JonathanLivengood

Karma: 524

JonathanLivengood 6 May 2013 4:49 UTC
5 points
0
on: Pascal’s Muggle: Infinitesimal Priors and Strong Evidence

If the AI actually ends up with strong evidence for a scenario it assigned super-exponential improbability, the AI reconsiders its priors and the apparent strength of evidence rather than executing a blind Bayesian update, though this part is formally a tad underspecified.

I would love to have a conversation about this. Is the “tad” here hyperbole or do you actually have something mostly worked out that you just don’t want to post? On a first reading (and admittedly without much serious thought—it’s been a long day), it seems to me that this is where the real heavy lifting has to be done. I’m always worried that I’m missing something, but I don’t see how to evaluate the proposal without knowing how the super-updates are carried out.

Really interesting, though.

JonathanLivengood 2 May 2013 6:50 UTC
2 points
0
in reply to: Rob Bensinger’s comment on: What do professional philosophers believe, and why?
Ah, I see that I misread. Somehow I had it in my head that you were talking about the question on the philpapers survey specifically about scientific realism. Probably because I’ve been teaching the realism debate in my philosophy of science course the last couple of weeks.

I am, however, going to disagree that I’ve given a too strong characterization of scientific realism. I did (stupidly and accidentally) drop the phrase ”… is true or approximately true” from the end of the second commitment, but with that in place, the scientific realist really is committed to our being able to uniquely determine by evidence which of several literal rivals we ought to believe to be true or approximately true. Weakening to “most cases” or “many cases” deflates scientific realism significantly. Even constructive empiricists are going to believe that many scientific theories are literally true, since many scientific theories do not say anything about unobservable entities.

Also, without the “in every case,” it is really hard to make sense of the concern realists have about under-determination. If realists thought that sometimes they wouldn’t have good reasons to believe some one theory to be true or approximately true, then they could reply to real-life under-determination arguments (as opposed to the toy examples sometimes offered) by saying, “Oh, this is an exceptional case.”

Anyway, the kinds of anti-realist who oppose scientific realism almost never deny that tables exist. (Though maybe they should for reasons coming out of material object metaphysics.)

JonathanLivengood 2 May 2013 6:07 UTC
3 points
0
in reply to: Desrtopa’s comment on: What do professional philosophers believe, and why?
We do pretty well, actually (pdf). (Though I think this is a selection effect, not a positive effect of training.)

JonathanLivengood 2 May 2013 5:59 UTC
3 points
0
in reply to: Rob Bensinger’s comment on: What do professional philosophers believe, and why?
I’m guessing that you don’t really know what anti-realism in philosophy of science looks like. I suspect that most of the non-specialist philosophers who responded also don’t really know what the issues are, so this is hardly a knock against you. Scientific realism sounds like it should be right. But the issue is more complicated, I think.

Scientific realists commit to at least the following two theses:

(1) Semantic Realism. Read scientific theories literally. If one theory says that space-time is curved and there are no forces, while the other says that space-time is flat and there are primitive forces (so the two have exactly the same observational consequences in all cases), then the realist says that at most one of the two is true.

(2) Epistemic Realism. In every case, observation and experimentation can provide us with good epistemic (as opposed to pragmatic) reasons to believe that what some single theory, read literally, says about the world.

Denying either of these leads to some form of anti-realism, broadly construed. Positivists, instrumentalists, and pragmatists deny (1), as Einstein seems to have done in at least two cases. Constructive empiricists deny (2) in order to keep a commitment to (1) while avoiding inflationary metaphysics. Structural realists deny one or both of these commitments, meaning that they are anti-realists in the sense of the question at stake.

JonathanLivengood 10 Apr 2013 19:38 UTC
0 points
0
in reply to: Manfred’s comment on: Meetup Interest Probe: Urbana-Champaign, Illinois
Ah, gotcha.

JonathanLivengood 10 Apr 2013 17:09 UTC
0 points
0
in reply to: Manfred’s comment on: Meetup Interest Probe: Urbana-Champaign, Illinois
Are the meetings word of mouth at this point, then? When is the next meeting planned?

JonathanLivengood 10 Apr 2013 15:46 UTC
0 points
0
on: Meetup Interest Probe: Urbana-Champaign, Illinois
I have had some interest, but I never managed to attend any of the previous meetups. I don’t know if I will find time for it in the future.

