VincentYu

Karma: 2,203

VincentYu 1 Feb 2013 21:36 UTC
70 points
on: Rationality Quotes February 2013

In Munich in the days of the great theoretical physicist Arnold Sommerfeld (1868–1954), trolley cars were cooled in summer by two small fans set into their ceilings. When the trolley was in motion, air flowing over its top would spin the fans, pulling warm air out of the cars. One student noticed that although the motion of any given fan was fairly random—fans could turn either clockwise or counterclockwise—the two fans in a single car nearly always rotated in opposite directions. Why was this? Finally he brought the problem to Sommerfeld.

“That is easy to explain,” said Sommerfeld. “Air hits the fan at the front of the car first, giving it a random motion in one direction. But once the trolley begins to move, a vortex created by the first fan travels down the top of the car and sets the second fan moving in precisely the same direction.”

“But, Professor Sommerfeld,” the student protested, “what happens is in fact the opposite! The two fans nearly always rotate in different directions.”

“Ahhhh!” said Sommerfeld. “But of course that is even easier to explain.”

Devine and Cohen, Absolute Zero Gravity, p. 96.

VincentYu 6 Apr 2013 3:24 UTC
52 points
on: The Universal Medical Journal Article Error
You claim that medical researchers are doing logical inference incorrectly. But they are in fact doing statistical inference and arguing inductively.

Statistical inference and inductive arguments belong in a Bayesian framework. You are making a straw man by translating them into a deductive framework.

Rephrased to say precisely what the study found:

This study tested and rejected the hypothesis that artificial food coloring causes hyperactivity in all children.

No. Mattes and Gittelman’s finding is stronger than your rephrasing—your rephrasing omits evidence useful for Bayesian reasoners. For instance, they repeatedly pointed out that they “[studied] only children who were already on the Feingold diet and who were reported by their parents to respond markedly to artificial food colorings.” They claim that this is important because “the Feingold diet hypothesis did not originate from observations of carefully diagnosed children but from anecdotal reports on children similar to the ones we studied.” In other words, they are making an inductive argument:
1. Most evidence for the Feingold diet hypothesis comes from anecdotal reports.
2. Most of these anecdotal reports are mistaken.
3. Thus, there is little evidence for the Feingold diet hypothesis.
4. Therefore, the Feingold diet hypothesis is wrong.
If you translate this into a deductive framework, of course it will not work. Their paper should be seen in a Bayesian framework, and in this context, their final sentence

The results of this study indicate that artificial food colorings do not affect the behavior of school-age children who are claimed to be sensitive to these agents.

translates into a correct statement about the evidence resulting from their study.

This refereed medical journal article, like many others, made the same mistake as my undergraduate logic students, moving the negation across the quantifier without changing the quantifier. I cannot recall ever seeing a medical journal article prove a negation and not make this mistake when stating its conclusions.

They are not making this mistake. You are looking at a straw man.

Full-texts:
What links here?
- Douglas_Knight's comment on The Universal Medical Journal Article Error by PhilGoetz (6 Apr 2013 15:47 UTC; 22 points)

VincentYu 13 Jun 2012 7:23 UTC
46 points
on: The New Yorker article on cognitive biases
Full paper (West, Meserve, & Stanovich, 2012).

The New Yorker article (“Why smart people are stupid”) by Lehrer misrepresents the main point of West et al.’s paper, which is that – as the original paper title asserts – “cognitive sophistication does not attenuate the bias blind spot”, instead of the article’s claim that smart people are more susceptible to cognitive biases.

Three negligent misrepresentations by Lehrer:

First, contrary to the general message of the article, West et al. state twice that most cognitive biases are negatively correlated with cognitive ability (emphasis mine):

The finding that the bias blind spot is more apparent among the more cognitively sophisticated individuals is contrary to much of the rest of the heuristics and biases literature where most biases are negatively correlated (or at least independent) with cognitive abilities (Bruine de Bruin et al., 2007; Chiesi et al., 2011; Del Missier et al., 2010, 2011; Finucane & Gullion, 2010; Kokis, Macpherson, Toplak, West, & Stanovich, 2002; Stanovich & West, 1998, 2000; Toplak et al., 2011; Weller, Levin, & Denburg, 2011; West et al., 2008).

