[{"data":1,"prerenderedAt":4},["ShallowReactive",2],{"article-body-the-schedule-is-obsolete-by-the-morning-meeting":3},"\nEvery manufacturing site runs a version of the same ritual. The\nscheduling run completes overnight: an optimised sequence,\nchangeovers minimised, due dates protected, capacity balanced. By\nthe time the morning meeting convenes, reality has filed its\namendments. An inbound truck is late, a machine failed its start-up\nchecks, two operators called in absent, and a priority order landed\nfrom sales. The supervisors do what experienced supervisors always\ndo, repair the plan by hand, and the overnight optimum spends the\nrest of its day as a reference document.\n\nNone of this is a failure of effort, and it is not fixed by a\nbetter solver. A schedule is a forecast about the near future of a\nphysical system, and physical systems disagree with forecasts\nhourly. The design question was never how to compute a perfect\nschedule. It is how to live well with the fact that no schedule\nsurvives contact with the shift.\n\n## The rescheduling trap\n\nThe instinctive software answer is frequency, rescheduling at every\ndisruption, continuously, optimally. This produces the disease\nplanners call nervousness, and it is worse than the ailment. A full\nreoptimisation is entitled to reshuffle everything, and it will\nhappily rebuild the entire sequence to harvest minutes of\ntheoretical efficiency. The costs of that churn are real and mostly\ninvisible to the solver: materials staged for a sequence that no\nlonger exists, changeovers begun for jobs that just moved, teams\nwho learn, correctly, that the plan is not worth reading.\n\nA plan that changes completely twice a day is not twice as\nresponsive. It is a plan nobody follows, which returns the plant to\nmanual scheduling with extra steps.\n\n> Optimality and instability come bundled. A schedule's value on\n> the floor is a product of its quality and its credibility, and\n> churn spends the credibility.\n\n## Disruption as triage, not recomputation\n\nThe mature architecture treats disruptions the way a good operation\ntreats any event stream: triage first, and reserve recomputation\nfor the events that earn it.\n\n1. **Absorb silently where buffers exist.** Most disruptions\n   threaten nothing: the late truck arrives inside the slack, the\n   short stoppage fits the buffer. The correct response is no\n   response, and the system's job is to establish that quickly and\n   quietly, rather than announcing every tremor.\n2. **Repair minimally where commitments are threatened.** When an\n   event does endanger a due date or a constraint, the objective\n   changes from \"optimal schedule\" to \"smallest stable change that\n   protects the commitment\": swap two jobs, shift a batch, hold\n   the rest of the sequence intact. Stability becomes an explicit\n   term in the objective, priced against the theoretical minutes a\n   full reshuffle would recover.\n3. **Rebuild rarely, and knowingly.** Some events, the line down\n   for days, the material out for a week, genuinely invalidate the\n   plan. Then a full reoptimisation is right, taken as a visible\n   decision with its disruption cost acknowledged, not as the\n   default response to every event.\n\nTwo disciplines make the triage honest. The thresholds are\nprobabilistic. \"The late inbound threatens the due date\" is a\nstatement with odds, computed from actual arrival and processing\nvariability, not a binary flag. And the triage is scored\nafterwards: which absorbed events should have triggered repairs,\nwhich repairs proved unnecessary, so the thresholds learn from the\nplant instead of from configuration defaults.\n\nRun this way, the morning meeting becomes a review of the handful of\ngenuine decisions, each arriving with options and odds attached, and\nthe schedule retains the property that makes any plan valuable on a\nshop floor: the people executing it believe it will still be true\nafter lunch.\n\nA diagnostic worth running at your own site: for one week, count\nhow many of the overnight schedule's first ten jobs actually ran in\nsequence. That number is your plan's real credibility, and raising\nit is what Prophesee's manufacturing module is for.\n[See what your schedule churn is costing](/contact).\n",1786833838578]