The question
When something goes wrong at work, the first question asked is usually whose hand it was. A coil comes off the mill below grade; a visit ends at a shut door; a date is copied into the wrong slot. A count of mistakes, or of talk about mistakes, starts from the same place: it gathers the moments when someone said that something was missed, forgotten or misheard, and asks how often they happen and to whom. What such a count cannot hold is the step before it, the moment in which a loss becomes a mistake at all.
This study asks how a scene written for comparison takes that step. It reads ten scenes from GAEKJU, a studio of the RISCON family that studies talk about people's flaws at work. GAEKJU sorts the talk it studies into seven kinds, and calls one of them "Missteps": the talk of slips, wasted effort, makeshift fixes and things got wrong. Many of GAEKJU's questions of this kind are scenes it has written itself, set in trades from a forest-fire lookout to a hospital pharmacy. In GAEKJU's own words, such scenes are written for comparison and are not testimony from those trades. The ten read here are the scenes of that kind whose evidence, the part where a scene says what it rests on and what it leaves open, holds no digit.
The research question has two parts. First, what in these scenes decides whether a loss counts as a misstep: the hand, the size of the loss, or something else? Second, where does each scene place the correction, and who carries the work that follows it? The scenes are the only evidence. They are evidence of how GAEKJU frames the question, not of what happens in any steelworks, hotel or building site, and this study draws no fact about any trade from them.
The contribution
The study makes three claims about the form of the scenes. The first is that the scenes let a record decide what counts as an error. A loss written in advance in a production plan is no one's mistake; the same loss with no line in the plan becomes an incident report with a name on it. Six of the ten scenes set two records, two marks, two times or two appointments against each other, and find the slip in the gap between them.
The second claim is that half the scenes read a slip, a workaround or a wasted trip as news: a sign that the design missed a size, that a door is shut because of the hour and not the resident, that a status screen has merged two meanings. Where a scene reads the slip this way, its judgment mostly declines to end in a person's fault, and in one case it explicitly puts off both praise and blame.
The third claim is that the scenes' judgments and proposals ask the correction to land in a place that can keep it, such as the official drawing, a slot in the plan or a shared schedule, and that most proposals divide the work after the slip so that it does not stay with the one person who found or admitted it.
Together these claims suggest a location for error that neither of the two familiar answers quite names. In the scenes that set two records side by side, they do not blame the hand, and they do not dissolve the slip into a whole system. They point to the seam between two records that were supposed to agree. That location is the paper's contribution, and the conclusion returns to it.
The debate
Research on error at work has argued for decades about where to look when something goes wrong. Four lines of that argument bear on the scenes.
The first concerns the person and the system. James Reason, Human Error: Models and Management (2000), sets out two ways of viewing the problem. The person approach focuses on the unsafe acts of people on the front line, sees them as arising mainly from aberrant mental processes such as forgetfulness, inattention and carelessness, answers them with procedures, discipline, retraining and blame, and tends to treat errors as moral issues. The system approach starts from the premise that humans are fallible and errors are to be expected even in the best organisations; it sees errors as consequences rather than causes, with origins in upstream systemic factors, including recurrent error traps in the workplace. Diane Vaughan, The Dark Side of Organizations: Mistake, Misconduct, and Disaster (1999), approaches the question through sociological constructs: reviewing four literatures, Vaughan treats mistake as one form of routine nonconformity, how things go wrong in socially organised settings, and finds such nonconformity systematically produced by the interconnection between environment, organisations, cognition and choice.
The second line concerns what gets counted. Erik Hollnagel, Safety-I and Safety-II: The Past and Future of Safety Management (2018), observes that safety has traditionally been defined as a condition in which adverse outcomes are as few as possible, so that safety is measured by counting the cases where it fails rather than those where it succeeds. Hollnagel proposes a second view, in which safety is the ability to succeed under varying conditions and is measured by what goes right in everyday work. The difference matters for any record of mistakes: a ledger that lists only what was lost has no column for what was kept.
The third line concerns what to do after an error. Michael Frese and Nina Keith, Action Errors, Error Management, and Learning in Organizations (2015), argue that organisations often focus on error prevention as a single strategy, and that prevention needs to be supplemented by error management, an approach directed at dealing effectively with errors after they have occurred, minimising their negative consequences and maximising their positive ones, such as learning and innovations. Cathy van Dyck, Michael Frese, Markus Baer and Sabine Sonnentag, Organizational Error Management Culture and Its Impact on Performance: A Two-Study Replication (2005), describe such a culture as norms and common practices, communicating about errors and detecting, analysing and correcting them quickly, and report it positively related to firm performance in two studies in two European countries. This line asks less where an error came from than where it goes next.
The fourth line concerns the workaround, the move that keeps work going when the intended way is blocked. Steven Alter, Theory of Workarounds (2014), observes that workarounds are mentioned often but understudied and undertheorised, and offers a process theory of how and why they are created, meant among other things for understanding compliance and noncompliance and for studying how workarounds sometimes lead to larger planned changes in systems. Jonathon R. B. Halbesleben, Douglas S. Wakefield and Bonnie J. Wakefield, Work-arounds in Health Care Settings: Literature Review and Research Agenda (2008), set out how work-arounds can be told apart from errors, mistakes and deviance, link them to safety outcomes, and discuss how open discussion can let them improve work processes. Deborah S. Debono, David Greenfield, Joanne F. Travaglia, Janet C. Long, Deborah Black, Julie Johnson and Jeffrey Braithwaite, Nurses' Workarounds in Acute Healthcare Settings: A Scoping Review (2013), describe workarounds as behaviours that circumvent or temporarily fix perceived hindrances, and conclude that they enable, yet potentially compromise, the execution of care: in some contexts necessary, in others counterproductive.
One further work stands at an angle to the others. Stephen R. Dixon, Christopher D. Wickens and Jason S. McCarley, On the Independence of Compliance and Reliance: Are Automation False Alarms Worse Than Misses? (2007), report an experiment in which automation prone to false alarms hurt overall performance more than automation prone to misses. Their alarms came from a machine, not from a person; the scene of the fire lookout below asks what happens when the one who raises a false alarm is a colleague and the one who answers it is another.
These lines meet the scenes at different points. The person and the system frame the question of blame. The count of failures and of successes frames what a record can see. Error management frames what comes after. The workaround frames the slips that are also fixes. The scenes, this study argues, add a question the lines leave in the background: which record decides that a loss was a mistake, and which record has to change for the correction to hold.
The shape of the paper
Chapter 1 sets out the data and the method, then reads the scenes in three sections: what names a misstep, the slips the scenes read as news, and where the correction lands and who carries the rest. An appendix gives the ten scenes word for word in this study's English. The conclusion discusses what the scenes add to the debates, states the contributions again, and names the limits of reading written scenes and the research that could follow.