JonathanLivengood 6 Jan 2013 13:56 UTC
0 points
0
in reply to: IlyaShpitser’s comment on: A reply to Mark Linsenmayer about philosophy
That question raises a bunch of interpretive difficulties. You will find the expression sine qua non, which literally means “without which not,” in some medieval writings about causation. For example, Aquinas rejects mere sine qua non causality as an adequate account of how the sacraments effect grace. In legal contexts today, that same expression denotes a simple counterfactual test for causation—the “but for” test. One might try to interpret the phrase as meaning “indispensable” when Aquinas and other medievals use it and then deflate “indispensable” of its counterfactual content. However, if “indispensable” is supposed to lack counterfactual significance, then the same non-counterfactual reading could, I think, be taken with respect to that passage in Hume. I don’t know if the idea shows up earlier. I wouldn’t be surprised to find that it does.

JonathanLivengood 6 Jan 2013 10:36 UTC
8 points
0
in reply to: Peterdjones’s comment on: A reply to Mark Linsenmayer about philosophy

I’ll say it again: there is no point in criticising philosophy unless you have (1) a better way of (2) answering the same questions.

Criticism could come in the form of showing that the questions shouldn’t be asked for one reason or another. Or criticism could come in the form of showing that the questions cannot be answered with the available tools. For example, if I ran into a bunch of people trying to trisect an arbitrary angle using compass and straight-edge, I might show them that their tools are inadequate for the task. In principle, I could do that without having any replacement procedure. And yet, it seems that I have helped them out.

Such criticism would have at least the following point. If people are engaged in a practice that cannot accomplish what they aim to accomplish, then they are wasting resources. Getting them to redirect their energies to other projects—perhaps getting them to search for other ways to satisfy their original aims (ways that might possibly work) -- would put their resources to work.

JonathanLivengood 6 Jan 2013 10:18 UTC
1 point
0
in reply to: IlyaShpitser’s comment on: A reply to Mark Linsenmayer about philosophy
That is being generous to Hume, I think. The counterfactual account in Hume is an afterthought to the first of his two (incompatible) definitions of causation in the Enquiry:

Similar objects are always conjoined with similar. Of this we have experience. Suitably to this experience, therefore, we may define a cause to be an object, followed by another, and where all the objects similar to the first are followed by objects similar to the second. Or in other words where, if the first object had not been, the second never had existed.

As far as I know, this is the only place where Hume offers a counterfactual account of causation, and in doing so, he confuses a counterfactual account with a regularity account. Not promising. Many, many people have tried to find a coherent theory of causation in Hume’s writings: he’s a regularity theorist, he’s a projectivist, he’s a skeptical realist, he’s a counterfactual theorist, he’s an interventionist, he’s an inferentialist … or so various interpreters say. On and on. I think all these attempts at interpreting Hume have been failures. There is no Humean theory to find because Hume didn’t offer a coherent account of causation.

JonathanLivengood 2 Jan 2013 18:32 UTC
3 points
0
in reply to: Pablo’s comment on: What is the best paper explaining the superiority of Bayesianism over frequentism?
Wow! Thanks for the Good Thinking link. Now I won’t have to scan it myself.

JonathanLivengood 2 Jan 2013 18:30 UTC
2 points
0
in reply to: Axel’s comment on: What is the best paper explaining the superiority of Bayesianism over frequentism?
Yes, that’s the letter!

JonathanLivengood 2 Jan 2013 3:34 UTC
19 points
0
on: What is the best paper explaining the superiority of Bayesianism over frequentism?
It might help if you told us which of the thousands of varieties of Bayesianism you have in mind with your question. (I would link to I.J. Good’s letter on the 46656 Varieties of Bayesians, but the best I could come up with was the citation in Google Scholar, which does not make the actual text available.)

In terms of pure (or mostly pure) criticisms of frequentist interpretations of probability, you might look at two papers by Alan Hajek: fifteen arguments against finite frequentism and fifteen arguments against hypothetical frequentism.

In terms of Bayesian statistics, you might take a look at a couple of papers by Dennis Lindley: an older paper on The Present Position in Bayesian Statistics and a newer one on The Philosophy of Statistics.

Lindley gives a personalist Bayesian account. If you want “objective Bayes,” you might take a look at this paper by James Berger. (The link actually has a bunch of papers, some of them discussing Berger’s paper, which is the first in the set.)

You might also find Bradley Efron’s paper Why Isn’t Everyone a Bayesian? useful. And on that note, I’ll just say that the presupposition of your question (that Bayesianism is straightforwardly superior to frequentism in all or most all cases) is more fraught than you might think.

JonathanLivengood 30 Dec 2012 20:05 UTC
2 points
0
in reply to: Philipp Risius’s comment on: Brain-in-a-vat Trolley Question
Oh, the under-specification! ;)

JonathanLivengood 30 Dec 2012 7:04 UTC
13 points
0
on: Brain-in-a-vat Trolley Question
Seems to me that curing cancer swamps out everything else in the story. Supposing that World War 2 was entirely down to Hitler, the casualties came to about 60-80 million. By comparison, back of the envelope calculations suggest that around 1.7 million people die from cancer each year* in the U.S., E.U., and Japan taken together. See the CDC numbers and the Destatis numbers (via Google’s public data explorer) for the numbers that I used to form a baseline for the 1.7 million figure.