Most cognitive biases in the heuristics and biases literature are negatively correlated with cognitive sophistication, whether the latter is indexed by development, by cognitive ability, or by thinking dispositions (Bruine de Bruin et al., 2007; Chiesi et al., 2011; Finucane & Gullion, 2010; Kokis et al., 2002; Toplak & Stanovich, 2002; Toplak et al., 2011; Weller et al., 2011; West et al., 2008).

Second, West et al. refer to the bias blind spot as a metabias because it is a pattern of inaccurate judgment in reasoning about cognitive biases. In this context, the distinction between bias and metabias (which the article neglects to maintain) is vital because the whole point of West et al.’s paper is that this particular metabias does not exhibit the same negative correlation with cognitive ability that many cognitive biases exhibit. They do not contest or assert to have a counterexample to the claim that most cognitive biases are negatively correlated with cognitive ability.

There is an important metabias that remains unexamined in all of the previous work on individual differences, however. It is the so-called bias blind spot—explored in an important article by Pronin, Lin, and Ross (2002). They found that people thought that various cognitive and motivational biases were much more prevalent in others than in themselves. Bias turns out to be relatively easy to recognize in the decisions of others, but often difficult to detect in one’s own judgments.

Third, West et al. are rather cautious (in contrast with the article) in stating the correlation strengths of bias blind spots with cognitive ability. For instance, they have as their paper title “Cognitive sophistication does not attenuate the bias blind spot”, rather than a stronger “Cognitive sophistication augments the bias blind spot”. However, they do find a slight positive correlation (two-tailed p < 0.05):

We found that none of these bias blind spot effects displayed a negative correlation with measures of cognitive ability (SAT total, CRT) or with measures of thinking dispositions (need for cognition, actively open-minded thinking). If anything, the correlations went in the other direction.

Whatever explains the slightly positive correlation, a conservative way to characterize the findings here is to say that cognitive ability provides no inoculation at all from the bias blind spot—the tendency to believe that biased thinking is more prevalent in others than in ourselves. In our data, cognitive ability did not attenuate the tendency toward a blind spot at all. Thus, the bias blind spot joins a small group of other effects such as myside bias and noncausal base-rate neglect (Stanovich & West, 2008b; Toplak & Stanovich, 2003) in being unmitigated by increases in intelligence.

Social news posts about the New Yorker article:
ETA: OB post
What links here?

VincentYu 3 Jul 2012 16:13 UTC
44 points
on: CFAR website launched
Nitpick/question: The ‘rationality camps’ seem to have been renamed to ‘retreats’. Why is this? A ‘retreat’ brings to mind religious connotations (and OED confirms this), and seems to augment phygish impressions.

VincentYu 11 Nov 2012 13:45 UTC
42 points
on: Rationality Quotes November 2012

Often a person uses some folk proverb to explain a behavioral event even though, on an earlier occasion, this same person used a directly contradictory folk proverb to explain the same type of event. For example, most of us have heard or said, “look before you leap.” Now there’s a useful, straightforward bit of behavioral advice—except that I vaguely remember admonishing on occasion, “he who hesitates is lost.” And “absence makes the heart grow fonder” is a pretty clear prediction of an emotional reaction to environmental events. But then what about “out of sight, out of mind”? And if “haste makes waste,” why do we sometimes hear that “time waits for no man”? How could the saying “two heads are better than one” not be true? Except that “too many cooks spoil the broth.” If I think “it’s better to be safe than sorry,” why do I also believe “nothing ventured, nothing gained”? And if “opposites attract,” why do “birds of a feather flock together”? I have counseled many students to “never to put off until tomorrow what you can do today.” But I hope my last advisee has never heard me say this, because I just told him, “cross that bridge when you come to it.”

The enormous appeal of clichés like these is that, taken together as implicit “explanations” of behavior, they cannot be refuted. No matter what happens, one of these explanations will be cited to cover it. No wonder we all think we are such excellent judges of human behavior and personality. We have an explanation for anything and everything that happens. Folk wisdom is cowardly in the sense that it takes no risk that it might be refuted.

