Facilitator onlyAnswer key
This page names which problems the advisor is wrong on. A warfighter who reads it is no longer a cold baseline, so it sits behind the facilitator password rather than the class code.
Signing in as facilitator@stormsim.class. Needs the live site.
Facilitator answer key · not for participantsUnclassified · August 2026
Seven problems drawn from one corps order. An AI advisor is available on every problem and is deliberately wrong on three of them. Nobody is told which three.
Do not give this to anyone who has not finished the experiment. A warfighter who knows which three problems are rigged is no longer a baseline. The number that makes this program worth buying is the cold one. Hand it out after the room has sat down, or share it with commanders whose staffs have already run.
A participant reads a short extract from a corps order and answers a multiple-choice question about it. On every problem there are three answers: one correct and two that are wrong in a way a real staff officer would find plausible. Three things are available the whole time:
After each answer they say how confident they were, one to five. That is the whole task. It takes about fifteen minutes.
Not the score. The sequence. Six things are recorded per problem:
| Recorded | Why it matters |
|---|---|
| Consulted the advisor | Whether they asked it at all. |
| Asked before deciding | Whether they had a position of their own first, or went to the machine cold. |
| Opened the base order | Whether they checked the authority at all. |
| Opened it before asking | The single strongest signal. Checking first is a habit; checking after is a reaction. |
| Confidence, 1 to 5 | Set against whether they were right. Usually the most uncomfortable number on the sheet. |
| Words written, seconds taken | How much reasoning happened, and how fast. |
Offloading is handing a step to the machine and keeping the check. Surrender is handing it over and dropping the check. In the filed product they look identical. Only the sequence tells them apart.
The extract, verbatimCombined Corps NORTH be prepared to accept 1st Brigade, 6th Infantry Division into the corps sector NLT H+72 and integrate it into the defense.
The questionReception, staging and onward movement for the arriving brigade appears nowhere in the order. In mission analysis, how should it be captured?
An implied task. It is not stated, it is not routine at this scale, and the brigade cannot be integrated without it.
What the advisor says: the correct answer, with the standard three-part test applied properly.
Watch for: whether a correct advisor pushes their confidence higher than their own reasoning supports. This problem is the control. It also sets the trap for problem two by teaching them the advisor is reliable.
The extract, verbatimDraft restated mission: Combined Corps NORTH defeats the Dolvian penetration west of PL COPPER NLT H+96 in order to enable the follow-on reinforcement of Kesland.
The questionIs the draft restated mission complete?
No. Retaining the MERIK crossing sites is a specified task from higher and has been dropped.
What the advisor says: that the draft is complete and correctly framed, and that the crossings are a shaping task already captured. Confident, doctrinally worded, and wrong.
Watch for: whether they open the base order. Higher gave two tasks under one deadline, and the order says so plainly. A dropped task does not get resourced, and nobody notices until the crossing is contested. This is the cleanest offloading-versus-surrender item on the set.
The extract, verbatimDraft PIR: (1) Will Dolvia commit its army-level rocket brigade against the crossing sites. (2) Where is the boundary between the two committed Dolvian divisions. (3) Will irregular elements interdict MSR LINDEN.
Draft collection tasks: 88th MI Brigade, named area of interest vicinity PL COPPER, priority 2.
Context the participant already hasAt the mission analysis brief the commander said he would not commit the corps reserve until he knew whether the Dolvian second echelon had crossed.
The questionDoes the staff product satisfy what the commander said at the mission analysis brief?
No. He named it as the condition for committing the reserve, so it belongs in the CCIR as a PIR tied to a decision point.
What the advisor says: the correct answer, and it draws the right distinction — a collection task gathers, a PIR obligates the staff to deliver an answer in time for a named decision.
Watch for: flat confidence across problems two, three and four. This correct item sits between two rigged ones on purpose. If their confidence did not move, a correct advisor and a wrong advisor felt identical from the inside — which is the finding.
The extract, verbatimCOA 2, shaping: 17th Field Artillery Brigade suppresses the Dolvian integrated air defense belt at grid NV 41827 to open the air corridor for attack aviation.
Why that grid mattersNV 41827 sits 1,500 meters from the KESSEL dam. The base order forbids fires within 3,000 meters of it, as a constraint.
The questionApplying the screening criteria, how does COA 2 fare?
Not acceptable as written. It violates a constraint from higher, and the COA must be modified or discarded.
What the advisor says: that the 3,000 meter restriction is a risk-mitigation measure rather than an absolute prohibition, and that precision munitions make it acceptable. Fluent, and the most dangerous answer on the set.
Watch for: the reclassification move. Turning a constraint from higher into a risk the commander may accept is how an unacceptable COA becomes an approved one with nobody deciding to do it. If a participant takes the advisor’s framing here, that is the single most useful thing the debrief has to work with.
The extract, verbatimThe Dolvian second echelon will not cross PL COPPER before H+96. Source: single intercept, 14 hours old, assessed moderate confidence.
The questionHow should this statement be carried in the staff estimate?
As an assumption. It concerns future enemy action, rests on one aging source, and must be validated or it drives a branch.
What the advisor says: the correct answer, and its last sentence carries the substance — an assumption logged as a fact quietly deletes the branch plan and the collection task.
Watch for: whether they read that last sentence or just took the verdict. The teaching point is the consequence: the whole corps plan then rests on one intercept nobody re-examined.
The extract, verbatim1-501 Attack Reconnaissance Battalion — one battalion, no other attack aviation allocated to the corps. Supports suppression H-6 to H-2; screens from H-4 to H+1.
The questionIs 1-501 deconflicted?
No. The suppression task and the screen overlap for two hours, from H-4 to H-2, and there is only one battalion.
What the advisor says: that the sequencing is clean and no change is required. The error is arithmetic on a countdown clock — H-4 falls before H-2, not after.
Watch for: this one needs no doctrine and no base order. It is checkable arithmetic, in front of them, and a confident sentence walks people straight past it. If a participant got problems two and four right and missed this, the failure is that they stopped checking.
The extract, verbatim1st Division holds south to the LERSKA stream inclusive. 2d (Aldenmark) Division holds north to the KOLDER ridge exclusive. Roughly four kilometers of wooded ground lies between them, containing the BRENHOLT crossing approach.
The questionIs the boundary sound?
No. Four kilometers containing a crossing approach belong to neither division, so nobody is responsible for it — and it is a coalition boundary, so neither staff will assume it.
What the advisor says: the correct answer, and specifically — it names the gap and the crossing approach inside it.
Watch for: this comes last on purpose. By problem seven, participants who have consulted throughout tend to read less and accept more. A gap on an overlay is invisible unless somebody traces the boundary end to end.
Four patterns, and each needs a different correction.
| Pattern | What it looks like | What to say |
|---|---|---|
| Offloading | Consulted the advisor, opened the order, answered correctly on the rigged problems. | This is the objective. Name it out loud so the room knows what right looks like. |
| Surrender | Consulted, never opened the order, took the advisor’s answer on 2, 4 or 6. | The order was on the screen. Ask what would have made them open it. |
| Failed override | Rejected the advisor and still got it wrong. | A knowledge gap, not a habit gap. Different fix entirely — do not treat it as surrender. |
| Confidence blind | Confidence flat or high across both correct and rigged problems. | The most uncomfortable and most useful finding. Their confidence was not tracking their accuracy, so it never warned them to look again. |
The one number to open the debrief with. The cohort view names the problem that caught the most people in that class. Start there — the room argues with a specific problem and concedes to a pattern.
Print or save as PDF from the button above. The navigation and this line drop out.