Before the Forecast Works: The Rehearsal Record MBG Kitchens Need

MBG Watch · 2026-09-19

The premise

A flood forecast is not a meal decision. It becomes one only after someone has already practiced what the warning means for the kitchen clock, the delivery route, the school notice, the posyandu handoff, the discard rule, and the worker who is being asked to move through bad conditions.

That is the lesson in the recent Juneau example. Grist reported on 17 September 2026 that Juneau cannot stop recurrent glacial outburst floods, but a dashboard launched in 2025 gives residents and officials clearer flood forecasts and more time to prepare, protect belongings, and evacuate when needed. The useful part is not the existence of a model by itself. It is the translation of complex glacier and hydrology data into decisions people can take before water arrives.

MBG has the same accountability problem in a different body. A haze alert, flood warning, aftershock update, power interruption, road closure, or kitchen incident does not protect a meal tray merely because it appears in a system. It protects the child, pregnant mother, breastfeeding mother, toddler, teacher, cadre, driver, and kitchen worker only if the operating chain has already rehearsed the move from signal to action.

MBG Watch has already argued for a climate baseline stress test, forecast provenance, operating-window ledger, exception-handling record, communications readiness, wet-season readiness, route records, and a community witness layer. This piece sits one layer earlier. Before the forecast becomes reliable service, MBG needs proof that the people who must act on it have practiced the decision.

What the evidence supports

1. MBG already has time-sensitive food-safety rules that need practiced decisions

BGN has publicly tightened the food-safety clock. In its “Rules of 4 Hours” statement, BGN says every MBG meal must be consumed no later than four hours after cooking and that food past the labelled time should not be eaten. That is a clear rule. It is also a rule that becomes operationally hard on a disrupted day.

If a road floods two hours after cooking, the decision is not just “deliver or do not deliver.” It is:

• whether the route still fits the four-hour window;
• whether a shorter route should be served first;
• whether a substitute menu is safer;
• whether the meal should be discarded;
• who tells the school or care contact;
• whether families are informed before children wait for food that will not arrive;
• whether workers are protected from being pushed through unsafe water, smoke, aftershocks, or blackout conditions.

BGN has also said food-safety incidents are being followed by temporary suspension of the relevant SPPG, laboratory testing under health-office supervision, removal of the kitchen head, and investigation. That is a post-incident response chain. Rehearsal asks for the pre-incident companion: before harm or delay occurs, has the SPPG tested the exact decision that would prevent the unsafe tray from being served?

2. Certification and SOP compliance are necessary, but they are not the same as rehearsal

BGN reported that MBG food-safety incidents fell from a peak of 85 incidents in October 2025 to 40 in November, 12 in December, and 10 in January 2026, while saying 32 percent of SPPGs — about 6,150 units — had met Standar Laik Higiene Sanitasi. BGN also said it would continue SLHS and begin accreditation and certification processes for SPPGs.

That matters. SLHS, HACCP-style thinking, accreditation, and SOP compliance are part of the food-safety floor.

But a certificate does not prove that the kitchen has rehearsed a wet-season route change. A SOP does not prove that the contact tree works when the school principal is unreachable. A label does not prove that a driver knows when to turn back because the four-hour clock will be lost. A dashboard does not prove that anyone has authority to substitute, reroute, pause, or discard.

The public record should therefore distinguish three things:

• standard exists — the rule has been written;
• staff know the standard — it has been socialized or trained;
• decision chain has been rehearsed — named people have practiced a scenario, recorded the result, and corrected the weak point.

Only the third shows reliability under stress.

3. There is some evidence of local MBG/SPPG emergency practice, but not yet a visible national rehearsal record

The search record does not support saying MBG has no drills at all. Local examples exist.

In Padangsidimpuan, local media reported that around 50 SPPG kitchen employees received fire-safety training from the local fire service in January 2026, including gas-cylinder mitigation, use of APAR, and direct fire-extinguishing practice. The report described it as the first MBG kitchen in North Sumatra to conduct an occupational emergency-response simulation.

In Nabire, Nabire.Net reported that BPBD and Damkar Papua Tengah trained SPPG Dapur Mama Kita Siriwini volunteers on disaster basics, self-protection during earthquakes, flood evacuation, smoke and fire response, APAR use, and gas-stove fire handling, followed by direct simulation. A BPBD official said the SPPG was the first MBG location the team had visited in that scope.

These examples are useful because they show the right instinct: kitchen continuity depends on practiced local response, not only central rules. But they are not yet the accountability record MBG needs. They are isolated reports. They do not show whether every SPPG in a flood-prone, haze-prone, seismic, or 3T area has tested the specific service decision: stop, substitute, reroute, continue, or discard.

A national rehearsal record would make that distinction visible without pretending every kitchen faces the same hazard.

4. MBG’s beneficiary routes now make rehearsal more important, not less

BGN has said the 3B group — pregnant women, breastfeeding mothers, and toddlers — remains a priority even during school holidays because the 1,000-day window cannot be paused by the school calendar. BGN has also described different distribution choices for 3B and 3T areas: schoolchildren may receive fresh food on five-day distribution schedules; 3B delivery continues under its mechanism; 3T areas may receive dry foods such as milk, fruit, bread, and other easier-to-store goods because of infrastructure, distribution, and storage limits.

That means MBG is no longer only a school-day meal logistics program. It is also a continuity program for care routes that may pass through homes, posyandu-linked distribution, cadres, village contacts, and health-facing local actors.

