The ambiguity problem is an obstacle — one of many — to the effective pursuit of truth.
Humans often disagree. Sometimes the disagreers are people who want to agree, and try to agree, but they fail. This happens a lot. John Nerst joked that 82% of the internet is people arguing, and the rest is cats and porn.
Sometimes when people disagree they’re not really trying to agree. Sometimes they’re trying to win, or to be right. They might actually be averse (consciously or not) to changing their mind.
But sometimes people — or at least parts of them — are genuinely trying to resolve a disagreement. And they just simply fail to do it.
And sometimes these people are skilled truth-pursuers. Intellectuals. Scientists. Philosophers. Lovers of truth. Sometimes they’re friends and comrades who genuinely respect one another. And they simply fail to agree.
What’s going on in those cases?
I want to suggest that a large percentage of the time, the problem is semantic ambiguity. Natural language has a nice feature in that it lets us be vague and gestural. It lets us speak in metaphors and allusions. It lets us describe part of a thing while leaving other parts ambiguous. But sometimes we need more precision than that. Sometimes natural language is too vague — sometimes it’s critically vague. This can get us into unproductive disagreements. Sometimes — perhaps very often — two people who disagree, think that they are understanding one another when they are not. They think that one believes p and the other believes not-p, when really, one person is saying that they believe p1, and the other is saying that they believe not-p2, where p1 and p2 are orthogonal. But because they hear “p” and “not-p”, they think they’re understanding one another and having a debate, when really they’re just talking past one another.
I want to give a provisional name to this phenomenon. Let’s say that the ambiguity problem is the propensity for natural language to lead to unproductive disagreement rooted in misunderstanding about the meanings of terms.
David Chalmers has called disagreements of this sort “merely verbal” disputes. In the linked paper he offers a technique for resolving them: the conversants should temporarily taboo the ambiguous term and re-phrase it in less ambiguous terms.
I have a hunch that a more powerful solution is possible. For people who are genuinely interested in pursuing truth and who are hindered by this ambiguity problem, my provocation is that we can create a tool that:
- lets people create bespoke semantic objects disambiguated to the appropriate degree for their desired use case, and
- reference these objects unambiguously.
My hunch is that such a tool has the potential to take motivated people a long way towards dissolving the ambiguity problem.