Four Types of Disagreement
post by silentbob · 2025-04-13T11:22:38.466Z · LW · GW · 2 commentsContents
Facts How to resolve factual disagreements? Values How to resolve values disagreements? Strategy How to resolve strategic disagreements? Labels How to resolve label disagreements? Why This Matters None 2 comments
Epistemic status: a model I find helpful to make sense of disagreements and, sometimes, resolve them.
I like to categorize disagreements using four buckets:
They don’t represent a perfect partitioning of “disagreement space”, meaning there is some overlap between them and they may not capture all possible disagreements, but they tend to get me pretty far in making sense of debates, in particular when and why they fail. In this post I’ll outline these four categories and provide some examples.
I also make the case that labels disagreements are the worst and in most cases can either be dropped entirely, or otherwise should be redirected into one of the other categories.
Facts
These are the most typical disagreements, and are probably what most people think disagreements are about most of the time, even when it’s actually closer to a different category. Factual disagreements are those that revolve around empirical questions that have some truth value and could, at least in principle, be answered decisively.
I’d personally count disagreements about what happens in the future as factual disagreements as well, even though these technically don’t have a fixed truth value yet.[1]
Example: Alice and Bob discuss vegetarianism. Alice is a convinced vegetarian. Bob isn’t, as he believes that animals aren’t sentient. So Alice and Bob have a factual disagreement about the state of the world (albeit one that can be very hard to resolve, as sentience is, of course, a tricky subject).
How to resolve factual disagreements?
- Research the answer
- Design an experiment
- Quantify your belief and state what your conviction depends on
Values
Values disagreements occur when people have different core beliefs about what's important, worthwhile, or morally relevant. These aren't empirical questions with truth values in the same way facts are – they're fundamentally about subjective preferences and priorities[2].
I find that values disagreements are often disguised as factual disagreements, because it feels more comfortable to argue about something "objective" than to admit we just care about different things and that the other party may even – on some level – be justified in their viewpoint. So we sometimes sneakily take our values as generally accepted, possibly without even noticing this step.
Example: Continuing the vegetarianism discussion, let's say Bob actually agrees that animals are sentient but believes human convenience morally outweighs animal suffering. Now Alice and Bob don't disagree about facts but about values – specifically how much moral weight to assign to the preferences of animals versus humans.
How to resolve values disagreements?
- Recognize that different values can be valid even if you don't share them
- Look for higher-order values you might share
- Not all values are “axiomatic”, some may depend on facts, which once again allows us to move things into the territory of empirical questions
Strategy
Strategic disagreements occur when people differ on the best approach to achieve a desired outcome. These are disagreements about means rather than ends.
When disagreements are strategic, that’s generally a good thing, as it implies there’s some common ground as you agree about the desired outcome. However, people sometimes have strong opinions on their strategic preferences, failing to see that they are discussing inherently empirical questions.
One can argue that strategy disagreements are pretty much the same as facts disagreements. I think that’s a fair point and would respond that it ultimately comes down to which categorization is most useful in practice. As strategy disagreements can be addressed differently than general values disagreements, I tend to differentiate between them. Furthermore, while many factual questions refer to the state of the world in some form, strategy questions are often more complex and depend on many factors, such as how and where the strategy is deployed, which can make a huge difference.
Example: Alice and Bob both agree animals are sentient (fact) and that reducing animal suffering matters morally (value). However, Bob thinks ethical consumption only has a tiny effect and factory farming should rather be addressed from supply side.
How to resolve strategic disagreements?
- Make sure you really have similar outcomes in mind and only disagree about how to get there
- Examine past cases and available evidence
- Make small experiments to test different strategies
- Sometimes, it may be viable to just run different strategies in parallel rather than necessarily identifying the better one
Labels
Label disagreements are about how we name or categorize things, and they're often the most frustrating yet least substantive type (which I admit makes this entire post slightly ironic, as it’s fundamentally about what names to attach to certain disagreements). These disagreements occur when people are using similar terms with implicitly very different semantics. They’re often not actionable and have little relevance for what’s actually going on in reality [LW · GW].
