Two other things:
RP introduces the prior sentence (defining extreme internalism) with “perhaps”, so mb he’s hedging both sentences.
RP could have also said something like ‘And one of the most extremely externalist positions says that a belief is justified only in the case that it is true.’
If RP had chosen Y as an example, then it might be neutral.
But RP says Y is “the [single] most extremely externalist position”.
Thinking a bit more about it:
RP could have written “And the most extremely externalist positions say that whether a belief is justified depends only on external factors, and the most extreme of these positions depends on only one factor: whether the belief is true.”
I think that would have been okay. I think RP could have written that (like he has the capacity to).
This end of the spectrum also seems like it doesn’t have much to do with the stated goal/focus of the paper, so there’s not much value to RP’s argument by being specific about the definition.
I can see some ppl ignoring the X=Y thing, or doing some error-correction on RP’s behalf. Then, given that X/Y isn’t really relevant, I think ppl could see either of the two candidate reasons as neutral and not think it’s a problem.
What do you mean by “most extreme”? Do you mean further to one end of a spectrum? If so, which spectrum?
I think I meant that there would be a second spectrum, that was something like notability, or purity, or relevancy. Like that the “most extreme” position was an example of like the ~essence of extreme externalism. Given that the main spectrum is already ~percentage based, I think RP might have been okay with either mixing in more things, like notability, or putting those positions with fewer factors/conditions further towards the extremes.
I don’t think my alternate definition was okay in hindsight, but I do think it was an improvement over RP’s. Replacing “most extreme” with “most notable” would be a further improvement (and doesn’t require mixing incompatible units or including it in the existing spectrum). Maybe with “notable” (or something like it) it would be okay.
I think using “most extreme” was sloppy on my part, and it occurs to me that mb I was making the same mistake as RP: I put less effort into that bit because it “doesn’t have much to do with [Max’s] goal/focus of the [post], so there’s not much value to [Max’s] argument by being specific about the [alternate] definition.”
There’s some symmetry between your example and me using “extreme” in place of those other options, too. (Possibly also the same mistake as RP)
It’s problematic in general to try to merge two spectrums to a single overall conclusion. That’s actually the issue of combining factors from different dimensions into a single number Multi-Factor Decision Making Math
You can have two spectrums and keep them separate, but trying to call something the most extreme according to both spectrums in a mixed way is problematic.
Yeah, I’m aware of this and was aware of including the mistake (wrt MFDMM) while writing the alt definition. I didn’t fix it because I thought it was a mistake that RP would make. I shouldn’t have said I thought it was okay, though (too vague for what I was trying to do).
I think mb I was trying to do too many things at once: write something that RP could have plausibly said, expand on a definition about a topic I don’t know much about, and use that in a chain of reasoning.
Yeah I think he might have been mixing up more than one spectrum. My higher level conclusion was that he says the purpose of the article in the abstract, and wasn’t seriously trying to think or talk about this other stuff. These introductory comments are not the ideas in the article he was trying to enable engagement with, which is part of why they’re hard to engage with. Author Goals and Errors
One issue that raises is: Why does he have an introduction at all instead of only writing about the stuff he cares about?
some brainstorming:
(Note: before this I did skim the paper and ctrl+f for key words like ‘position’, ‘choose’, ‘edge’, ‘map’, ‘landscape’, ‘extreme’ and mb some others. There were not many hits and nothing substantial WRT the spectrum. There are 107 instances of the word “concept” though.)
‘normal’ reasons for having an intro:
other reasons:
I think the main theme in my other reasons list is: to frame the “disagreement … between internalists and externalists” such that RP can later come to a conclusion that is compatible with this new framing, but isn’t with the old framing (two camps). His conclusion looks more reasonable if he says all the various “concepts” are on a spectrum, and he can then claim that two concepts are close to one another even if they’re not related or they’re otherwise incompatible. (Being vague about the spectrum and defining extreme end-points also mb provides an illusion that ideas are “closer” than they’d otherwise appear.)
This quote seems relevant since it fits with some of the other reasons.
