Chapter 1 diagnosed the short light cone — how far into the future a decision's consequences can still reach the person who made it: leaders capture upside now and diffuse downside onto their successors. You do not fix that by exhorting better character. You fix it by rewiring when the money arrives.
Start with the unglamorous part: leaders draw a solid base wage, independent of performance — civic duty, properly funded. This is not generosity; it is anti-corruption engineering with a long evidence trail. Underpay the people who hold discretionary power and they will price their influence on the open market — the pattern is visible everywhere from underpaid police forces, Russia's being the canonical study in how a state manufactures its own corruption, to the quiet consultancy-and-favors economy of underfunded legislatures. Post-Soviet Russia ran the full experiment: a state that stopped paying its enforcers watched them privatize enforcement, insiders strip-mined the commons through loans-for-shares (the 1995 scheme that signed the state's oil and metals giants over to a handful of bankers for token loans), and deaths from external causes — homicide, suicide, accidents, poisoning — nearly doubled across the first half of the decade — the cost of unpaid discretion, denominated in lives. Northern Europe's clean civil services run the experiment in the other direction — and Singapore made the bet explicit: ministerial pay competitive with the private market, corruption prosecuted without fear or favor, and within a generation one of the cleanest governments on earth. The point of paying well is selection as much as protection: the system should be able to recruit the most capable, most aligned people it can find away from the private market — and keep them aligned once recruited. Pay enough to take money off the table; make the honest career the comfortable one. The wage buys integrity's floor. It deliberately buys no more than that.
The performance upside is where the light cone gets stretched. A leader whose proposal passes earns an execution payoff — and then persistence pay: continued quarterly payments for as long as the policy stands. The trigger is deliberately simple — did this get repealed? A policy that survives its reconfirmations keeps paying; the moment the public repeals it or declines to reconfirm it — and chapter 8 makes that cheap to do — the stream ends. There is no measured formula between the people and the payout, no index to argue with, nothing for clever counsel to litigate: the review vote is the condition. And because every policy faces scheduled reviews and is challengeable between them, survival is never the inertial default it is today — it is a standing verdict, repeatedly renewed. Pay accumulates the way a builder's rents do: each durable thing put in place and kept in place adds a stream, and a career becomes a portfolio of standing law — which is exactly the fortune this design wants its most capable people chasing. Durable, effective, widely accepted proposals pay dividends; ineffective or harmful ones get repealed and stop paying. Leaders profit from creating policies that continue to work — the exact inverse of the current payoff structure.
Three consequences fall out. Because the public verdict controls the dividend (chapter 8), a leader's only way to defend their income is public persuasion that the law still deserves to stand — so someone is always paid to keep making the case for every standing law, and nothing survives purely by being forgotten. Repealing your own obsolete law becomes rational, since a policy the public has soured on was paying zero anyway. And the payoff is symmetric with history's judgment: if your action — a proposal, a repeal — is later reversed by the community, your future payments stop. You don't get paid for being on the wrong side of governance history. Past payments are never clawed back; experimentation must stay safe to attempt.
Policies do not act instantly, and a review that arrives before effects do measures noise. So every proposal declares its own maturation window — the settling period before its first review — and delegates sanity-check the claim at passage. The incentive design does the rest: the maturation window is unpaid. Claim too short a horizon and your law hits review before the effects arrive, and fails. Claim too long and you have bought yourself dead, unpaid years. The only profit-maximizing declaration is an honest estimate of the policy's real complexity horizon. Where a leader expects a J-curve — pain before payoff — the staked-claim instrument from chapter 3 applies: declare the pain window and the vindication test up front, and put standing behind it.
And why no performance metrics, no KPI bonuses, nothing "objective" in the pay formula? Chapter 3 already answered: any named metric that binds compensation is a Goodhart target: a measure stops measuring once someone is paid to move it. The broad did-this-work signal is the only input that cannot be gamed at the measurement layer — because it is the terminal value itself — the thing actually wanted, not a proxy standing in for it.