An Overview of Human Error Drawn from J. Reason, Human Error, Cambridge, 1990OutlineDependability and human errorLearning from other fields: PSTNSlide 5Learning from experimentsSlide 7Slide 8A theory of human error (distilled from J. Reason, Human Error, 1990)A theory of human error (2)Origins of error: the GEMS modelGEMS and errorsError frequenciesError detection and correctionSlide 15Human error and accident theoryAccident theory (2)Slide 18Addressing human errorThe Automation IronyBuilding robustness to human errorSlide 22Slide 23Human error: the ROC approachRepairing the Past with UndoUndo detailsSummaryAn Overview of Human ErrorDrawn from J. Reason, Human Error, Cambridge, 1990Aaron BrownCS 294-4 ROC SeminarSlide 2Outline•Human error and computer system failures•A theory of human error•Human error and accident theory•Addressing human errorSlide 3Dependability and human error•Industry data shows that human error is the largest contributor to reduced dependability–HP HA labs: human error is #1 cause of failures (2001)–Oracle: half of DB failures due to human error (1999)–Gray/Tandem: 42% of failures from human administrator errors (1986)–Murphy/Gent study of VAX systems (1993):0%10%20%30%40%50%60%70%80%90%100%1985 1993Causes of system crashesTime (1985-1993)% of System CrashesSystemmanagementSoftwarefailureHardwarefailureOther53%18%18%10%Slide 4Learning from other fields: PSTN•FCC-collected data on outages in the US public-switched telephone network–metric: breakdown of customer calls blocked by system outages (excluding natural disasters). Jan-June 20019%47%17%5%22%Human-co.Human-ext.Hardware FailureSoftware FailureOverloadVandalismHuman error accounts for 56%56% of all blocked calls–comparison with 1992-4 data shows that human error is the only factor that is not improving over timeSlide 5Learning from other fields: PSTN•PSTN trends: 1992-1994 vs. 2001Cause Trend1992-942001Human error: company98 176Human error: external100 75Hardware49 49Software15 12Overload314 60Vandalism5 3Minutes Minutes (millions of customer minutes/month)Slide 6Learning from experiments•Human error rates during maintenance of software RAID system–participants attempt to repair RAID disk failures»by replacing broken disk and reconstructing data–each participant repeated task several times–data aggregated across 5 participantsError type WindowsSolaris LinuxFatal Data Loss Unsuccessful RepairSystem ignored fatal inputUser Error – Intervention Required User Error – User Recovered Total number of trials 35 33 31Slide 7Learning from experiments•Errors occur despite experience:•Training and familiarity don’t eliminate errors–types of errors change: mistakes vs. slips/lapses•System design affects error-susceptibilitySlide 8Outline•Human error and computer system failures•A theory of human error•Human error and accident theory•Addressing human errorSlide 9A theory of human error(distilled from J. Reason, Human Error, 1990)•Preliminaries: the three stages of cognitive processing for tasks1) planning»a goal is identified and a sequence of actions is selected to reach the goal2) storage»the selected plan is stored in memory until it is appropriate to carry it out3) execution»the plan is implemented by the process of carrying out the actions specified by the planSlide 10A theory of human error (2)•Each cognitive stage has an associated form of error–slips: execution stage»incorrect execution of a planned action»example: miskeyed command–lapses: storage stage»incorrect omission of a stored, planned action»examples: skipping a step on a checklist, forgetting to restore normal valve settings after maintenance–mistakes: planning stage»the plan is not suitable for achieving the desired goal»example: TMI operators prematurely disabling HPI pumpsSlide 11Origins of error: the GEMS model•GEMS: Generic Error-Modeling System–an attempt to understand the origins of human error•GEMS identifies three levels of cognitive task processing–skill-based: familiar, automatic procedural tasks»usually low-level, like knowing to type “ls” to list files–rule-based: tasks approached by pattern-matching from a set of internal problem-solving rules»“observed symptoms X mean system is in state Y”»“if system state is Y, I should probably do Z to fix it”–knowledge-based: tasks approached by reasoning from first principles»when rules and experience don’t applySlide 12GEMS and errors•Errors can occur at each level–skill-based: slips and lapses»usually errors of inattention or misplaced attention–rule-based: mistakes»usually a result of picking an inappropriate rule»caused by misconstrued view of state, over-zealous pattern matching, frequency gambling, deficient rules–knowledge-based: mistakes»due to incomplete/inaccurate understanding of system, confirmation bias, overconfidence, cognitive strain, ...•Errors can result from operating at wrong level–humans are reluctant to move from RB to KB level even if rules aren’t workingSlide 13Error frequencies•In raw frequencies, SB >> RB > KB–61% of errors are at skill-based level–27% of errors are at rule-based level–11% of errors are at knowledge-based level•But if we look at opportunities for error, the order reverses–humans perform vastly more SB tasks than RB, and vastly more RB than KB»so a given KB task is more likely to result in error than a given RB or SB taskSlide 14Error detection and correction•Basic detection mechanism is self-monitoring–periodic attentional checks, measurement of progress toward goal, discovery of surprise inconsistencies, ...•Effectiveness of self-detection of errors–SB errors: 75-95% detected, avg 86%»but some lapse-type errors were resistant to detection–RB errors: 50-90% detected, avg 73%–KB errors: 50-80% detected, avg 70%•Including correction tells a different story:–SB: ~70% of all errors detected and corrected–RB: ~50% detected and corrected–KB: ~25% detected and correctedSlide 15Outline•Human error and computer system failures•A theory of human error•Human error and accident theory•Addressing human errorSlide 16Human error and accident theory•Major systems accidents (“normal accidents”) start with an accumulation of latent errors–most of those latent errors are human errors»latent slips/lapses, particularly in maintenance•example:
View Full Document
Unlocking...