Keith E. Stanovich, How to Think Straight About Psychology, 10th ed. (2013), 14.

ETA: Should have included the subsequent paragraph:

That folk wisdom is “after the fact” wisdom, and that it actually is useless in a truly predictive sense, is why sociologist Duncan Watts titled one of his books: Everything Is Obvious—Once You Know the Answer (2011). Watts discusses a classic paper by Lazarsfeld (1949) in which, over 60 years ago, he was dealing with the common criticism that “social science doesn’t tell us anything that we don’t already know.” Lazarsfeld listed a series of findings from a massive survey of 600,000 soldiers who had served during World War II; for example, that men from rural backgrounds were in better spirits during their time of service than soldiers from city backgrounds. People tend to find all of the survey results to be pretty obvious. In this example, for instance, people tend to think it obvious that rural men would have been used to harsher physical conditions and thus would have adapted better to the conditions of military life. It is likewise with all of the other findings—people find them pretty obvious. Lazarsfeld then reveals his punchline: All of the findings were the opposite of what was originally stated. For example, it was actually the case that men from city backgrounds were in better spirits during their time of service than soldiers from rural backgrounds. The last part of the learning exercise is for people to realize how easily they would have explained just the opposite finding. In the case of the actual outcome, people tend to explain it (when told of it first) by saying that they expected it because city men are used to working in crowded conditions and under hierarchical authority. They never realize how easily they would have concocted an explanation for exactly the opposite finding.
What links here?
- beoShaffer's comment on Rationality Quotes July 2013 by Vaniver (23 Jul 2013 2:32 UTC; 5 points)

VincentYu 20 Jan 2014 15:01 UTC
36 points
in reply to: jefftk’s comment on: 2013 Survey Results
What if the people who have taken IQ tests are on average smarter than the people who haven’t? My impression is that people mostly take IQ tests when they’re somewhat extreme: either low and trying to qualify for assistive services or high and trying to get “gifted” treatment. If we figure lesswrong draws mostly from the high end, then we should expect the IQ among test-takers to be higher than what we would get if we tested random people who had not previously been tested.

This sounds plausible, but from looking at the data, I don’t think this is happening in our sample. In particular, if this were the case, then we would expect the SAT scores of those who did not submit IQ data to be different from those who did submit IQ data. I ran an Anderson–Darling test on each of the following pairs of distributions:
- SAT out of 2400 for those who submitted IQ data (n = 89) vs SAT out of 2400 for those who did not submit IQ data (n = 230)
- SAT out of 1600 for those who submitted IQ data (n = 155) vs SAT out of 1600 for those who did not submit IQ data (n = 217)
The p-values came out as 0.477 and 0.436 respectively, which means that the Anderson–Darling test was unable to distinguish between the two distributions in each pair at any significance.

As I did for my last plot, I’ve once again computed for each distribution a kernel density estimate with bootstrapped confidence bands from 999 resamples. From visual inspection, I tend to agree that there is no clear difference between the distributions. The plots should be self-explanatory:

(More details about these plots are available in my previous comment.)

Edit: Updated plots. The kernel density estimates are now fixed-bandwidth using the Sheather–Jones method for bandwidth selection. The density near the right edge is bias-corrected using an ad hoc fix described by whuber on stats.SE.

[Link] Does a Simulation Really Need to Be Run?

VincentYu22 Jun 2011 23:28 UTC

28 points

17 comments1 min readLW link

VincentYu 2 Jun 2012 23:11 UTC
28 points
in reply to: lukeprog’s comment on: Reaching young math/compsci talent
SPARC and a number of mostly homogeneous math camps are all looking for pre-college students with strong mathematical ability. Since SPARC’s syllabus is notably different from that of math camps, it seems like a bad idea to compete with these camps for the top students. But competition is inevitable if SPARC runs at the same time as these camps; below I have found and listed the 2012 start and end dates for the most prominent math camps:
- Ross: June 18 – July 27
- PROMYS: July 1 – August 11
- HSMC: June 17 – July 28
- Mathcamp: July 1 – August 5
- HCSSiM: July 1 – August 11
- SUMaC: July 15 – August 11
- RSI*: June 24 – August 4
- (SPARC: August 6 – August 13)
SPARC’s starting date this year conflicts with the end dates of three of these seven camps. Perhaps there are other scheduling constraints, but if not, wouldn’t it be a good idea to run SPARC a week later to avoid conflicts? (It is too late to change this year, of course.)

