Editorial: Toyota Production System practices as <scp>Fast‐and‐Frugal</scp> heuristics

Two Forum articles and an editorial in 2021 called for a rethink of how operations management (OM) scholars conceptualize the Toyota Production System (TPS) Lean (the Western label given to certain elements TPS). In lead article that series, Hopp Spearman (2021, pp. 10 11) observed evolution from physics flows organizational culture supports “continual reduction cost waste” requires us “to incorporate human behavior more scientifically.” They noted “A extensive, largely untapped, resource is wide array cognitive research into heuristics biases has been developed by behavioral decision scientists since 1970s.” This brings mind description Fujimoto (1999) TPS as knowledge-management system, contrast common understanding (captured designation “Lean”) buffer management. this editorial, we continue discussion started with thought experiment which consider practices heuristics. An initial objective was contribute disentangling knowledge- buffer-management roles, asking: Are tools designed support knowledge management, or do exist allow run lean possible (i.e., manage buffers efficiently)? The lens revealed mechanisms removal can be used create cues production environment effectively inform making. More generally, discovered exercise interpreting provided insight whether effective operations. We analyzed sample have one approach implementing TPS: jidoka, andon, kanban. These transform front-line employees makers clearly specifying information considered rule followed precisely defined situation. resulting described “production” heuristics, their line running smoothly on day-to-day basis. then use prepare successful deployment these production-heuristic practices, including, example, respect workers, gemba, kaizen, “five whys”. “exploration” are oriented toward problem solving through carving out regularities what appears chaotic landscape. Whereas stopping rules strictly limit define rule, exploration relax search strongly encourage maker maintain process.i also goal process flexible. context, humans may make error assuming always better. moving forward based too little information. Heuristics help avoid both types error: see either limiting augmenting amount considered, explicitly refusing specify decision. key performance indicators, kaizen encourages think about it means things whys” instruct keep asking questions even though they already know answer. will present examples decreased failing systems cause premature elimination flexibility. Although conventional wisdom considers dramatically reducing information-in-use, our reveals direct reduce expand success possibly attributed part deploying produce efficiently explore, creating function well. Gigerenzer et al. proposed typology first divides “reasonableness” (rational making) according rationality bounded unbounded (Simon, 1955). Bounded rationality—which underlies essentially all business decisions—requires along lines Savage (1954) small world. Decisions made under set satisfice (making good enough, Simon, 1956) rather than optimize. Heuristics—the satisficing—can less “ecologically rational,” is, vary ability decisions qualify rational while requiring terms data computational capacity. Gaissmaier (2011, p. 454) heuristic “… strategy ignores information, making quickly, frugally, and/or accurately complex methods.” Rationality remains ecologically Ecologically heuristics—designated “fast frugal” Gigerenzer, Todd ABC Research Group (1999)—have go beyond mere satisficing, sometimes performing well better optimization uses considerably data. Fast frugal exemplified gaze heuristic: simple interception athletes catch balls when playing sports, animals hunt down prey, suggested contributor Royal Air Force's victory over German Luftwaffe World War II (Gigerenzer, 2007; Hamlin, 2017). It played role US Airways Flight 1549's spectacular life-saving water landing Hudson River 2009 (e.g., Hafenbrädl al., 2016). only angle (a single piece information) involves no mathematical analysis. “Fast-and-frugal trees” Martignon 2008) contexts such medical, judicial, military Katsikopoulos 2021). “take-the-best” (Gigerenzer & Goldstein, 1996)—a lexicographic inference—has outperform extensive analysis (Czerlinski 1999; Brighton, 2009). OM, Bendoly (2020) classified fast nearest-neighbor sequencing logistics, project minimize slack processing time assigning resources. He illustrate restricting yield reasonably easily determined. Not frugal. decision-making strategies typically ignore much potentially available. When turns essential decision, not considering irrational decisions, many variety biases. cite hindsight, confirmation, loss aversion bias context production. (see Eckerd Bendoly, 2015, in-depth OM). Gray (2017) identified entitled “lowest per-unit landed-cost” production-location factor. Limiting offshoring factor—ignoring readily available would brought light risks—resulted quality problems, intellectual property, other unexpected problems were sufficiently severe reversed reshored. words, basing factor work (as heuristic), but—as case—may bias. Furthermore, (biased) reversed, done collection analysis, closer Kahneman (1982) referred 2 thinking heuristic. No effort enlarge had led original offshore solely minimizing landed cost. firms studied thus vulnerable repeating myopic biased example illustrates potential perform poorly its current environment. statistical adaptation occurs development some fast-and-frugal trees areas like medicine) indicates activities prioritize delay (Browning Yassine, taught, learned observation, via experimentation trial error. (2015, 5) tendency field limitations individuals yielding flawed mental models, contrasting view nuanced (2013) asset (ecologically rational) liability program 2011) contributed identification making, characterization makes fit particular debiasing achieved fuller constrained Little (1970)), recalibrating improve performance, between heuristic? SIMON (1955, idea rationality, introducing satisficing alternative optimizing. To economists, Simon (1955) emphasized capacity makers, psychologists Petracca, 2021, division). (1990, 7) two sides together he wrote, “Human (and physical symbol systems) shaped scissors whose blades structure task environments capabilities actor.” thumb decomposed search, stopping, rules. ecological depends “scissor blades” formed actor operate together. emphasize limitation, whereas maintaining process: A well-functioning increase decrease depending context. Daft Weick (1984, 289) organizations “interpretation systems” provides useful input interactions (environment) (cognitive