In sum: I don’t think there are good reasons to resist pluralism about justification. This, then, opens the door to a view on which there are legitimate internalist, externalist, and hybrid versions of the concept of justification, perhaps answering to different intuitions, perhaps playing different theoretical roles in our epistemology, perhaps serving different practical, moral, or political ends.
Also it’s immediately followed by: (emphasis mine)
Of course, you might say that what this really supports is not pluralism about justification, but nihilism.[1] In the end, I think these positions are reasonably close to one another. Of course, they seem to be as far apart as possible: […]
How can he say that two things are close together or far apart unless there’s some prior explanation of what ‘close’ and ‘far’ might mean? This fits with some of the other reasons too.
Note: WRT these two quotes, I’m conscious that I might be finding stuff that already supports reasons I brainstormed, rather than taking a more critical approach.
↩︎Thanks to an anonymous referee for pushing me to consider this possibility. [Note: this part of a quote from RP – footnote 1 in RP’s paper]
A major purpose of introductions is to begin with high level ideas that most people would be familiar with or understand, and then go from there to more specialized, detailed or low level ideas. You start at common ground and then tell people how to get from there to the stuff you’re going to write about. This connects your specifics to something of general interest.
This isn’t directly related to the main RP quote, but I found 6-7 more typos or editing errors (missing or extra words) and/or grammatical mistakes. They’re bolded in the quotes below.
2 errors (“a belief … might count it as rational” and “because formed”)
For those who count the boundedly rational among the rational, a belief formed on the basis of a base rate fallacy, for instance, might very well count it as rational because formed by a method that proves reliable in our evolutionary niche, but most will agree that it is not justified.
missing “it was” between “because” and “formed” (or something like that)
Is ‘a belief’ the subject of ‘might very well count it as rational’? (I think the intention is that “those who count the boundedly rational among the rational” are the ones who count the belief as rational.)
a simple fix would be to replace “count it” with “be counted”.
however, reorganizing things makes more sense and fixes the meaning:
(not a real quote, but made of phrases/clauses taken from RP’s quote above)
[Regarding] a belief formed on the basis of a base rate fallacy, for instance, those who count the boundedly rational among the rational might very well count [that belief] as rational because it was formed by a method that proves reliable in our evolutionary niche, but most will agree that it is not justified.
idk if the idea itself makes sense but at least this way is easier to read.
Apparently some people count beliefs based on fallacies as rational?
On the pluralist view, an internalist notion of justification need not serve a practical, moral, or political end at all in order to be included in our suite of justification concepts. And it certainly needn’t serve a more important such end than the end served by an externalist notion.
Feels wrong, mb there’s a dialect thing going on here? I think it can be fixed with word order alone. “… serve such an end more important than …”
Part of what is surprising about the lack of support for pluralism is that it seems the natural response when competent users of the concept—as I assume analytic epistemologists of the past fifty years to be—disagree so irreconcilably on certain of its basic features.
either “… on some of its basic features” or “… on certain basic features.”
The arguments he enumerates there can be divided into three categories based on the considerations they adduce in favour of the various accounts: in the first sort of argument, …
“adduce in favour of” – to adduce is to provide evidence/argument in support of something, so “in favour of” is redundant and superfluous, but changing it would require restructuring the sentence and the verb form. (or RP could have just replaced ‘adduce’ with ‘advance’, ‘provide’, ‘put forward’, etc)
So, for instance, a concept might help us politically if we value or disvalue for the same reasons everything that falls under it.
‘for the same reasons’ – should be parenthetical or comma separated. I’m not sure it’s ever grammatical without something kind of separation. I think it reads “… if we value or disvalue […] everything that falls under it”.
Or it might be because the nihilist thinks that the various concepts that the pluralist counts as different concepts of justification are not sufficiently closely related to count as different versions of the same concept.
alternatives that would work:
Sometimes free online versions of papers are not the final draft. They’re often pre-publication versions which are never updated. I think the final, published version is this: https://link.springer.com/content/pdf/10.1007/s11098-021-01660-x.pdf So you might want to see if that fixes some of the typos or not.
It’s a problem because a few fixes may be made before publication but then, even after publication, I think even most academics with journal access read the online version over the journal version.
