The Decay Protocol

A busy marketplace attracts more buyers. More buyers attract more sellers, and each new seller gives people another reason to stay. That loop can make a service more useful. It can also leave a new competitor with little chance to catch up.

Preferential attachment describes one way this happens: new connections tend to form around the people or places that already have many. In economics, the shorthand r > g describes a different mechanism: returns on capital outpace overall economic growth. These ideas are not interchangeable, but each points to a question worth asking: when does an early advantage start deciding who gets the next one?

When an advantage compounds

Accumulated power can change the rules around it. A large platform can set the terms for reaching its users. A wealthy owner can absorb losses that would close a smaller business. A political institution with few checks can make decisions that strengthen its own position. These are different problems, but they share a risk: the people who benefit from the arrangement may also gain more influence over whether it changes.

AI could speed up some of these loops. A company with more users may have more data to improve its systems, and better systems may draw more users. The strength of that loop depends on what information a company collects and how it uses it. When software, capital, and access to a market reinforce one another, waiting for concentration to correct itself may not be enough.

In The Algorithmic Leviathan, I wrote about the authority people can hand to a system when its recommendations become difficult to question. Concentrated ownership creates a related problem: even when a system does not make the decision, its owner may have enough influence to set the terms.

Restraint at a human scale

Anthropologist Christopher Boehm used the phrase “reverse dominance hierarchy” to describe how a group can restrain someone who tries to dominate it. People can object, withhold cooperation, or push an overreaching member back toward shared expectations. That kind of accountability depends on people having ways to act together.

Those methods do not transfer neatly to a platform used by millions. Most participants do not know one another, and leaving may mean giving up access to customers, work, or a community. A rating system or a promise to “vote with your feet” cannot do much if the alternatives are weak. The question is how to make collective limits possible when the people affected are scattered and the owner controls the space where they meet.

Nature makes room through decay

In a forest, fallen leaves and dead wood are broken down by fungi, insects, and microbes. Their nutrients return to the soil, and dead wood provides habitat for other organisms. Ecologists describe dead wood as part of the structure and functioning of forest ecosystems. Decay makes material available for new growth and changes the conditions in which life can take hold.

That is not the same as steady improvement. A change can create room for one species and make life harder for another. Nature offers a cycle of turnover and reuse, not a promise that every loss leads to something better. The useful comparison for institutions is that renewal needs room: privileges can expire, rules can be revisited, and decision-making power can move before it hardens into permanent control.

Decay already limits some kinds of abuse

Software uses time-based limits when an unlimited resource would invite abuse. A leaky-bucket rate limiter drains a fixed allowance at a steady rate and rejects requests that arrive faster than it can drain. That makes it harder to overwhelm a service with a sudden burst of requests. It does not stop someone from spreading requests across many accounts or staying just below the limit; it puts a bound on one channel of abuse.

Scoring systems can also reduce the weight of old activity over time. For example, a reputation score could give recent behavior more weight than ratings from years ago, so an early burst of favorable ratings does not count forever at full strength. This creates a different risk: an inactive but trustworthy participant may lose standing, while a determined attacker may keep generating new activity. The decay rate and a way to contest the score matter as much as the formula.

These examples do not show that decay makes a system un-gameable. They show a narrower principle: when influence or access accumulates, reducing the lasting effect of old inputs can limit how much a single burst or early advantage controls the future. The mechanism works best when the quantity being limited is clear and the people affected can see and challenge the rule.

What should decay?

One answer might be to make certain advantages temporary. A privilege could expire unless it is renewed; decision-making power could be shared when it crosses a threshold; a company could face stronger duties as its control over a market grows. Each approach tries to make concentrated power answerable to others.

But automatic limits come with their own dangers. A rule needs to measure power, and any measure can be gamed or miss what matters. A cap that blocks a growing company may protect its established rivals. A decision to redistribute control can put too much authority in the hands of whoever writes the rule. Even a well-meant system can punish success without making the underlying market fairer.

People need room to build something useful and benefit from the work. The task is to stop one success from becoming permanent control over everyone else's choices. That means deciding what kinds of power should fade, who gets to set the limits, and how people can challenge a bad decision.

A “decay protocol” could give institutions a way to notice when advantage is compounding and reconsider the rules before people lose any meaningful say. The first design question is how people subject to concentrated power can recognize when their influence is shrinking and take part in changing the rules.