*I know RSI is not a math camp in the spirit of the others, but it’s well-known and attracts some students away from math camps.

ETA: And since SPARC is free and relevant to math students, if it can guarantee that it will not conflict with the other program dates, I think many math programs will be happy to link to or otherwise mention SPARC to current and past students – this should help spread the word to more potential students.

VincentYu 19 Jan 2014 21:47 UTC
23 points
in reply to: jefftk’s comment on: 2013 Survey Results
Here is a kernel density estimate of the “true” distribution, with bootstrapped) pointwise 95% confidence bands from 999 resamples:

It looks plausibly bimodal, though one might want to construct a suitable hypothesis test for unimodality versus multimodality. Unfortunately, as you noted, we cannot distinguish between the hypothesis that the bimodality is due to rounding (at 500 M) versus the hypothesis that the bimodality is due to ambiguity between Europe and the EU. This holds even if a hypothesis test rejects a unimodal model, but if anyone is still interested in testing for unimodality, I suggest considering Efron and Tibshirani’s approach using the bootstrap.

Edit: Updated the plot. I switched from adaptive bandwidth to fixed bandwidth (because it seems to achieve higher efficiency), so parts of what I wrote below are no longer relevant—I’ve put these parts in square brackets.

Plot notes: [The adaptive bandwidth was achieved with Mathematica’s built-in “Adaptive” option for SmoothKernelDistribution, which is horribly documented; I think it uses the same algorithm as ‘akj’ in R’s quantreg package.] A Gaussian kernel was used with the bandwidth set according to Silverman’s rule-of-thumb [and the sensitivity (‘alpha’ in akj’s documentation) set to 0.5]. The bootstrap confidence intervals are “biased and unaccelerated” because I don’t (yet) understand how bias-corrected and accelerated bootstrap confidence intervals work. Tick marks on the x-axis represent the actual data with a slight jitter added to each point.
What links here?
- VincentYu's comment on 2013 Survey Results by Scott Alexander (20 Jan 2014 15:01 UTC; 36 points)

VincentYu 4 Aug 2012 17:58 UTC
22 points
on: Solving the two envelopes problem

But we know from basic probability theory that for two independent random variables, E(X/Y) > 1 does actually imply E(X) > E(Y).

This does not follow; a counterexample:

Suppose X and Y are independent random variables, with X taking the values {2,100}, and Y the values {1,150}, each with equal probability (i.e., each of X and Y follows the Bernoulli distribution with p = 0.5). Then we have
E(X/Y) = (2/1 + ¹⁰⁰⁄₁ + ²⁄₁₅₀ + ¹⁰⁰⁄₁₅₀) / 4 = 25.67 > 1,
but
E(X) = 51 < 75.5 = E(Y).

Keep in mind that the equation E(1/Y) = 1/E(Y) does not hold in general, because taking the inverse is not a linear transformation. To evaluate the expectation after a nonlinear transformation, one requires not just the orignal expectation, but also the pdf of the distribution. (I can’t be sure this is what you did, but misapplying this gives: E(X/Y) = E(X)E(1/Y) = E(X)/E(Y). The first equality holds if X and Y are independent or by definition if X and 1/Y are uncorrelated, but the second equality does not hold in general.)
What links here?
- Kelly Criteria and Two Envelopes by Wei Dai (16 Aug 2012 21:57 UTC; 11 points)

VincentYu 28 Jan 2014 18:21 UTC
21 points
on: [Error communicating with LW2 server]
A fundamental issue here is that von Neumann–Morgenstern (VNM) utility functions (also called cardinal utility functions, as opposed to ordinal utility functions) are not comparable across entities; after all, they are only invariant up to positive affine transformations.