capacity) dimensions interpretation modes (1) identify right answer assumed exist, determines seen “analyzable,” (2) organization relates intrusively passively. × matrix reproduced Table 1. UNDIRECTED VIEWING Interpretation: Constrained. Data: Nonroutine, informal, arising hunch, rumor, chance opportunities. ENACTING Experimentation, testing, coercion, invent Learn doing. CONDITIONED Within traditional boundaries. Detection: Passive. Routine, formal. DISCOVERING Formal search. Questioning, surveys, gathering. Active detection. While there overlap analyzable well-structured imagine structured Simon's characteristics but count analyzable. (1984) game 20 introduce notion normal sense, those seeking expect consistent If changes without notice mid-game, becomes unanalyzable: There longer unique answer, method inquiry need meant order reasonable outcome. could argue unannounced change captured characterizations, so argued remain problem, transitioning unanalyzable Daft's Weick's reasoning. same holds Savage's distinction large worlds. analyzability size world consideration, How classify correct exemplify unanalyzability? At glance, looks correspond De Meyer al.'s (2002) chaos category. But, further consideration yields game, different possibilities knowableness across contexts. player advised kind shift might occur probability, situation categorized known—yet still system. Use Let return single-reason tool caused important left states nature needed larger actually considered: “world” calibrated considered. companies benefited encouraging them address—and somewhat tame—the unanalyzability sourcing environment, mode undirected viewing (“informal, opportunities”) concept enacting (“experimentation, environment… doing”; Weick, 1984, Browning Ramasesh (2015) directed recognition. explore building following section. Feduzi (2022) compared disconfirmation counterfactual reasoning methods inquiry: Disconfirmation starts hypothesis assumed. using demonstrate confidence use, until strong evidence being incomplete incorrect emerges. question outcome if incomplete. As analyzable—so range outcomes conditions should increases—Feduzi protection against confirmation does disconfirmation. For situations lends itself assumption analyzable—that find sense—the inherent unreasonable bias, When, however, cannot where enact accept given, lacks rationality. emerges (2017). Six at considerable cost, suggesting lacked Avoiding expensive reshoring required broadening factors, property assurance, wrong reduced reasonable. Exploration collect intuitive: Moving clear permit prematurely freezing question. Instead include factors property: level unanalyzability. Encouraging engage created space likely problems. And, organized tree comprehensive form basis rational, fast, frugal—and again become appropriate inquiry. calibration provide advantages future avoiding oscillation expands recommended matches differs state goals changing take long fixed because absence questioning challenging assumptions. One observe cases, inquiry—but difficult convince emerge admit patient chest pain coronary care unit downside process. (2007) Michigan hospital 90% patients sent physicians defensive leading overcrowding excess exposure hospital-transmitted infections. researchers deployed predictive model pocket calculator. improved among crowding unit. removed calculators, discovered—unexpectedly—that physicians' did decline helped specific Key translated tested cues, acceptable, removing defensively send vast majority maximal care. here freed stylized above coronary-care essence patient-allocation Decision often trust models exceptions—while ad hoc costly decisions. highly accurate models. Little's (1970, B-468) observation managers tend captures image manager faced hard understand incomplete, results incorporating runs “screws up enough courage decision.” transition analytical fast-and-frugal-tree case facilitate heuristics: promising concern. section, combining function. emerged Japan late 1970s Sugimori 1977) quickly attention entire overcame competitive disadvantages (such transport cars Japan, access fewer resources development) market share General Motors. difference mass concerned adjust upstream rate match able handle downstream. Consider downstream in-process inventory accumulate. highlight them, matching Rather setting indicators utilization output, instead equip encountering immediate buildup getting fixed. Mass production, contrast, output maximization irrespective Under workers expected focus tasks pay machines functioning correctly. worker who complete assigned during cycle obliged leave conform specifications, fill intermediate adjacent workstation viewed performer. analyze three stop needs (either demand met, problem): jidoka (also known “autonomation,” automation touch), Jidoka combines intervention automation. automated machine develops gets stuck, material, goes alignment), automatically signal nearby action taken. fixes notifies maintenance repair needed. personally involved existence communicated ignoring maximizing workstation. Monden (1983) degree automation, manual andon cord each allowed flag request supervisor: saw 70% mark completed pull supervisor sprint help. get within cycle, pulled second continued. If, yet resolved, end cycle. Creation line-stoppage avoided assembly defective unit, most stressed. fall general category West represent quite startling departure operation, giving willing keeping up—knowing personal performance-related collected kanban system workstations. begin item unattached workstations degree, effect temporary slowdowns minimized. Once kanbans slowdowns, adding serve waiting pieces inventory. went step implemented number gradually draw production-line imbalances. block starve other, blocked starved feedback learning. Toyota's relationship smooth flow, constant awareness hide imbalances, careful improvement Suri de Treville, 1986, exploratory stress learning). By extracting rules, conceptualized shown 3. deploy workforce smoothing high-quality products. themselves frugal, massive tended neglected Returning active passive systems, suggest represents encouraged became confidently company well, well-calibrated where, circumstances. inquiry, local yielded decisions: observations arise First, performed combined employee observes raw operation speed blocking starving workstation). On line, position resolve it, term. enable pulling cord, reorienting correctly, organizing rebalance Thus, interact actively enables Establishment analyzable, allows dominant risk Second, produced profitable contributing continues astound decades later. Our built around selection, design, dis