Most people who feel worn down by apps, dashboards and timelines wonder whether it is them or the work. We’re Good Citizens, a design and software studio, and this is our study of where that frustration goes, built so it can come out either way. It is not open yet.
When your day runs on someone else’s software, what happens inside you, and where does it go next?
Sometimes it lands on the tool. Sometimes it lands on the people around you, a coworker, a customer, someone at home. Sometimes it lands back on you. And sometimes it just goes. The usual advice is better habits, better boundaries, a better attitude. The work itself may also play a part, and the two are hard to tell apart from the inside. Neither explanation gets the benefit of the doubt here: the study is built to see where the frustration goes, whatever that turns out to be.
So the study asks one thing of each frustrating moment: where did most of it go? And it compares each person with themselves, one moment against another, so that who you are, what you do and how your week is going all cancel out.
One check-in, read honestly: whether software was part of the moment is read from what people wrote, so the question never puts the idea in their heads.
02 · Taking part
Built for people who are already tired
Taking part means a short screener to see whether the study fits you, about a week of check-ins by text or voice note, and two short surveys. There are no essays, no login and no homework.
It is for anyone whose day is shaped by apps, dashboards, tickets, routes or metrics, in any job and at any comfort with technology: support, healthcare, delivery and the trades, office and remote roles, and jobs that barely touch a computer but still run on other people’s schedules. You do not need to know anything about AI. We want people with a lot of automation in their day, a little, and almost none.
You can skip a day, skip a question, or stop, without telling us why. It is unpaid, and we would rather say so here than let you find out later. What you get back is what everyone gets: the findings in plain language, the methods, and the limits. We never contact anyone’s employer.
How each person takes part: about a week each, five to ten minutes on the days people take part. Recruitment stops at 90 people or six weeks, and the analysis runs once, after the last week closes.
03 · What we expect
And what would prove us wrong
Before anyone takes part, we write down what we expect to find and how big a difference has to be to count. If we turn out to be wrong, we publish that too. Read the plan.
The exact claims, and the size each has to reach, are in the plan rather than here, so that nothing you read before deciding whether to take part tells you what answer we expect. Read the plan after you have decided, or before, if you would rather. Either way, it can be checked: the questions, the rules and the weak spots are public.
04 · The fine print
What it can’t tell you
No study is neutral. These are its limits, written down before there is any data to explain away.
01
People who sign up already feel something
An unpaid, public study draws people who have opinions about work and technology, and they will have read what we think the answer might be first, which can change what they notice. That is why the main question compares each person with themselves, why the pages that recruit stay neutral about the answer, and why nothing we publish will claim how common anything is.
02
People report what they notice
If frustration with software is more memorable rather than more common, the moments people choose to tell us about will carry that. No design of this kind can remove it, so we say it first.
03
Association, not cause
A moment with an app in it may also be a busier moment. Comparing people with themselves removes what stays the same about them, not what changes from hour to hour.
04
People will drop out, and not at random
Unpaid daily check-ins lose people, and in a study about frustration, the frustrated may be the first to go. We will report who finished against who started, and we will not fill in the gaps.
05
Keeping track changes things
A week of noticing your frustration may change your frustration a little.
06
The result gets published either way
If the pattern is not there, that is the result, and it goes up in the same place, at the same length.
05 · Not open yet
What has to happen first
1 of these 7 steps are done. Nobody can take part until all of them are, and nothing has been collected yet.
A study about people living inside systems they did not design should not ask them to trust one either. So the order is fixed: the plan first, then the people who check our work, then you.
1 of 7 done. Each step is drawn as what it produces, in full once it exists. The order is fixed: the plan, then the people and safeguards that check our work, then you, drawn as an outline until someone joins.
01
Done
Write the plan down
The question, the hypotheses, what would count as no effect, the analysis and its limits, all fixed before any data. Reviewed once against the experience-sampling literature by an AI reviewer, and disclosed as such. Audited again on November 12, 2026, against survey-science practice on wording, scale design and order; the findings are folded into the plan.
02
Not yet
Try the questions on a few people
Ten to fifteen people we know, two or three days each, answering on a private page of their own (texting needs a registered number, which takes weeks), to check that the four places frustration can go cover what people actually say. Each answer also says when it happened and can say that none of the choices fit. Before the wording is frozen, five to eight people in the jobs the study is for (drivers, nurses, support agents) read the questions aloud and tell us what each one means to them. Their answers are never analyzed, and none of them can join the study.
03
Not yet
File the plan with an independent registry
OSF Registries timestamps it and gives it a DOI, so anyone can check that we did not move the goalposts after seeing the answers. Before it is filed, a survey methodologist who is a person reads the plan and the questions, and a simulation shows how many people it takes for the main question to be answerable.
04
Not yet
Independent ethics review
An outside review board reads the protocol, the consent and the plan for anyone who is upset by taking part, and approves them or sends them back.
05
Not yet
Consent and privacy, checked by a lawyer
Including what we would have to report, a separate permission for voice recordings, and how long anything is kept.
06
Not yet
A named person for anyone in distress
Someone who reads difficult entries and knows what to do, with crisis lines in every reply.
07
Not yet
Somewhere safe to keep answers
Contact details kept apart from what people tell us, names removed before analysis, and a deletion that actually deletes.
Read the plan, every hypothesis, rule and limit, in the words we will register. Or read the consent you would be asked to agree to, as it stands today. Or check the study: the questions word for word, where we think it is weakest, and a way to object.
We especially welcome questions, objections and methodologists: study@goodcitizens.us.