Urgent Coordination Problems Require Narrow Mandates

...

I

In Scott Alexander’s essay Archipelago and Atomic Communitarianism, he poses a thought experiment about a world made up of small diverse nations, that allows individuals to seek out the society which most closely matches their belief system. To make it compelling, the thought experiment includes a global institution called UniGov that can infringe the sovereignty of the nations. But UniGov has a very narrow mandate: ensure that all societies respect the Right of Exit. This is the core protection that allows the nations to experiment with different systems of government while preventing terrible human rights abuses. Without exit rights, we open the door to slavery, oppression of minority groups, forced collectivization, and all the other exhibits in the Museum of Human Cruelty. So in the world of Archipelago, protecting Exit Rights is an Urgent Coordination Problem, which is solved by UniGov and its narrow mandate.

Of the many agencies related to the United Nations, one of the most justifiable is the International Atomic Energy Agency. The primary mandate of the IAEA is to prevent the proliferation of nuclear weapons, while also helping countries develop nuclear power for peaceful purposes. Relative to smaller nations, the IAEA is in some sense supersovereign, since it can force them to accept rules and regulations related to nuclear power (more powerful nations can ignore the agency). So far, it has been decently successful, and that success is directly linked to its narrow mandate. If feminists or gay rights activists tried to use the IAEA to push their agenda in places like Iran or Kazakhstan, its credibility would collapse, leading not to a feminist Iran but a nuclear Iran. Again, an Urgent Coordination Problem (preventing nuclear proliferation) is (partly) solved by a global agency with a narrow mandate.

Famed rationalist author Eliezer Yudkowsky recently wrote a book with the dramatic title “If Anyone Builds It, Every Dies”. Yudkowsky wishes to impose a global law restricting AI research, and his organization (called MIRI) has a written a proposal for a global agency, similar to the IAEA, that would monitor adherence to the law. Many more mainstream thinkers wish only to pause AI research until the safety problem is better understood and principled safeguards can be put in place.

One severe obstacle to AI-research restrictions is that it requires multinational coordination, since AI has potentially decisive geopolitical implications. Scott Alexander teamed up with several AI experts to write a narrative-driven forecast called AI 2027; a major “plot hook” in this story is accelerating competition between the US and China. In May 2025, the Ukrainian Ministry of Defense posted:

For the first time, Ukraine deploys AI-powered carrier drone for strike on enemy targets… We continue to develop Ukrainian innovations that are changing the rules of technological warfare.

It would be foolish for liberal Western democracies to unilaterally restrict AI research, because of the threat that autocratic regimes will blaze forward and seize the advantage (this dynamic mirrors the situation in World War 2, where many scientists had strong objections to building nuclear weapons, but felt that the US must get the bomb before Germany did). So slowing AI research is an Urgent Coordination Problem, and MIRI’s proposal for a global monitoring organization may be necessary to solve it.

In 2026, it appears that the threat of climate change has been substantially mitigated by the development of cheap solar panels, but for many years this problem was taken seriously by many people. Billions of dollars of research money was directed towards this topic (scientists often half-joked that if you want to get your grant funded, you needed to add the phrase “climate change” to it somewhere). Achieving a global consensus on carbon reduction is difficult, because countries that refuse to participate will realize significant economic advantages. If only 80% of nations agree to restrict emissions, carbon-intensive industries will simply move to the remaining 20%. To some extent, this has already happened, with carbon-happy China growing far faster than carbon-reluctant Europe over the last 30 years. So reduction of carbon emission is an Urgent Coordination Problem.

II

In a previous post, I stated the following ethical principle:

Every human has the right to live in a society that reflects their own values.

This statement is immediately controversial, because if it is accepted, it necessitates substantial changes in the way our governments are organized. Many people, when first hearing this principle, reject it by citing one of the Urgent Coordination Problems mentioned above. We cannot build an Archipelago that satisfies this principle, because we must have some form of global government that controls nuclear proliferation, restricts AI development, and reduces carbon emissions. In a US context, people might argue that we need a powerful central government to oppose authoritarian countries such as Russia and China.

This objection is wrong for two reasons, one general and one specific. The first, general counterargument is that the impossibility of complete achievement of a philosophical principle never implies that we should not reach towards it. For example, the US does not fully protect Freedom of Speech: you cannot shout “fire” in a crowded room or publish detailed instructions about constructing a nuclear weapon. Most Americans agree with these restrictions, but still insist that Freedom of Speech must be protected in other areas. This is not a contradiction, it is an illustration that one can strongly support a principle while acknowledging it has limits. The general American support of capitalism is similar: most people believe in economic freedom, but are willing to accept limits on it, in areas like healthcare.

The second counterargument is more specific to the issue of centralization and global coordination. All of the problems mentioned above are urgent: if unaddressed, they pose severe threats to humanity. Because of that urgency, we contemplate the creation of some form of global coordination agency, even though this sacrifices personal and national sovereignty. The right response is:

Well, we need a powerful central government to handle all of these Urgent Coordination Problems. So we’d better give that government the narrowest possible mandate, to ensure it does not get distracted or disorganized by lesser priorities.

The mistake is to say:

Well, we need a powerful central government to handle all of these Urgent Coordination Problems. Since we’re going to centralize anyway, we might as well use the central government for lots of other priorities, like fighting racism and micromanaging the health care system.

In the case of the US, this mistake has certainly been made. Under the original design of the US federal government, the main priorities were shared military defense and economic integration. The 10th Amendment of the Bill of Rights stated very clearly that the powers of the DC government were limited to those enumerated by the Constitution. Unfortunately, this principle was trampled by Franklin Roosevelt, using the urgency created by the Great Depression as a pretext. So the federal government now includes a dizzying range of incompetently-run programs related to health, education, social welfare, and so on. DC is unable to manage these programs while also running the military and foreign policy, which is why US military adventures have been an unbroken string of costly failures for the last 70 years.

Conservative American skeptics of climate-change alarmism often cited the fact many of those pushing for carbon restrictions were also pushing for far wider restrictions on other parts of the economy. Former VP Al Gore, who won a Nobel Peace Prize for his climate activism, wrote on his web site that “The need for climate action is bound together with the struggle for racial equality and liberation.” This totally illogical connection severely damaged the credibility of the activists, and strengthened the suspicion that carbon reduction was a Trojan horse for far-reaching, UN-mandated control over society and the economy.

The moral of this story is that the more urgently you believe that coordination problems call for central-government institutions, the more serious you should be about narrowing the mandate of those organizations. Those of us who dislike centralization are extremely wary of ambitious politicians like Roosevelt who exploit crises to justify increasing their power. If you really want a global treaty to restrict carbon emissions, nuclear proliferation, or AI research, you’d better be willing to axe one of the other central agencies that plague our political existence, like the Social Security Administration. If you’re not willing to make that trade, then the Urgent Coordination Problem you’re invoking must not really be that urgent.