Thinking About the Global Maximum in a World of Local Maxima

On Facebook I read a post called “The Two Faces of the Dollar” by Simon Kim (김서준), CEO of Hashed, the Seoul-based crypto venture fund. Programmable money, the stablecoin wars, the collision between DeFi (decentralized finance) and traditional finance. The analysis is sharp and the metaphors are vivid. One observation in particular stayed with me.

Aave accepting JPMorgan dollars as collateral is not crypto surrendering. It just wants to raise its TVL (total value locked) and bring in institutional liquidity. JPMorgan putting dollars into DeFi is not a scheme to take over crypto. It is just a new revenue channel. But as each follows its own interest, most of DeFi’s collateral ends up filled with assets issued by Wall Street.

Nobody meant harm. Each actor simply pursued its own interest, rationally. Each climbed its own local maximum. And yet the sum converged on a structure nobody wanted. To borrow the post’s image, like a praying mantis infected by a horsehair worm, the parasite that steers its host into the water. Each actor believes it is choosing of its own free will, and ends up moving in a direction that benefits someone else.

Which raises a question. In a world of actors who pursue only local maxima, is thinking about the global maximum a futile exercise?

This is not a question of futility. It is a question of role.

In a world of actors who pursue only local maxima, thinking about the global maximum is not how that world normally behaves. So it always remains a minority role. But that does not make it optional. When the role is left empty, the system automatically converges on the simplest function: short-term profit, the logic of scale, the concentration of power. This is not a moral matter. It is where optimization leads.

Here is the important point. If you try to pursue the global maximum directly, you fail. Most actors do not have the global maximum in their objective function. For them it is incomputable, or an external variable that only adds cost. So persuasion, enlightenment, and moral appeals do not work well.

The key lies in the next step. Not forcing the direction, but designing the surface.

This is closer to the language of mathematics and engineering than of ethics. Each actor still climbs its own local maximum. But if the slope it climbs is a curved surface rather than a flat plane, then the steepest path upward leads, on its own, to a broader outcome. You are not giving the actors eyes. You are shaping terrain that brings them to a different landscape even without eyes.

This is also the core issue of the age of planting conditionals in money. The conditionals themselves are unavoidable. The question is who writes the conditions, and in which coordinate system. Right now that coordinate system is defined almost entirely in the language of local optimization: efficiency, risk management, regulatory compliance, yield. The global maximum does not appear in it. And what does not appear does not get optimized.

So this role is not the “powerless idealist.” It is the surface designer. Not a policymaker, not a moral teacher. Someone who designs incentive structures, protocols, governance, and default options.

This role is always seen late, and it does not win right away. But paradoxically, once a system crosses a certain scale, it cannot move to the next stage without this role, because the local optimizations start colliding with one another. The reason friction is appearing at the same time in stablecoins, DeFi, RWA (real-world assets brought on-chain), and oracles is that we have reached that point.

Pursuing the global maximum directly is futile. But designing the terrain so that the global maximum gets optimized indirectly is something someone absolutely has to do.

Original Korean version written on February 18, 2026. Translated into English with AI in October 2026.