Tragedy of the Commons

A commons was once a pasture that belonged jointly to an entire village. Every farmer was allowed to let his cows graze there. For the individual farmer, it pays off to put one more cow on the meadow: he alone reaps the benefit, while the damage from the shorter grass is shared by everyone. But if everyone calculates this way, the meadow is soon grazed bare and useless. This exact pattern is called the tragedy of the commons. What is tragic about it is that no one acts maliciously. Each person behaves reasonably when considered alone, and yet in the end everyone loses.

Why no one alone can save the meadow

The term is so well known because it makes visible a problem that is easily misread without it. When a resource is destroyed, people usually look for someone to blame. The tragedy of the commons shows that the problem lies in the structure, not in the character of those involved. Even a group made up entirely of decent people falls into the trap as long as the incentives are distributed this way.

On top of that: a single individual cannot save the situation by abstaining. A farmer who voluntarily keeps fewer cows loses income. The meadow gets grazed bare regardless, just by the others instead. So whoever holds back is punished twice over. This is exactly where many voluntary appeals fail, for instance when it comes to saving water during a drought.

Economist Elinor Ostrom received the Nobel Prize in Economics in 2009 for disproving this automatism. She studied real-world fisheries, forests, and irrigation systems. Her finding: communities are indeed capable of protecting their commons if they establish their own rules and monitor compliance themselves. The tragedy is thus a danger, not a law of nature.

The ingredients of the trap

For the problem to arise, two conditions must come together. First, the resource must be finite: it can become depleted or degraded. Second, no one may be excludable from it. There is no fence and no bill that keeps users away. Experts call such goods common-pool resources.

At the core lies a lopsided calculation. The benefit of one additional cow goes one hundred percent to its owner. The costs are spread across all the villagers—with twenty farmers, the one responsible bears only a twentieth of the cost. As long as this ratio remains this way, overuse is the individually smarter choice. Economists speak of external costs, that is, damages that one imposes on others.

Countermeasures therefore target this calculation. One can divide the resource into private ownership, so that everyone feels the consequences of their own actions directly. One can introduce fixed caps and penalties, as with catch quotas in fishing. Or one can make usage subject to a fee, for instance through a price on greenhouse gas emissions. One distinction is important: in the related free-rider problem, someone uses something without paying for it. In the case of the commons, usage additionally destroys the good itself.

From fishing grounds to AI training data

The best-known example today is the climate. The atmosphere belongs to no one, every country benefits from cheap energy, and the consequences of warming are borne by everyone together. Overfished oceans, falling groundwater levels, or overcrowded city centers work in a similar way. Antibiotics also belong in this category: the more often they are used, the faster resistant bacteria emerge, and their effectiveness is lost for everyone.

In the tech industry, the term now comes up regularly. The open internet was for a long time a shared pasture of knowledge: people wrote articles, answers, and guides, and others read them. AI systems are trained on exactly these texts. When users get their answers only from the chatbot, visits to the original sites decline. Their operators earn less and write less. This dries up the very source from which the models draw their knowledge.

A second example is web crawlers, i.e. programs that automatically retrieve web pages. For a single AI company, mass retrieval is cheap and useful. For small websites, it means high server costs. Many operators respond by blocking all automated access, including the harmless kind. In business news, you’ll also encounter this keyword in connection with licensing agreements between publishers and AI companies. Such agreements are an attempt to turn the commons into a paid good.