly, about once per day we are getting these prompts and there is NO option to disable them.
And now you are not happy. Why are you not happy ?
You're holding it wrong. /s
Are you sure that people are having a good experience ?
(after some random time) Are you sure that people are having a good experience ?
(after some random time) Are you sure that people are having a good experience ?
(after some random time) Are you sure that people are having a good experience ?
(after some random time) Are you sure that people are having a good experience ?
It should at least wait until I've finished my journey and parked up. Not that I'm going to bother giving ad hoc feedback then either.
1. We use session quality feedback as a signal to make sure Claude Code users are having a good time. It's helpful data for us to more quickly spot & prevent incidents like https://www.anthropic.com/engineering/a-postmortem-of-three-.... There was a bug where we were showing the survey too often, which is now fixed (it was annoying and a misconfiguration on our part).
2. Giving session quality feedback is totally optional. When you provide feedback, we just collect your numerical rating and some metadata (like OS, terminal, etc.). Giving feedback doesn't cause us to log your conversation, code, or anything like that. (docs: https://code.claude.com/docs/en/data-usage#session-quality-s...)
3. We don't train on quality feedback data. This is documented in the link above.
4. If you don't want to give feedback, you can permanently turn it off for yourself by setting CLAUDE_CODE_DISABLE_FEEDBACK_SURVEY=1 in your env or settings.json file. (docs: https://code.claude.com/docs/en/settings)
5. To permanently turn off the feedback survey for your whole company, set CLAUDE_CODE_DISABLE_FEEDBACK_SURVEY=1 in the settings.json checked into your codebase, or in your enterprise-managed settings.json (docs: https://code.claude.com/docs/en/settings)
6. You can also opt out of both telemetry + survey by setting CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC=1, or you can more granularly opt out with DISABLE_ERROR_REPORTING=1, DISABLE_TELEMETRY=1, etc. (also documented in the settings docs)
Security and privacy are very important, and we spend a lot of time getting these right. You have full control over data usage, telemetry, and training, and these are configurable for yourself, for your codebase, and for all your employees. We offer all of these options out of the box, so you can choose the mechanism that makes the most sense for you.
If there is a setting or control that is missing, or if anything is unclear from the docs, please tell us!
I recommend people always respond with the lowest possible score (1, not 0) when presented with popups like this.
Your entire answer ignores the fact that this is irritating behavior that ensures users are not having a good time. We don't want to chase down secret config values. We want to click "stop bothering me" and be done with it.
"Never ask me again"
(asks again the next day)
1. Does it take as much effort to opt-in to your feedback mechanism as it takes to opt-out? If not, why not?
2. If you want a thing ('feedback', 'a signal', data that is helpful to YOU), but getting it has this negative effect on others, what would happen if you preferenced others over yourself, and did with less of the thing?
Again, I think you are making the best product out there. I want to keep using it. Privacy is my #1 feature request to keep using it so transparency is crucial.
The setting is "leave me alone and don't ask again".
Get it in an air-tight legal agreement with some kind of audit provision, actual enforcement and penalties, or don't give out data you care about.
Yes they can, because data privacy laws forbid collecting data for one purpose and then using it for new ones without notice or consent.
And of course the EU does nothing but fine tech companies bazillions of euros for GDPR violations.
But it does lead to privacy protections.
(This is similar to how Americans have a religious belief that "regulations are written in blood" and so think it's immoral to ever undo any safety regulation; as a result their building code bans anything anyone has ever thought was bad with not a moment spared for cost-benefit analysis.)
Anthropic in particular is a PBC founded by weenies who thought the other labs weren't being safe enough. I believe up until a few months ago they never used any user data for training, even if you opted in.
No, i would have love to see them (Hello Microsoft, Google) but, no.
https://en.wikipedia.org/wiki/GDPR_fines_and_notices#Fines_a...
Plus this $377 million one.
https://www.cnil.fr/en/cookie-regulation-cnil-continuing-act...
How would you even know your data is being used in a way you didn’t authorize?