This means that the relations in your post that involve more than one utility function are meaningless under the VNM framework. Contrary to popular misconception, the inequality

u_v(v(1), b(0)) > u_b(v(0), b(1))

tells us nothing about whether Veronica likes apple pies more than Betty does, and the equality

u_b(v(1), b(0)) = u_v(v(0), b(1)) = 0

tells us nothing about whether Betty and Veronica cares whether the other gets a pie.

A quick way to see this formally is to note that you may transform one of the utility functions (by a positive affine transformation of the form ax + b, which leaves VNM utility functions invariant) and get any relation you want between the pair of utility functions at the specified points.

A quick way to see this informally is to recall that only comparisons of differences in utility within a single entity are meaningful.

Von Neumann and Morgenstern address some of these common misunderstandings in Theory of Games and Economic Behavior (3rd ed., p. 11; italics original, bold mine):

A particularly striking expression of the popular misunderstanding about this pseudo-maximum problem [of utility maximization] is the famous statement according to which the purpose of social effort is the “greatest possible good for the greatest possible number.” A guiding principle cannot be formulated by the requirement of maximizing two (or more) functions at once.

Such a principle, taken literally, is self-contradictory. (In general on function will have no maximum where the other function has one.) It is no better than saying, e.g., that a firm should obtain maximum prices at maximum turnover, or a maximum revenue at minimum outlay. If some order of importance of these principles or some weighted average is meant, this should be stated. However, in the situation of the participants in a social economy nothing of that sort is intended, but all maxima are desired at once—by various participants.

One would be mistaken to believe that it can be obviated, like the difficulty in the Crusoe case mentioned in footnote 2 on p. 10, by a mere recourse to the devices of the theory of probability. Every participant can determine the variables which describe his own actions but not those of the others. Nevertheless those “alien” variables cannot, form his point of view, be described by statistical assumptions. This is because the others are guided, just as he himself, by rational principles—whatever that may mean—and no modus procedendi can be correct which does not attempt to understand those principles and the interactions of the conflicting interests of all participants.

Sometimes some of these interests run more or less parallel—then we are nearer to a simple maximum problem. But they can just as well be opposed. The general theory must cover all these possibilities, all intermediary stages, and all their combinations.

To directly address the issue of utility comparison across entities, refer to this footnote on p. 19:

We have not obtained [from the von Neumann–Morgenstern axioms] any basis for a comparison, quantitatively or qualitatively, of the utilities of different individuals.

I highly recommend reading the first sections of the book. Its copyright has expired and the Internet Archive has a scan of the book.

A quick note for anyone confused over why the utility functions here are so much weaker than what they are used to seeing: You are probably used to seeing “utility” in discussions of utilitarianism, in which “utility” generally does not fall under the VNM framework (they are often intended to be much stronger than VNM utilities, so that they are no longer invariant under positive affine transformations and can be compared across entities—the trouble here is that there is no sensible formalization that captures these properties); that is, “utility” in utilitarianism suffers from a namespace collision with “utility” in economics and decision theory (“utility” in economics and decision theory also often refer to different things: ordinal utility is more common in the former whereas cardinal utility is more common in the latter).

VincentYu 11 May 2012 15:38 UTC
21 points
in reply to: Mitchell_Porter’s comment on: Holden Karnofsky’s Singularity Institute critique: other objections
A variant of this contest with less catastrophic unfriendliness actually ran for a few years. The (now defunct) Underhanded C Contest (description below from the contest web page):

The Underhanded C Contest is an annual contest to write innocent-looking C code implementing malicious behavior. In this contest you must write C code that is as readable, clear, innocent and straightforward as possible, and yet it must fail to perform at its apparent function. To be more specific, it should do something subtly evil.

Every year, we will propose a challenge to coders to solve a simple data processing problem, but with covert malicious behavior. Examples include miscounting votes, shaving money from financial transactions, or leaking information to an eavesdropper. The main goal, however, is to write source code that easily passes visual inspection by other programmers.

VincentYu 7 Apr 2013 17:09 UTC
20 points
in reply to: PhilGoetz’s comment on: The Universal Medical Journal Article Error
I don’t think you responded to my criticisms and I have nothing further to add. However, there are a few critical mistakes in what you have added that you need to correct:
Now pay attention; this is the part everyone gets wrong, including most of the commenters below.