What makes label disagreements particularly insidious is that they often appear very fundamental and important. Words carry emotional and social baggage that can make conversations heated even when there's no practical difference in positions. They additionally are enticing because they allow people to distinguish themselves through strong and smart-sounding opinions that can’t be falsified. In a sense, I find this category to be the most important one to recognize as disagreements of this nature are both fairly common and can waste a lot of time.
Naturally, exceptions apply – some labels disagreements do hint at substantive questions underneath the surface. In these cases it’s important to unravel them and figure out what the actual crux is. When discussing “is a fetus a person”, then this is most likely not just about semantics but is rather a wrapper for some fundamental values question with impact on real world decisions.
Example: Alice and Bob both care deeply about animals and want to do their best to help and protect them. However, they end up in a heated debate about whether ordering the plant-based options at a big fast food chain is “vegan”. Alice claims that giving money to an exploitative capitalist company that makes most of its profits from meat products goes against everything veganism stands for, and hence cannot be vegan. Bob claims that food made out of plants is vegan no matter who produces or sells it. They both find it very important to be right about this.
Notably, when reframing things as a values or strategy question, Alice and Bob might actually get somewhere in their debate. But focusing on whether something can rightfully be labeled as “vegan” most likely isn’t very worthwhile and just leaves everyone annoyed.
More examples:
- “LLMs don’t actually think”
- “Nuclear energy is not renewable”
- “Is a hot dog a sandwich?”
- “AI-generated images can never be art”
How to resolve label disagreements?
- Focus on how something impacts the actual, physical world and the beings within it
- Be willing to use your conversation partner's preferred terminology if it doesn't compromise clarity
- If necessary, taboo certain loaded words
- Try to find a way to redirect the disagreement into a facts, values or strategy disagreement and proceed from there
Why This Matters
Understanding which type of disagreement you're having is crucial because each type requires a different resolution approach. Trying to resolve a values disagreement with more facts is often futile; likewise, arguing about labels often leads nowhere.
Frequently disagreements are mistaken for factual ones even when they’re not. Figuring out where exactly the disagreement lies greatly improves the odds of having a productive conversation and avoids wasting time by talking past each other.
The next time you find yourself in a heated disagreement, try asking: "Is this about facts, values, strategy, or labels?" Just identifying the category can lower the temperature and open pathways to resolution – or at least help you recognize when true resolution isn't possible.
Sometimes the most productive outcome isn't agreement but clear understanding of what you disagree on. Facts can be researched, strategies can be tested, and labels can be negotiated – but values differences may simply need to be acknowledged and respected.
- ^
Unless, of course, if you're into determinism. Or, if you believe in the multiverse, then maybe they have truth values already, but they are rather decimal numbers or distributions indicating the ratio of branches where the property in question will have a certain outcome.
- ^
Naturally, moral realists might see pretty much all values disagreements as facts disagreements.
2 comments
Comments sorted by top scores.
comment by romeostevensit · 2025-04-14T05:08:46.563Z · LW(p) · GW(p)
Type errors:
Map-territory confusion (labels facts)
Is-ought confusion (fact value)
Means-ends confusion (value strategy)
Implementation-classification confusion (strategy label) eg "if you classify this as an emergency that must mean you support taking immediate action"
Semantic-normative confusion (label value) eg "if you classify this as art you must think it is valuable"
Empirical-procedural confusion (fact strategy) eg "recidivism rates are highest among those without stable employment, therefore job training programs are the most important intervention"
Replies from: danielechlin↑ comment by danielechlin · 2025-04-14T15:54:17.331Z · LW(p) · GW(p)
Some of these follow from the "central fallacy," e.g. just because penguins are birds doesn't mean they're typical birds, which typically can fly. I nicknamed this "semantic bounty" in a short post -- if you spend 45 minutes convincing somebody something is X, e.g. X = discriminatory because X is probably gonna be something values-infused rather than feel like an arbitrary label, you're more likely to win the argument that something is technically X and therefore doesn't get a whole lot of properties of X, when you were hoping you get all the properties of X as a bounty for your opponent conceding the is-ness.