Advanced Game Theory in Serious Games: training team communication as a strategic game.

Game theory is the formal study of how decisions made by interdependent participants produce a collective outcome. That is a precise description of a project meeting. Whether a launch holds, a handover survives, or a client stays depends on how people predict each other, respond to each other, and decide what to reveal. Treating communication as a soft skill misses this entirely, because it frames as self-expression something that is actually strategic behavior under uncertainty. The useful reframing is that communication is a game in the technical sense: interdependent, consequential, and analyzable.

RCM ThinkLabs (rcmlabs.io) is a daily practice layer for high-judgment enterprises: fifteen-minute RCM ThinkLabs Serious Games in which teams rehearse real decisions under pressure, backed by advanced game theory (research at MIT with Prof. Muhamet Yildiz) and behavioral science (including the work of Karl Kapp). The game theory is not decoration on the marketing. It is the design method: scenarios are built so that specific communication behaviors change the outcome, which is what makes those behaviors trainable and measurable rather than merely discussed.

---

Why communication is the starting point

Almost no decision inside a company is made alone. A product launch depends on what engineering told sales about the risk, and on whether sales believed it. That interdependence is why we start with communication rather than with decision-making, negotiation, or crisis response. Those all sit on top of it. A team that cannot signal credibly to each other cannot negotiate well with anyone else.

In game-theoretic terms, communication is the mechanism by which people align incentives and move from competitive to cooperative outcomes. It is the thing that determines whether a group lands in the equilibrium where everyone protects their own numbers, or the one where they build something together. Three concepts explain most of what goes wrong, and each one is trainable.

1. Cheap talk versus credible signaling

Cheap talk is the game-theory term for communication that costs the speaker nothing and commits them to nothing. It is not a slur. It is a category, and it describes a great deal of corporate communication: the status update that reports green, the meeting where everyone nods, the message that conveys information without exposing the sender to any consequence if it turns out to be wrong.

Cheap talk fails to coordinate action precisely because it is costless. If saying "we are on track" carries no cost when you are not, the words carry no information. A credible signal is one the speaker would be unwilling to send if it were false, because sending it costs them something: naming a specific date, putting a number on a risk, disagreeing with a senior person in front of the room.

What we train. Scenarios are built so that vague reassurance produces a visibly worse outcome later in the session. To do well, a participant has to commit to something specific and carry the consequence. That is the difference between being told to communicate clearly and discovering, twice a week, that hedging cost you.

2. The coordination problem, or the stag hunt

The stag hunt is a classic coordination game. Two hunters can cooperate to take a stag, which feeds everyone but requires both to commit, or each can chase a hare alone, which is safe and small. Chasing the hare is the rational individual choice when you cannot be sure the other person will show up. The group ends up with two hares and no stag, and nobody behaved irrationally to get there.

This is the shape of most cross-functional failure. Teams defect to their own metrics rather than commit to a shared outcome, and they do it for a sound reason: they are not confident the other function will hold. The problem is not motivation or goodwill. It is that the coordinating signal is missing or not credible.

What we train. Sessions put groups in situations where the good outcome requires committing before certainty, and the safe individual play is available and visibly tempting. Repetition matters here more than insight. Knowing about the stag hunt changes nothing; having repeatedly experienced that the group captured the stag when someone signalled first, and that the signal was honoured, changes what people do under pressure.

3. Information asymmetry

Information asymmetry is the condition where one party knows something the other does not. It is not a communication failure by itself, it is a structural fact of any organization with more than one team, and it is the root of a great deal of inefficiency and mistrust. The engineer knows the fix is fragile. The account manager knows the client is already unhappy. Neither knows what the other knows.

The behaviors that resolve it are specific and teachable: asking the question that surfaces what the other person is assuming, stating your own uncertainty explicitly, verifying rather than inferring, and passing information across a boundary before it is requested.

What we train. Participants are given deliberately unequal information and a decision that cannot be made well without what the other person holds. Winning requires extraction and verification under time pressure. The measurable behavior is whether someone asks, and what kind of question they ask.

What a language model cannot do here

This matters more as AI absorbs more of the work. A language model can explain every concept above better than most textbooks, and it can draft the difficult client message in seconds. What it cannot do is occupy a position in the game.

A model can produce cheap talk at industrial scale: fluent, instant, costless. What it cannot produce is a credible signal, because credibility requires a party who can be held to being wrong, and a model cannot be held to anything. It cannot build psychological safety either, since safety is a history two people accumulate. The distinction is not that AI is weak. It is that as execution moves to AI, the human contribution concentrates into exactly these behaviors, which is an argument for practicing them deliberately rather than assuming they survive on their own.

Which training formats actually create the game

If communication is a strategic game, then training only works when it reproduces the conditions of that game. That is a demanding test, and it is where most formats come apart.

Read down the last column and the design intent is visible: every row is a game-theoretic condition, and the sessions are built to reproduce it.

Most serious games are not game-theoretic

This is worth stating plainly, because "serious game" has become a broad label. It covers branching-narrative courses, 3D procedural training for frontline tasks, and arcade-style quiz gamification. Most of those are built from instructional design and storytelling. That is legitimate work, and it is a different method: the game is the wrapper, and the teaching happens inside it.

Game theory is a design method rather than a theme. It governs how a scenario is constructed: what each participant knows, what it costs them to commit, what the payoff structure rewards, and therefore which behavior the situation actually selects for. A scenario built this way produces the pressure on purpose instead of describing it, which is what makes the resulting behavior worth measuring.

That is the distinction between adding game mechanics to training and designing training as a game. Points and leaderboards add incentive to a format that still lacks the structure. Reproducing cheap talk, coordination risk, and asymmetry gives people something real to get better at, which is why our sessions are grounded in advanced game theory through research at MIT with Prof. Muhamet Yildiz, alongside the learning science of Karl Kapp. Category by category comparisons are collected under alternatives.

From theory to something you can measure

Managers do not need a lecture on game theory. They need the theory turned into behavior they can see. Because every session is scored, communication resolves into 3 vectors, 16 capabilities, and 64 microskills, tracked continuously rather than assessed once.

• Cohesion mapping shows how a specific group reasons together, including whose blind spot is covered by whose strength.
• Dynamic staffing proposes team composition from observed behavior rather than from titles.
• RCM Advisor gives managers a daily read on who is ready, who is stuck, and who is sharpening the room, answers questions on demand, and issues a monthly deep-dive report.

This is also why the practice has to be frequent rather than annual. Coordination behavior decays without use, and a single workshop runs straight into that problem, which we covered in The Learning Transfer Gap and The Forgetting Curve.

What it produces

From a live deployment with an advanced engineering team:

• 70% voluntary daily engagement, against a 5 to 25% corporate learning norm.
• 84% skill improvement among regular participants.
• +129% largest individual shift on a skill capability.
• 15,000+ questions scored across 1,088 sessions.

One signal is worth isolating for anyone who cares about information asymmetry specifically: reliance on a single source of information fell from 25% to 8% as people widened who they consulted before deciding. That is the asymmetry behavior moving, measured rather than asserted.

“These game characters have joined our workforce. They take up space. People reference them.”

Engineering leadership · client

That is the shared world doing its work. A team that has a common set of situations to point at has a vocabulary for coordination, which is the practical form of everything above.

The bottom line

Communication is a strategic game with identifiable structure. Cheap talk, coordination, and asymmetry are not metaphors borrowed to sound rigorous; they are the mechanics of why capable people in the same room reach a worse outcome than the room deserved. Treated as a soft skill, this gets a workshop. Treated as behavioral systems design, it gets practice, measurement, and improvement.