Formal agreements need to be unambiguous about the main features of the promises being made: dates, fees, allocation of risk, etc. To be able to rely on the promises, there needs to be certainty about what the promises mean.
But ambiguity also has its place. Sometimes the parties to an agreement disagree on peripheral issues, edge cases, speculative hypotheticals, and leaving ambiguity on those questions allows the parties to secure the promises they care most about.
This is an example of efficient ambiguity. Leaving these questions unresolved leaves some risk, and risk is costly – sometimes these unlikely questions will come up and create significant expense. But most of the time, they won’t. Much more often, the parties get the benefit of the promises they care about, and avoid the cost of the disagreements they leave unresolved, so that on net they are much better off to accept a certain amount of ambiguity – the perfect is the enemy of the good.
Efficient ambiguity applies much more broadly. Consider the common discouragement of discussing politics and religion. In much of life, a person’s political or religious affiliation is not directly relevant. You can a perfectly cordial relationship with your neighbors, your co-workers, your children’s friends’ parents. But many political and religious systems require that their members shun non-members, or try to convert them. Even absent those explicit requirements, many people just feel strongly enough about those subjects that they can’t let it go when they know someone disagrees with them.
A norm of generally not discussing those topics allows for the part of life where they have no direct relevance to happen where it could not if political and religious differences were made salient. Those differences are left ambiguous, and that helps us get other things done.
Efficient ambiguity becomes harder to achieve when the information ecosystem is such that most people have a public online presence from which their positions on many topics can be readily determined. It’s harder to remain willfully ignorant of topics like politics and religion.
What’s more, social connections are also often public, whether that’s direct connections on social media, or appearing together in images, conversations, etc. So it’s not only relatively easy to find out what an individual believes, but what their friends believe too.
This has a dual effect on efficient ambiguity. First, it directly resolves ambiguity, by establishing individuals’ positions on many more issues, so that they are known and discoverable by potential partners even when they would not be central to a specific partnership (this is one effect of context collapse).
Second, it increases the importance of the kinds of peripheral issues that efficient ambiguity usually serves to exclude. Even where an issue is not directly relevant, the fact that my cooperation with someone else will be publicly available means that their positions may be attributed to me and could influence future potential partners, meaning that those issues become indirectly relevant always and everywhere.
This is probably part of what’s driving political polarization. Cooperation across party lines becomes much more difficult as every position of every member of a coalition becomes permanently recorded and easily discoverable, and the lack of ambiguity increases the salience of every partner’s position on every issue in negotiations around every other potential partnership. This would encourage strict alignment withing coalitions and discourage cooperation outside of those strictly aligned coalitions.
A final thought, oddly optimistic, is that the advent of AI misinformation may solve this problem somewhat, by making all claims of fact much more ambiguous (this introduced other problems, of course). The prevalence of convincing lies makes willful ignorance much easier, which would make certain forms of cooperation easier. The post-truth era might see a revival of efficient ambiguity.