The methodology used in this study, and in most studies, is as follows:
- Divide subjects into a test group and a control group.
No, Mattes and Gittelman ran an order-randomized crossover study. In crossover studies, subjects serve as their own controls and they are not partitioned into test and control groups.

If you don’t understand why that is so, read the articles about the t-test and the F-test. The tests compute what a difference in magnitude of response such that, 95% of the time, if the measured effect difference is that large, the null hypothesis (that the responses of all subjects in both groups were drawn from the same distribution) is false.

No, the correct form is:
- The tests compute a difference in magnitude of response such that if the null hypothesis is true, then 95% of the time the measured effect is not that large.
The form you quoted is a deadly undergraduate mistake.

ADDED: People are making comments proving they don’t understand how the F-test works. This is how it works: You are testing the hypothesis that two groups respond differently to food dye.

Suppose you measured the number of times a kid shouted or jumped, and you found that kids fed food dye shouted or jumped an average of 20 times per hour, and kids not fed food dye shouted or jumped an average of 17 times per hour. When you run your F-test, you compute that, assuming all kids respond to food dye the same way, you need a difference of 4 to conclude that the two distributions (test and control) are different.

If the food dye kids had shouted/jumped 21 times per hour, the study would conclude that food dye causes hyperactivity. Because they shouted/jumped only 20 times per hour, it failed to prove that food dye causes hyperactivity. That failure to prove is then taken as having proved that food dye does not cause hyperactivity, even though the evidence indicated that food dye causes hyperactivity.

This is wrong. There are reasonable prior distributions for which the observation of a small positive sample difference is evidence for a non-positive population difference. For example, this happens when the prior distribution for the population difference can be roughly factored into a null hypothesis and an alternative hypothesis that predicts a very large positive difference.

In particular, contrary to your claim, the small increase of 3 can be evidence that food dye does not cause hyperactivity if the prior distribution can be factored into a null hypothesis and an alternative hypothesis that predicts a positive response much greater than 3. This is analogous to one of Mattes and Gittelman’s central claims (they claim to have studied children for which the alternative hypothesis predicted a very large response).

VincentYu 13 Jan 2015 10:44 UTC
19 points
in reply to: NatPhilosopher’s comment on: What topics are appropriate for LessWrong?
Regardless of the conflicts between your beliefs and that of others, it is dishonest to misrepresent the writings of other people to bolster your arguments. Do not manipulate quotes from a report to make it say what you want it to say, even if you believe that the original report is incorrect.

VincentYu 15 Mar 2014 4:46 UTC
19 points
on: What are you working on? March / April 2014
Since December, I have been working on a proposed extension to OpenPGP to add support for some sort of designated verifier signature. Most of this work involved doing a literature review and carefully reading the OpenPGP specification to see what cryptographic schemes can be implemented successfully in OpenPGP. I settled on a ring signature scheme by Abe et al. (2004), and recently posted a proposal to the OpenPGP mailing list. Note that this is a proposal for a draft of a proposed standard, so things remain very flexible. This is the best time for anyone to comment on the proposal, before I write up an I-D.

The following is a snippet of the description of ring signatures that I gave in my overview of the proposal:

Ring signatures allow a signer to specify a set of possible signers without revealing who actually signed the message.

The current OpenPGP specification lacks a straightforward way for Alice to send Bob an authenticated message that can only be verified by Bob. Right now, if such a signature is desired, the best that Alice can do is to send Bob a signed-then-encrypted message. Unfortunately, this means that Bob can show others the decrypted signature to prove to others that Alice signed the message. This ability of Bob to transfer Alice’s signature to others is often undesirable from Alice’s point of view.

Ring signatures provide an easy way for Alice to authenticate a message to Bob without concomitantly giving Bob the ability to transfer her signature. Alice simply creates a ring signature that specifies Alice and Bob as the possible signers of a message. Upon receipt of the message and ring signature, Bob is assured that the message is signed by Alice, but he cannot prove to others that Alice signed the message because Bob could have created the ring signature himself.