Yeah gp. Looking at that version, all the bolded parts of what I quoted are still there (I didn’t check the full quotes).
I did compare the original PDF link against the html version before posting those errors, too.
Just now I used pdf2txt, wdiff (word diff), and colordiff (highlighting) to check for differences between the two PDFs. (I also removed newlines and replaced groups of spaces with a single space.) There were no significant differences, mostly just artifacts from PDF conversion or hyphenation over line-breaks or page numbers.
One thing that occurs to me more generally: if an author wanted to update the web copy of their article PDF, how difficult would that be? How difficult would it be for an author to be better than a journal when it comes to updates, retractions, etc? (Particularly if that author doesn’t have clout.)
I guess that the journals set the standard here, and it’s v hard for an author to exceed those standards. Would journals push back against an author that tried to include a link to a forum (e.g., at the end of an abstract) for discussion / criticism of the work? Do they have ways for authors to make minor corrections without going through a review process? What if the author wants to retract research proactively?
I guess that sites like arXiv would allow many of these sorts of things b/c it’s a preprint server rather than a journal, but I can’t imagine nature or science making it easy. So, depending on journal culture, the journals are like a limiting factor on academic culture. Academics who try to go against that waste their time at best and piss off journals at worst (which might really negatively impact their career).
Last year I told an author about a math error in their philosophy paper and they said:
Yes, whoops! Thanks for pointing that out! Unfortunately, it’s too late to make changes now.
Sad that the journals just won’t let them fix stuff… What a bad system that hasn’t been updated for internet publishing.
It was basically a typo, repeated in four places, that made the math wrong but didn’t affect the point of the paper. It does raise questions about the editing processes used, as do the typos/errors Max found. It was a simple arithmetic error that was visible just with loose estimation, and trivial to figure out with a calculator, so if readers/reviewers/etc were actually paying attention and trying to follow the details they should notice.
I noticed something while reading RP’s paper the other day – an explicitly anti-yes/no and anti-paths-forward bit. I think, more broadly, it’s anti-Popper, too (including being anti-CR/CF). I thought it was notable:
Upon reflection, it surprises me that so few epistemologists have heeded Alston’s call for pluralism about the concept of justification (Alston, 2005). On this view, there are a number of concepts that have equally good claim to be our concept of justification and there is no concept that has a better claim than these.
It’s notable because the claim “there are a number of concepts that have equally good claim to be our concept of justification” is explicitly something that yes/no and paths forward would help resolve. More broadly, CF/CR both reject the idea that we should accept multiple contradictory ideas simultaneously (an idea that RP seems to be adducing, or attempting to, at least).
Arguably, what RP said actually contradicts the main goal of his article: how to choose between those concepts. If the concepts all have an equally good claim, then there is no rational way to choose (since there’s no way to differentiate them). But, if some ideas worked in some contexts, and other ideas in other contexts, then the ~quality of each idea’s claim to be “our concept of justification” would be context dependent, and thus there would be some justification for choosing one over another for a given context.
IDK if RP tries to resolve this or even notices it (atm I don’t care to read enough to find out).
Since this is anti-CF I’ll cross-link Critical Fallibilism Real World Examples here.
Accuracy and the Laws of Credence, by Richard Pettigrew, page 2:
No Drop If an agent has opinion set {A, B} and A entails B, then rationality requires that c(A) ≤ c(B).
(The name “no drop” is because a credence shouldn’t drop over a logical entailment.)
Comments?
Working out that A entails B is an extra step so you could make a mistake. So why shouldn’t the credence of B be lower in his system?
Yeah, that’s the same main idea as what I was thinking.
Is “A entails B” a premise that you have 100% certainty of, or an idea that you have some credence for? It doesn’t clearly specify.
If it’s a premise that can be treated as a certainty, then this law can never be used to guide real life action, because you’re never presented with a decision-making situation where you have infallible certainty that A entails B.
It actually says his “opinion set” only contains A and B, which I think actually means the agent does not know that A entails B – and therefore should not take it into account when forming credences, even if it’s true. Or do agents have some knowledge set other than their opinion set?