That means that within a generation or two, the cancer guy would have saved as many lives (to speak with the vulgar) as the Hitler guy would have killed. Plus, the cancer guy would have improved quality of life for a lot more people than that. Maybe we have to go another couple of generations to balance life years if the Hitler casualties are all young and the cancer savings are all old. But under the assumption that a solution to cancer is very unlikely without the cancer guy, the right decision seems clearly to be to steer the trolley left.

* Things get more complicated if we suppose that the Hitler guy will bring about a new world war and attempted genocide, which might involve full-on nuclear war, rather than a sort of repeat of real-Hitler’s consequences. I am choosing to understand the Hitler guy as being responsible for 60 or 80 million deaths—or make the number a bit larger, like 100 million, if you like.

JonathanLivengood 27 Dec 2012 21:01 UTC
1 point
0
in reply to: ddxxdd’s comment on: Morality Isn’t Logical
Interesting piece. I was a bit bemused by this, though:

In fact Plato wrote to Archimedes, scolding him about messing around with real levers and ropes when any gentleman would have stayed in his study or possibly, in Archimedes’ case, his bath.

Problematically for the story, Plato died around 347 BCE, and Archimedes wasn’t born until 287 BCE—sixty years later.

JonathanLivengood 26 Dec 2012 18:59 UTC
2 points
0
in reply to: BrassLion’s comment on: New censorship: against hypothetical violence against identifiable people
I’m not sure what you count as violence, but if you look at the history of the suffrage movement in Britain, you will find that while the movement started out as non-violent, it escalated to include campaigns of window-breaking, arson, and other destruction of property. (Women were also involved in many violent confrontations with police, but it looks like the police always initiated the violence. To what degree women responded in kind and whether that would make their movement violent is unclear to me.) The historians usually describe the vandalism campaigns as violent, militarism, or both, though maybe you meant to restrict attention to violence against persons. Of course, the women agitating for the vote suffered much more violence than they inflicted.

JonathanLivengood 26 Dec 2012 6:18 UTC
17 points
0
on: META: Deletion policy

When a certain episode of Pokemon contained contained a pattern of red and blue flashes capable of inducing epilepsy, 685 children were taken to hospitals, most of whom had seen the pattern not on the original Pokemon episode but on news reports showing the episode which had induced epilepsy.

At the very least, this needs a citation or two, since the following sources cast doubt on the story as presented:

WebMD’s account

CNN’s account

Snopes’ account

And CSI’s account, which includes the following:

At about 6:51, the flashing lights filled the screens. By 7:30, according to the Fire-Defense agency, 618 children had been taken to hospitals complaining of various symptoms.

News of the attacks shot through Japan, and it was the subject of media reports later that evening. During the coverage, several stations replayed the flashing sequence, whereupon even more children fell ill and sought medical attention. The number affected by this “second wave” is unknown.

And then goes on to argue that the large number of cases was due to mass hysteria.

JonathanLivengood 24 Dec 2012 19:55 UTC
−1 points
0
on: New censorship: against hypothetical violence against identifiable people
Do you have worked out numbers (in terms of community participation, support dollars, increased real-world violence, etc.) comparing the effect of having the censorship policy and the effect of allowing discussions that would be censored by the proposed policy? The request for “Consequences we haven’t considered” is hard to meet until we know with sufficient detail what exactly you have considered.

My gut thinks it is unlikely that having a censorship policy has a smaller negative public relations effect than having occasional discussions that violate the proposed policy. I know that I am personally much less okay with the proposed censorship policy than with having occasional posts and comments on LW that would violate the proposed policy.

JonathanLivengood 21 Dec 2012 8:55 UTC
2 points
0
in reply to: fubarobfusco’s comment on: Gun Control: How would we know?
What is the evidence that 2 is out? Suppose there are five available effective means to some end. If I take away one of them, doesn’t that reduce the availability of effective means to that end? Is the idea supposed to be that the various means are all so widely available that overall availability of means to the relevant end is not affected by eliminating (or greatly reducing) availability of one of them? Seems contentious to me. Moreover, what you say after the claim that 2 is out seems rather to support the claim that 2 is basically correct: poison, bombs, and knives are either practically less effective for one reason or another (hard to use, hard to practice, less destructive—in the case of knives) or practically less available for one reason or another (poisons not widely available).