Comp Negotiation as an Information Entropy Problem
Most comp negotiations fail because both parties are operating with incomplete information. The candidate doesn’t know the company’s budget constraints or how desperate they are to fill the role. The company doesn’t know the candidate’s BATNA or what they actually value beyond salary.
This isn’t a conflict of interests problem. It’s an entropy problem.
In thermodynamics, entropy measures disorder—the amount of missing information in a system. A negotiation with high entropy is one where both parties are guessing. They don’t know what the other side needs, what they’ll accept, or what they’re willing to trade. The negotiation either collapses or settles at a suboptimal point because neither party has enough signal to find the actual zone of agreement.
I’m calling this the Information Entropy Model of negotiation. The goal isn’t to “win.” It’s to systematically reduce uncertainty until both parties can see the deal structure that maximizes joint value.
The Standard Advice Is Wrong
The standard advice on comp negotiation is tactical: anchor high, never give a number first, leverage competing offers. This treats negotiation as a zero-sum game where information is a weapon. That works if you’re negotiating a one-time transaction with someone you’ll never see again.
Most comp negotiations aren’t like that. You’re negotiating with someone who might become your manager, your report, or your co-founder. The relationship has value. The information you withhold now becomes friction later.
High-entropy negotiations produce brittle outcomes—agreements that break down the moment context shifts because they were never based on shared understanding.
Here’s the falsifiable claim: Comp negotiations fail in proportion to the amount of unresolved information asymmetry between parties. If you can reduce entropy faster than the other negotiator, you will consistently reach better outcomes—not because you’re a better bluffer, but because you’re operating with more complete information about the actual constraints and preferences in play.
The Entropy Reduction Strategy
Most negotiators do the opposite. They hoard information. They treat every question as a trap. They optimize for not revealing their position before the other side reveals theirs. This keeps entropy high. Both parties stay uncertain. The negotiation either stalls or closes on a narrow dimension (usually salary) while leaving value on the table across every other dimension.
The entropy reduction strategy inverts this. You actively gather information through inquiry. You reduce the other party’s uncertainty by making your constraints and preferences explicit. You map the full compensation space—not just salary, but equity, flexibility, title, scope, learning opportunities, team composition—and figure out where preferences are misaligned versus where they’re just unknown.
This creates three advantages.
You find non-obvious trades. The company might have flexibility on equity but not on salary. You might care more about remote work than about title. These trades are invisible in a high-entropy negotiation because neither party knows what the other values. Once you map preferences explicitly, you can construct deals that make both parties better off without either giving up something they actually care about.
You build trust faster. Sharing information is risky, but it’s also a signal. When you make your constraints visible—“I need X to make this work, but I’m flexible on Y”—you’re reducing the other party’s uncertainty about your intentions. If they reciprocate, entropy drops on both sides. The negotiation becomes collaborative instead of adversarial.
You avoid the anchoring trap. The “anchor high” strategy assumes the negotiation is one-dimensional. But if you’re negotiating on salary alone, you’re ignoring 80% of the compensation package. Anchoring works in high-entropy environments where neither party knows what else is on the table. In low-entropy environments, anchoring is just noise. The deal is determined by the actual constraints and the actual value creation, not by who said a number first.
The Math
In information theory, entropy is defined as H = -Σ p(x) log p(x), where p(x) is the probability distribution over possible states. In a negotiation, the “states” are the possible deal structures. High entropy means you have a wide, flat distribution—lots of possible deals, no clear signal about which one is likely. Low entropy means you’ve narrowed the distribution—you know which deals are feasible and which aren’t.
Every question you ask, every constraint you make explicit, every preference you reveal—these are entropy reduction moves. They collapse the probability distribution. The negotiation converges faster because both parties are working with the same information set.
What This Looks Like In Practice
I’ve trained 10,000+ PMs across India on negotiation. The consistent pattern: junior negotiators optimize for not losing. Senior negotiators optimize for information.
Here’s a concrete example from equity negotiation. A founder is negotiating with a potential technical co-founder. The founder thinks in terms of ownership percentage. The co-founder thinks in terms of outcome value. High entropy. They’re not even measuring the same thing.
The entropy reduction move: quantify the value contribution. If the co-founder joining improves the probability of a successful outcome by 10%, what’s that worth? If the company is currently valued at Rs 5 crore and has a 20% chance of reaching Rs 100 crore, the expected value is Rs 20 crore. A 10% improvement in success probability shifts expected value by Rs 10 crore. Now you have a shared frame. The conversation shifts from “what percentage do I deserve” to “what’s the marginal value of this hire.”
This doesn’t eliminate disagreement. But it eliminates ambiguity. Both parties are now negotiating over the same variable.
The Dimensions Most People Miss
Salary negotiations fail because people negotiate on one dimension. Here are the dimensions that actually matter in a comp package:
| Dimension | What It Actually Measures |
|---|---|
| Base salary | Guaranteed income, risk-free |
| Equity | Upside exposure, long-term alignment |
| Flexibility | Control over time and location |
| Scope | Learning rate, career trajectory |
| Team quality | Peer group, collaboration cost |
| Title | External signaling, internal authority |
Most candidates optimize for base salary because it’s the only number that’s certain. But if you’re joining a high-growth company, equity could be worth 5-10x your salary over four years. If you’re optimizing for learning, scope and team quality might matter more than either.
The entropy reduction move: make your preference weights explicit. “I care about equity 2x as much as salary. I care about remote flexibility 3x as much as title.” Now the company knows what levers to pull. If they can’t move on salary but can move on equity or flexibility, they have a path to yes.
What I Got Wrong
I used to think this strategy only worked with sophisticated negotiators. If the other party doesn’t understand information theory, they won’t reciprocate. They’ll just exploit your transparency.
That’s not what happens. Even unsophisticated negotiators respond to reduced uncertainty. When you make your constraints explicit, you’re lowering the cognitive load of the negotiation. The other party doesn’t have to guess what you want. They can focus on whether they can meet it. This speeds up the negotiation even if they don’t consciously understand why.
The failure mode isn’t exploitation. It’s mismatched time horizons. If the other party is optimizing for a one-time transaction, they won’t value the relationship benefits of low-entropy negotiation. But that’s a selection signal. If they won’t engage in information sharing, that tells you something about how they’ll operate after the deal closes.
The Civilizational Question
Here’s what I’m still working through: does this model hold when power asymmetry is extreme? In a negotiation between a candidate with no BATNA and a company with 500 applications for one role, does entropy reduction still work, or does it just expose the candidate’s weakness?
The standard answer is: don’t negotiate from weakness. But that’s not an answer. It’s an evasion. Most negotiations involve power asymmetry. The question is whether information sharing changes the outcome when power is unequal.
My hypothesis: it does, but not through fairness. It works because low-entropy negotiations are faster and cheaper. If you can show the company exactly what it would take to close you—and why that’s a good deal for them—you reduce their search cost. That has value even when they have other options.
But I don’t have enough data yet. The model works in peer negotiations and in negotiations where both parties have strong BATNAs. I’m less certain about its performance in asymmetric power situations.
More on this as I develop it.