I am interested to hear from the perspective of end users. For instance:
- Is there anything puzzling in my description of ring signatures?
- Are there any particular features that you desire?
Of course, I would also like to hear from anyone who has the time to go through the details of the proposal. Sections 3 and 4 contain the bulk of the actual cryptographic scheme, and it closely follows Abe et al.’s scheme with only small modifications to better fit OpenPGP. Some resources:
- Ring signatures are interesting by themselves, and I think some Less Wrong readers might enjoy reading the following papers. The standard reference for ring signatures is Rivest et al. (2006); I would recommend reading this paper first if you plan on reading Abe et al.’s paper, because it contains a very clear exposition of the ideas behind ring signatures. A third paper that I recommend is Bresson et al. (2002); Sections 1–3 are particularly useful.
- The current OpenPGP specification is given in RFC 4880. I don’t recommend reading it unless you have some burning desire to learn more about OpenPGP; it’s not particularly illuminating.
In the best case scenario, ring signatures get deployed on GnuPG within a few years and Alice can create a ring-signed message for Bob simply by running: gpg2 --clearsign --ring-signature --recipient Bob

(In the not-quite-worst case scenario, I write the I-D, everyone agrees it’s crap, and I vow never to touch math again.)

VincentYu 11 Oct 2013 23:01 UTC
19 points
on: Does the simulation argument even need simulations?
1. Bite the bullet: we are most likely not even a computer simulation, just a mathematical construct[3]
Biting the bullet here is roughly equivalent to accepting Tegmark’s Ultimate Ensemble. This was discussed on LW in ata’s post from 2010, The mathematical universe: the map that is the territory.

See Tegmark (2008). In particular, Section 6, “Implications for the simulation argument”. A relevant extract:

For example, since every universe simulation corresponds to a mathematical structure, and therefore already exists in the Level IV multiverse [the multiverse of all mathematical structures], does it in some meaningful sense exist “more” if it is in addition run on a computer? This question is further complicated by the fact that eternal inflation predicts an infinite space with infinitely many planets, civilizations, and computers, and that the Level IV multiverse includes an infinite number of possible simulations. The above-mentioned fact that our universe (together with the entire Level III multiverse) may be simulatable by quite a short computer program (Sect. 6.2) calls into question whether it makes any ontological difference whether simulations are “run” or not. If, as argued above, the computer need only describe and not compute the history, then the complete description would probably fit on a single memory stick, and no CPU power would be required. It would appear absurd that the existence of this memory stick would have any impact whatsoever on whether the multiverse it describes exists “for real”. Even if the existence of the memory stick mattered, some elements of this multiverse will contain an identical memory stick that would “recursively” support its own physical existence. This would not involve any Catch-22 “chicken-and-egg” problem regarding whether the stick or the multiverse existed first, since the multiverse elements are 4-dimensional spacetimes, whereas “creation” is of course only a meaningful notion within a spacetime.

A while ago, I posted a LW discussion link to John Regehr’s blog post about similar ideas: Does a simulation really need to be run?.
What links here?
- VincentYu's comment on Does the simulation argument even need simulations? by lmm (12 Oct 2013 17:09 UTC; 6 points)

VincentYu 5 Nov 2012 17:02 UTC
19 points
in reply to: gwern’s comment on: 2012 Less Wrong Census/Survey
Commenting here because this is the highest voted comment mentioning the Big Five.

The Big Five personality test linked to in the survey is an online implementation of the Big Five Inventory (BFI). It was used in two studies: Srivastava et al. (2003) and Gosling et al. (2004.

The most recent canonical citation of the BFI is John et al. (2008). The BFI is very widely used in the literature, so descriptive statistics (i.e., mean and SD) for different populations are available from many studies.

