Published on June 11, 2025 4:42 AM GMT
This is a followup to this earlier post, which I realized today didn't quite cover everything I wanted to.
To recap, the AB problem (note: I don't know who coined this name, does anyone know the origin?) is a problem that arises in arguments that leads people mischaracterizing their opponent's position. If Yvonne argues for A and Xavier believes A=>B, Xavier might assert that Yvonne is arguing for B. But this is incorrect -- it's Xavier who believes A=>B; Yvonne may disagree. You'll see this mistake commonly on the internet.
In my earlier post I touched on the distinction of "X as policy" vs "X as phenomenon", and how failure to distinguish to between them causes confusion in arguments. However, there are other related distinctions, more closely related to the AB problem, that I think also cause confusion, and that I failed to explicate.
Note that some of these I might describe as you talking about your position, some I might describe as a person talking about their opponents position, but all of these are applicable to both cases!
X as ends (phenomenon) vs X as ends (policy) vs X as means vs X as beneficial effect
This is basically an elaboration of the "X as policy" vs "X as phenomenon" distinction I described in my previous post.
When arguing for a policy, people will typically invoke what they consider to be beneficial effects of that policy. And, when arguing for a desired beneficial effect, people may argue for a policy that they believe will yield it. As such, the policy and its effect may get conflated into a single position -- even by those arguing for it, for convenience. If circumstances do not change, this may be harmless. However, if circumstances change, the link between the policy and the effect may be severed, causing a split.
By "X as ends (phenomenon)", I mean essentially the same thing as "X as phenomenon", saying that X is good, independent of means of achieving it (assuming they don't have massive downsides, etc). By "X as ends (policy)", I mean, "X is my preferred policy, and I think it's fundamentally right in the sense that even if circumstances were to change significantly (so that it no longer had the same effects) but not completely unreasonably, it would still be correct". By "X as means", I mean, "X is my preferred policy right now, but if circumstances were to change, it would no longer achieve my ends, and I would prefer a different policy". And by "X as beneficial effect", I mean, "X is a good thing that would be a consequence of what I think we should do, but it's not actually fundamentally the thing to focus on".
Why do I say all this? Because people will just say "I support X" or "that person supports X" and mean any one of these, and often not even be clear on which. This is related to the AB problem because the conflation is due to an assumed A=>B lurking in the background. Better to make it explicit and be clear whether it's A or B that you fundamentally favor! Or in other words:
- "B as ends (phenomenon)" is when you fundamentally favor B, and you say you favor B"A as ends (policy)" is when you fundmentally favor A, and you say you favor A"A as means" is when you fundamentally favor B, but you say you favor A"B as beneficial effect" is when you fundamentally favor A, but you say you favor B
Of course this relates to "X as policy" vs "X as phenomenon" in that, as I described it in my last post, A is a policy and B is a phenomenon. But I think this extra distinction is helpful. I think it is most natural to speak in terms of ends, what one fundamentally favors, but as always, people often won't be clear unless you make them clarify.
X in this world vs X in any possible world (X as ends)
I think this is one that's more of one that comes up with people talking about their opponent's positions, but it's not exclusive to that case.
In my original post I talked about policy vs phenomenon, above I added a dimension, expanding it to 4 cases. Let's now collapse it back down to only that dimension.
The question then is, if I hear someone say "Yvonne supports X", and supposing that the statement is in fact true in some sense and not purely an AB error, the question becomes, does Yvonne support X in this world, or in all possible worlds? (Maybe not all possible worlds, that's a bit extreme, but something like that. X as ends.) That is to say, is X actually the thing Yvonne fundamentally supports? Or is X merely a consequence of what Yvonne supports, even if Yvonne would indeed agree it is a consequence?
(I suppose here I'm excluding the "means" case, where X is actually a particular policy to achieve what Yvonne supports, but such cases are in reality probably an AB error. Still, you can lump it in in the case where it's not.)
You can guess here what I would suggest -- namely, talking about the things people fundamentally support. As always, other people will be unclear, and you will have to make them clarify. Doing things the other way can get pretty silly -- after all, every policy has downsides, so you can always smear someone by saying "Yvonne supports [downsides of Yvonne's preferred policy]". If they acknowledge that that downside is in fact a downside, but think it's worth it anyway; or if they consider it neutral; that's a misleading way to speak! If of course they consider it an upside it's less of a problem, but...
Speaking of neutrality, let me mention one particular variant of this...
Neutral on X vs opposed to X
One particular variant that comes up a bunch is people saying that someone is "anti-X" when they're actually just neutral on X. Thing is, being neutral on X in terms of your fundamental preferences, tends to lead to a reduction in X in terms of effect, because in a world of limited resources, anything not optimized for is optimized away. Still, there's a real difference between merely optimizing something away, and truly optimizing against it.
Discuss
