Hacker News Reader

Comment by aidenn0 | Hacker News Reader

Any time I talk to someone about this evidence it's always "yeah, most people suck at telling liars, but I'm good at it."

Lake Wobegon effect for people reading.

View on HN

There's also the very strong confirmation bias.

We're all very good at detecting the liars we detect.

(This is the fundamental diagnostics fallacy. Absent some known ground truth, identifying hard-to-detect, hard-to-confirm phenomena is, well, hard.)

(This then gets coupled to the treatment fallacy, post hoc ergo prompter. I did A, B was cured, therefore A cured B.)

I think people probably immediately think about how they react to their spouse or other very close relationships, and I suspect (haven't looked it up) that people are actually pretty good at detecting lies in those situations because we have so much experience with how that person acts.

Detecting the lies of strangers? Yeah... good luck.

This article primarily focuses on the ineffectiveness of using non-verbal cues. It doesn’t really go into that much depth on how easily (or not) liars can be detected by analyzing what they actually say. It briefly mentions that this method appears more effective (although the accuracy was still only 80% or so).

I should have included more details, but the claimed expert lie-detectors almost always name non-verbal cues as the trick to detecting liars.

Any time I read about spotting liars I thought "man, I don't know how other people do it, but I suck at detecting lies".

I mostly think this because every time someone on TV insists they're innocent, I believe them, and then they turn out to be guilty.

Why do you choose to "believe" or not to "believe" an actor based off of what you know to be a performance?

I think that is the joke.

Not actors, actual criminals from the news.

Lake Woebegon effect is such a terrible name. It is very easily possible that every student in a small town is above average compared to e.g. the national average.

And the eponymous commons didn't suffer from a tragedy of the commons. It's still the name.

I have to question whether you're trying to prove GPs point, as this is hilariously incorrect. Unless you're using very unlikely values for easy and/or small.

I think you're right but I had to think about it and might not have noticed. How did you have that ready to go?

I have a basic understanding of statistics and a basic understanding of logical fallacies, and I deal with this kind of sounds-good-if-you-don't-think-too-hard "insights" all the time.

> basic understanding of logical fallacies

Did you spot one here?

> I have a basic understanding of statistics

Perhaps an intermediate understanding would help. Entire schools tend to score higher or lower together, also wealth correlates to IQ and scholastic success and zip code. There are also other reasons an entire group would be biased higher or lower.

> I deal with this kind of sounds-good-if-you-don't-think-too-hard "insights" all the time

Uh-huh.

Could you help me understand what my fallacy was?

The probability is just pow(0.5, N) where N is the student population, assuming that the distribution has no skew, assuming independence between students, and assuming the students are drawn randomly from the general population. It's an exponentially decreasing function, so it becomes incredibly unlikely for large student bodies under these assumptions.

I disagree with this argument, but that's the basis of it.

Now I'm thinking of a horror fan fiction version of lake wobegon where they give all kids an IQ test at a young age and murder all those below the national average.

Just my crazy version of what "easy" could entail.

They're not incorrect.

Firstly, you're assuming independent probabilities between the students.

If it's a single class of 30, for certain skills it might be much more likely than pow(0.5, 30) if they were all specifically coached on that particular skill, or if there are selection effects impacting what type of person ends up in that group of 30.

Secondly, even if we grant the assumption of independence, the claim doesn't rely on "unlikely values for ... small". There's many small towns with a student body of only 5 or less people, and pow(0.5, 5) isn't that unlikely.

Reply on news.ycombinator.com