The percentiles from the online test were approximated under a Gaussian assumption (scores on Big Five inventories are typically not well-approximated by Gaussians, so the percentile rankings are bound to be off). Everyone was normed to the same distribution (i.e., age, gender, etc. do not affect the results; only the 44 questions on the BFI do). The exact mean and SD used by the test to calculate percentile rankings are as follows (converted to the usual 5-point Likert scale):
- E: mean = 3.30, SD = 0.88
- A: mean = 3.66, SD = 0.70
- C: mean = 3.40, SD = 0.76
- N: mean = 3.15, SD = 0.85
- O: mean = 3.85, SD = 0.65
In order to compare LW’s results to norms in the literature, we need to convert the percentiles back to the 5-point scale. This is easy; just use an inverse CDF. Unfortunately, we are unable to recover the full range of results because some scores at the tails were rounded to the same percentile bins (note that the “percentiles” were actually rounded values from the CDF, so each percentile bin was shifted by 0.5 percentiles from what they should have been; truncation/floor should have been used instead of rounding).

As a quick comparison to the norms used by the online test, here are norms for the BFI from Donnellan et al. (2006) (n = 300, 78.7% female, sampled from undergraduates enrolled in psychology courses at Michigan State University):
- E: mean = 3.43, SD = 0.72
- A: mean = 3.82, SD = 0.56
- C: mean = 3.63, SD = 0.60
- N: mean = 2.93, SD = 0.73
- O: mean = 3.50, SD = 0.57
Because we do not have the individual responses for each item on the BFI, we will be unable to test the fit of the five-factor model to LW data or to confirm the expected factor loading of each item.
What links here?
- VincentYu's comment on 2014 Less Wrong Census/Survey—Call For Critiques/Questions by Scott Alexander (13 Oct 2014 13:05 UTC; 9 points)

VincentYu 20 Jul 2012 10:29 UTC
19 points
in reply to: MixedNuts’s comment on: [link] Prepared to wait? New research challenges the idea that we favour small rewards now over bigger later
Why not quote the abstract of the original paper? It is, after all, the authors’ best attempt at accurately summarizing their results and hooking the reader:

Hyperbolic discounting of delayed rewards has been proposed as an underlying cause of the failure to stick to plans to forego one’s immediate desires, such as the plan to diet, wake up early, or quit taking heroin. We conducted two tests of inconsistent planning in which respondents made at least two choices between a smaller–sooner (SS) and larger–later (LL) amount of money, one several weeks before SS would be received, and one immediately before. Hyperbolic discounting predicts that there would be more choices of SS as it became more proximate—and, equivalently, that among those who change their mind, “impatient shifts” (LL-to-SS) will be more common than “patient shifts” (SS-to-LL). We find no evidence for this, however, and in our studies shifts in both directions were equally likely. We propose that some of the evidence cited on behalf of hyperbolic discounting can be attributed to qualitatively different psychological mechanisms.

VincentYu 8 Jan 2013 16:15 UTC
18 points
on: [Link] Your Elusive Future Self

Another possibility is that people “might well anticipate substantial change, yet not know how they would change, and thus, just predict the status quo”

To give a specific example: If personality can be modeled over time as a martingale)—meaning that the expected future personality is always equal to the current personality—then it is quite appropriate for participants to provide their current personality as a prediction of their future personality, even if they think that the standard deviation of the martingale is large.

More generally, given the current personality $c$ and random variable $p$ following the distribution of predicted future personality, the predicted change in personality can be characterized by the expected absolute deviation
) or the expected squared deviation
%5E2)). However, study 1 in the paper measured the absolute expected deviation
%7C) instead, which is a poor measure. (This limitation of the study is understandable because collecting data on $c$ and
) is relatively easy.)

The authors tried to circumvent this limitation in a replication of study 1 by asking each participant to directly predict change rather than inferring predicted change from $c$ and
). P. 97 column 2:

Instead of being asked to report or predict their specific personality traits, these participants were simply asked to report how much they felt they had “changed as a person over the last 10 years” and how much they thought they would “change as a person over the next 10 years.”

I think there is an interpretational issue here—we cannot be sure if the participants actually estimated something similar to the expected absolute deviation
). If the participants relied on imagining a future self, then this will again give the absolute expected deviation
%7C).

I’m surprised that the authors did not report whether any of the Big Five personality domains mediated this “end of history illusion”. That seems like an obvious thing to check. Maybe they’re saving that for a later paper.

“The single best way to make predictions about what you’re going to want in the future isn’t to imagine yourself in the future, … it’s to look at other people who are in the very future you’re imagining,” [Gilbert] says.

In other words, take the outside view.