I don’t think there are good Paths Forward for this kind of issue. Which makes it very hard to engage with the book because the book is a bunch of sub-ideas that are either premised on stuff like No Drop or else make the same kind of assumptions made when coming up with No Drop. The book is not about higher level issues like whether No Drop is correct in the first place and how it actually works.
IIRC the book does say something about there being multiple arguments for No Drop, so I think you could look up papers that argue for it. I’d expect them to be overly focused on math within their framework and given their premises, but some of them might specify what No Drop means in more detail.
Note: I wrote this on the weekend but am not that confident in it. I intended to review and re-edit it but I think that’s just been keeping me from posting it. I would change some stuff if I edited it now, but that’d partially be due to Alan and Elliot’s posts from a few hours ago, so I’ll leave it how it is.
Besides the whole credence function idea, No Drop seems reasonable. I can’t think of a counter-example. It reminds me of the conjunction fallacy. If the credence function is based on MFDMM and only outputs 0 and 1 then it seems to hold.
That said, it doesn’t really sit well for me. Not exactly sure why.
I did have to search for entail and entailment to figure out what those meant exactly. It means like “if A is true, then B is necessarily true”. I think something like this would be fine, too: if A entails B, then A&B is equivalent to A. (A&B meaning both A and B – in some cases this is equiv to boolean AND.) IDK if RP ever precisely defines it but it seems to be an established idea.
I read a bit of the source. RP also says that c maps to [0,1] which seems reasonable enough, but has some interesting consequences. (Also, IDK if c : \{\dots\} \to [0,1] is actually reasonable or not, why not c: \{\dots\} \to [-1,1] instead? RP says “convention” but that doesn’t seem like a good enough reason when dealing with foundational ideas.)
For the sake of discussion, RP’s A and B examples took the form:
A: Sonya is an X and a Y
B: Sonya is a Y
X and Y were chosen so that they’re independent / unrelated.
Say we assume that new supporting evidence for A always increases credence (and inversely: contradicting evidence decreases credence) – so if there’s new evidence that Sonya is an X then c(A) should increase, but c(B) should stay the same. In that case, it never makes sense to have c(B)=0 by default b/c you end up with a contradiction. This means that your degree of belief in anything and everything should always be >0. Especially if you have absolutely no evidence either way.
If you don’t want to have some degree of belief in everything by default, then you need to use the idea that supporting evidence might not increase credence, and you end up with something a bit like MFDMM. I suspect that maybe MFDMM is the only solution in this case (basically: fractional credences don’t make sense), but IDK how to prove that.
(You could also say that evidence supporting A increases the credence of A and B regardless of whether that evidence has anything to do with B, but that seems obviously broken, and getting into raven paradox territory)
I feel like there is maybe a way to force the credence function to output 0 for everything (like some construct based on an infinite number of mutually exclusive propositions), but IDK.
One thing that seems notable is that whether A entails B isn’t accounted for, it’s just taken as fact. This isn’t an issue for RP’s specific example, but I wonder how that sort of idea would be integrated in general.
Hmm, maybe this sort of thing is related to why it doesn’t sit right with me:
Using No Drop:
H: (No Drop) if A entails B, then rationality entails c(A) \le c(B)
I: rationality
G: c(A) \le c(B) (i.e., No Drop’s conclusion)
So, given H, c(I) \le c(G)?
This seems to be saying something like: we should be less confident in our methods of thinking than the ideas that our methods are based on. This doesn’t seem reasonable b/c it doesn’t take into account error checking (e.g., using redundant methods to check for the same answer) or finding a better competing theory that mostly overlaps.
“better competing theory” meaning one that mostly agrees but solves new cases too, like GR vs Newtonian Gravity.
I think there’s an interesting asymmetry WRT introductions:
Also, I have a guess about the problematic spectrum paragraph from Radical epistemology, structural explanations, and epistemic weaponry: I think it was added late in the editing process. It’s badly integrated (there’s very little mention of the idea after that paragraph), and it doesn’t really do much to serve the paper. But it would answer some possible review comments like needing more context about using different theories of justification like RP does. (Or help to, at least.)