This is untimed educational practice for one applied skill. It is not a clinical, diagnostic, employment, or professionally recognized assessment, it does not produce a score or credential, and for the judgement and safety families it is not a substitute for your employer's own policies, training, or legal duties.
A steady drift is the only one of the four readings that contains information about the future, and acting before the limit is what leaves time to act at all. Waiting for the crossing is the procedurally defensible option and throws away the warning: the alarm confirms what the trend already told you, twenty minutes later.
A fixed interval bounds how long any change can go unnoticed, and an audible alert converts the out-of-view display from a visual task into one that interrupts you. Watching the important one continuously is the intuitive allocation and gives the second display an unbounded detection delay, which is where the incident will come from.
Rare events give attention nothing to hold on to, and detection rates for infrequent signals fall steeply within the first part of a watch. The busy display feels harder because it is more effortful moment to moment, and effort is not the same thing as reliability. The busy display keeps re-engaging you, which is what the quiet one cannot do.
The decline in vigilance is quick and largely independent of motivation, which is why it is treated as a design constraint rather than a discipline problem. Assuming a motivated operator holds up is the comfortable answer and is the assumption behind most monitoring failures: it makes the fix an instruction rather than a rota.
You currently know one thing for certain, that at least one instrument is wrong, and acting on either number without resolving that is acting on unknown data. Believing the reading that matches expectation is the most dangerous option precisely because it feels like judgement: it is confirmation bias with an instrument attached.
Forty false alarms have already trained everyone that this signal means nothing, so the system is now less safe than if the alarm did not exist. Instructing people to keep investigating is the conscientious-sounding answer and is unenforceable: no amount of instruction survives a fortieth false positive.
Sensitivity and false alarms move together; there is no setting that reduces both, so the question is always which error you can better afford. 'As sensitive as possible' is the safety-flavoured answer and produces the alarm flood of the previous item, which converts a low miss rate on paper into a high one in practice.
Eleven simultaneous faults are far less likely than one fault with ten consequences, so the first move is to find the event rather than to work the list. Taking them in order of arrival is the disciplined-looking approach and can leave you addressing symptoms in sequence while the cause continues.
An alarm is a demand for action; anything that does not demand action belongs on a display or in a log where it can be consulted rather than responded to. Keeping it at low priority is the compromise that most alarm systems actually make, and it is how lists grow to hundreds of entries that operators learn to scroll past.
Silencing is legitimate and incomplete: the state has to stay on screen, because the sound was the only thing still reminding you it existed. Resetting it is the tidier-looking action and is the genuinely dangerous one. It clears the record of a condition that has not been dealt with.
Two tasks that both demand decisions compete for the same limited control resource, so they interleave rather than overlap, and the reassembly cost at each switch is real. 'Different senses' is the most technical-sounding wrong answer and only helps with the input bottleneck; the decision bottleneck is where the competition actually is.
Resumption errors come from losing your place, not from losing your knowledge, and a two-second note is what preserves the place. Finishing the current step is the sensible-sounding version and does not help: you still return with no record of what the step after it was, which is exactly the information that goes missing.
Automatic tasks consume little of the resource the demanding task needs, which is why walking and talking works and reading and talking does not. 'Any two familiar tasks' is the near-miss: familiarity helps, but a familiar task that still requires decisions competes just as hard as an unfamiliar one.
Redoing a short verified stretch is cheap; a skipped item in a checking task is exactly the failure the task exists to prevent. Carrying on from where it looks like you stopped is the efficient-seeming choice and relies on the one judgement you have just established you cannot make.
A record is the natural unit to finish because the interrupted state is small at its boundary and large in its middle. Answering immediately is the courteous option and is where transposed and omitted fields come from. The interruption lands inside the record rather than between two of them.
Re-ranking is the right response and telling the dependants is what stops your change from breaking their plans. Re-ordering quietly is the efficient half of the answer and is how somebody else discovers at the deadline that the thing they were promised moved three hours ago.
A limit encodes a judgement about a consequence you may not be able to see from the control room, and an hour outside it is a reportable condition regardless of whether anything visible has happened. Widening the limit is the most damaging option because it is self-justifying: every exceedance becomes evidence that the limit was too tight.
State, changes, open items, and open uncertainties is the set that lets the next person carry on rather than start again, and doing it in person lets them ask. A verbal briefing alone is the strongest wrong answer because it does carry more nuance, and it leaves nothing that a third person, or an investigation, can consult afterwards.
Two hours is still early enough to check whether anything downstream was affected, and the circumstances of the miss tell the designers something about the workload at that moment. Waiting for handover is the plausible middle option and delays both the check and the information by hours, for no gain at all.
Automating the routine leaves a human with a rare-event vigilance task and a manual skill they no longer use daily, the two things people are worst at, so practice has to be scheduled rather than assumed. Reducing attention in proportion is the intuitive reading of the change and inverts it: the remaining work is a smaller amount of much harder monitoring.