In the US, privacy policies change every week. And they don't gather consent - they just say "you're using our thing? Okay you consent". They'll send you an email with their new privacy policy. Which you don't read, because it's awful to read. And even if you do read it, it doesn't matter, because it's completely fake. The policy doesn't tell you how they actually use your data. It just says they can use any data for any purpose.
Per recent comment from Anthropic at https://github.com/anthropics/claude-code/issues/8036#issuec...
What is going on over there at Anthropic?
You can set CLAUDE_CODE_DISABLE_FEEDBACK_SURVEY either in your env, or in your settings.json. Either one works.
Stop asking me about accepting cookies.
Stop asking me to subscribe to your email list
Stop asking me to review my last purchase.
Stop telling me I need to subscribe to view this content.
Stop asking for my phone number
Stop asking for my income
If I want to do or tell you any of these things I will initiate that myself.
The problem is I'm pretty any one of us has (at most) 2-3 sites we actually want notifications from, and dozens asking.
Where I live, garbage disposal is a county contract. You get get whatever company your county has engaged. Do they think people would to move to another county for better garbage disposal?
Very occasionally these types of arrangements end up with an enthusiastically high performing company that does the right thing, but usually it's dumpster fires all the way down.
The purpose of the tool is to infer customer loyalty. What's the point of that in a captive market? I suppose whatever 3rd party is facilitating the survey gets paid and that's something.
Imagine a world where you don't need to click on anything because cookies are no longer being used for large scale tracking.
The banner is required every time there is processing of personal data where consent of required, whether that processing happened thanks to cookies or thanks to any other technical means (1px gifs, JavaScript fingerprinting, etc)
The banners are a fig leaf for behavior that violates the spirit of the GDPR, creating an aggravation where the simplest way to dismiss them is by agreeing.
Any site that doesn't offer a button to reject the tracking (with no more stops than angreeing) and still function as expected without the tracking, is in violation of the law.
Multiple times within a few minutes
During a damn incident I was trying to deal with
I left critical feedback. I wish someone would see it and feel ashamed, but it is rather clear that there haven't been decision makers in our industry capable of shame in many years.
I feel bad for you but... this is also kind of hilariously absurd/unaware of them.
It's very clear they didn't even take a moment to think through "Why do users access our tool, under what circumstance, and how does our tool treat them in that situation", where it would have been pretty clear that "interrupt and block access to the user until they provide feedback" is not a good UX for an engineer trying to do literally anything.
That wasn't their job I guess.
If it's anyone other than your bank or brokerage, that seems pretty weird and sketchy.
Every other time they ask it's voluntary on your part, and you should decline. They just use the information for advertising at that point.
Which sucks.
It's to the point where, when we broke down in a live lane on a dual carriageway the other day (flat tyre - actually shredded a run flat, newer car so no spare, all lay-bys closed so nowhere to pull off road and couldn't make it to next exit), the police came out and cordoned off the lane and then the AA guy who came and rescued us asked if we could write him a review when the feedback request came through.
Of course, on this occasion I did write him an absolutely glowing review (which he very much deserved, and which I was more than happy to do), because this was an incredibly dangerous situation - potentially life or death. I also sent a thank you to the local police force that helped us out.
But that's the point: it was life or death. It really mattered. So of course I wanted to say thank you, and the feedback mechanism provided a decent way to do that.
But most of these feedback requests are for things that don't matter that much, if at all, and are no better than spam, because of course everybody asks for it for every little interaction nowadays... and it's just endlessly tiresome.
So, yes: please stop.
(Btw, as someone who worked in market research for 7 years I can tell you that CX reviews skew towards the extremes - either very positive or very negative - and that you're much more likely to get a review if someone has a bad experience than if they have a good one. As a result, whilst these reviews can be good for qualitatively highlighting specific problems that might need to be solved, deriving any kind of aggregate score from them and expecting that to be representative of the average customer's experience is a fool's errand. Please don't do it. [Aside: I know, I know - this will stop no-one but I'd feel remiss if I didn't point it out, especially on this site where a lot of you will - I hope - get the point and apply it in your own businesses.])