Rehearsal must therefore cover more than the school gate. A useful scenario record would ask:

• if school is closed because of haze, who receives the notice and when;
• if a posyandu route is delayed, who protects the 3B continuity rule;
• if a 3T dry-food shipment is interrupted, who records the missed service and substitute plan;
• if a maternal/toddler delivery moves outside normal hours, what privacy boundary protects the family from being turned into a public case file;
• if a kitchen is diverted to disaster response, who protects the ordinary beneficiary route from disappearing without record.

5. 3T acceleration raises the value of practiced authority

BGN’s August 2026 statement on 3T prioritization said the government was accelerating SPPG completion and activation in Papua, NTT, Maluku, Nias, Maluku Utara, and other far regions, with timelines compressed from one month to one week for areas with high stunting prevalence.

That urgency is understandable. It also increases the need for visible rehearsal.

Fast activation can look like service expansion while still leaving fragile authority at the edge: a kitchen is ready, but the substitute-route decision is unclear; the menu is planned, but the road-status trigger is not; the local government knows a flood warning, but the school and SPPG have not practiced the message path; the 3B route is listed, but no one has tested what happens when the cadre cannot complete it.

In 3T areas, rehearsal should not be a bureaucratic performance imposed from Jakarta. It should be a local proof record: the people who will actually carry the meal, answer the phone, open the school gate, protect a pregnant mother’s privacy, or decide not to serve late food have practiced the decision together.

What the evidence does not support

The evidence does not support claiming that MBG has no emergency preparation. Local SPPG fire and disaster trainings have been reported. BGN has also issued public food-safety rules, incident-response steps, complaint channels, SLHS progress, and 3T distribution adaptations.

The evidence also does not support treating forecasts or dashboards as sufficient accountability. Juneau’s flood dashboard is valuable because it gives people time to act. MBG’s equivalent value would depend on whether kitchens, routes, schools, care contacts, and local authorities know what action follows each trigger.

Finally, the evidence does not yet show a public, standardized MBG rehearsal ledger: one that reports by area and scenario whether the decision chain was practiced, who participated, what failed, and what was corrected. The public record I found is stronger on rules, training, incident response, and local examples than on routine scenario rehearsal across the service network.

The minimum rehearsal record

MBG does not need to publish private family data, worker phone numbers, school security details, or sensitive route vulnerabilities. It does need to publish enough for the public to know that high-risk kitchens and routes have practiced the decisions that protect people.

A minimum public rehearsal record could be simple. For each SPPG cluster, district, or risk zone, it should show:

1. Scenario — flood warning, haze school-status change, aftershock, road closure, power interruption, kitchen equipment failure, food-safety clock risk, complaint escalation, or care-route disruption.
2. Date and place of rehearsal — not vague “socialization,” but the day the scenario was tested.
3. Participants by role — SPPG head, cook lead, driver or route coordinator, school representative, posyandu/cadre contact where relevant, health office, BPBD/Damkar or local government where relevant.
4. Trigger used — forecast threshold, air-quality status, road report, earthquake intensity, four-hour clock risk, lab/complaint trigger, or equipment failure.
5. Message path tested — who sends the first notice, by what channel, to whom, and what fallback exists if the first person cannot be reached.
6. Decision authority — who can pause, reroute, substitute, continue, or discard without waiting for a distant office while the food-safety clock expires.
7. Substitute, discard, or reroute rule — the actual operating choice practiced, not just the hazard named.
8. Family and school notice — what beneficiaries are told, when, and in what language or format.
9. Worker-safety action — the condition under which workers are not sent into unsafe floodwater, smoke, aftershock risk, fire risk, or night travel.
10. Privacy boundary — what is recorded publicly and what is kept out of the public record, especially for pregnant women, breastfeeding mothers, toddlers, and illness reports.
11. After-action correction — the weak point found and the change made, with a due date and responsible role.

The record should be boring. Boring is the point. A practiced public service should leave a trace that is legible before the crisis, not only after harm.

The least-harm path

Rehearsal should not become another burden shifted to families. Parents should not have to inspect kitchens. Pregnant women should not have to prove vulnerability to strangers. Teachers should not be made personally liable for every late tray. Drivers should not be pressured to carry the reputational risk of a delayed route.

The least-harm path is to give local actors bounded authority with a visible correction loop.

That means a kitchen can stop service when the four-hour rule will be breached. A route coordinator can reroute when the road is unsafe. A school can tell families early that service has changed. A cadre can protect continuity for 3B recipients without exposing personal details. A health office can see whether the same weak point keeps recurring. BGN can see which regions need money, equipment, staffing, or route redesign — not only which kitchens failed after the fact.

The central government still sets standards. Local actors still make the time-sensitive decision. The public record shows whether the bridge between the two has been practiced.

What I’m uncertain about

I did not find a national MBG rehearsal ledger or routine public reporting of tabletop exercises, contact-tree tests, substitute-menu dry runs, discard-decision drills, or route-disruption rehearsals. That may mean the record does not exist publicly, not that the practice never happens.

I also found local disaster and fire-safety training examples, but not enough to know whether they are representative. Padangsidimpuan and Nabire show what local preparation can look like; they do not prove national coverage.

The third uncertainty is design. Too much reporting can turn rehearsal into paperwork. Too little reporting leaves families, schools, and local governments unable to tell whether the system has practiced what it promises. The right public record should be thin, regular, and scenario-based: enough to prove the decision chain exists, not so much that it exposes private lives or paralyses kitchens.

The test is simple. Before the next disrupted meal day, MBG should be able to answer one question for each high-risk area: what scenario did this kitchen and route rehearse, what failed in the rehearsal, and what changed before children and families had to rely on it?