supposedlyfun

Posts

Sorted by New

Comments

A Semitechnical Introductory Dialogue on Solomonoff Induction

The language of physics is differential equations, and it turns out that this is something difficult to beat into some human brains

You rang?

I'm not sure why, but I find these dialogues easier to learn from than an article expressing the same ideas in the same order, even in an explicitly Q&A format.

Economic Class

businesses might be afraid of getting sued anyway

This is the correct explanation of why nobody does this.  

Here is what I hope is a gears-y rundown from a lawyer.  I would appreciate feedback as I've been pondering a post or sequence in this vein (probably a different topic).

If a company client came to me and said, "I want to hire based on IQ; to hell with Duke Power," I would say, 

"We can make this happen in a legally defensible way.  It will slightly increase the likelihood that a plaintiff-side employment lawyer will take the case of a person who doesn't get hired, because of Duke Power and because IQ tests are so unusual in the hiring context.  That, in turn, increases the likelihood that we end up in front of a jury, which might as well be an RNG for our purposes. [Plaintiffs without lawyers almost never make it to a jury, and they are terrible when they do.]  You're raising your annual risk of a big jury verdict by maybe half a percent per year.

"First, I need you to pretend I never went to college, then explain to me (1) why IQ is a good hiring measure for this job.  Then, (2) why some other measure isn't better than IQ (a college degree in your field, for example; remember, I am an average American, I like to think that college degrees mean something).  Then, (3) how you arrived at the IQ cutoff you plan to use for this job.  

"Next, we need to hire a psychometrician who will write a report endorsing your views on items 1 through 3.  The report will also state that whatever discrimination is baked into IQ test results based on {race, gender, religion, ethnicity, national origin, age, disability status*, military service} won't have a disparate impact on people in those categories. This needs to be a real psychometrician, maybe somebody who works for schools or is a tenured professor, not somebody we found on Expert Witness Search dot com. The report will go in my file and yours.  It should be updated every year.  If you can't get this report, I advise you not to include IQ in your hiring rubric.

"Now.  Are you sure about your answer to number 2?

I would be looking for an explanation like, "We're basically DARPA, but with harder problems, for [say] rocketry.  Three-quarters of the postdocs we hire with Ph.Ds in actual rocketry wash out because our rocketry is so complex.  They stare at the differential equations for five minutes, then vomit.  But we did Career Day with a Mensa club and ended up hiring several middle schoolers, who tripled our effectiveness.  I cry myself to sleep every night thinking about how even a Ph.D in rocketry is not a good enough cutoff for our hiring decisions.  I have tried everything else and still can't get good people.  I dream of rockets, and all I want is to--to rocket them.  There's just something about rocketeering that requires a high IQ."

Is this more than "demonstrably a reasonable measure"? Maybe. But it's good to have some safety margin built in, with employment law as with boxing an Oracle AI.

Texas Freeze Retrospective: meetup notes

This was extremely valuable content for me. This kind of object-level data doesn't come through in normal news reports. These people went longer without power, and got much colder, than the other specific stories I'd heard.

A whirlwind tour of Ethereum finance

One of the things I appreciate about this community is posts like this, where a smart person gets the itch to learn all about something interesting, then writes it up in a way that effectively conveys their synthesis.

Economic Class

Hiring directly based on IQ is illegal.

This is not correct, in the U.S. states where I've practiced employment law.  Are you talking about a particular state?

If this is your interpretation of the US Supreme Court's decision in Griggs v. Duke Power Company, I would disagree.  

Nothing in the Act precludes the use of testing or measuring procedures; obviously they are useful. What Congress has forbidden is giving these devices and mechanisms controlling force unless they are demonstrably a reasonable measure of job performance.

401 U.S. 424, 436 (1971).  What had happened was, Duke Power enjoyed its explicitly racially segregated work force and wanted to keep it, but the newish Title VII of the Civil Rights Act of 1964 said that was illegal, so Duke imposed an intelligence test on certain jobs it wanted to keep white but which didn't really require high intelligence-as-tested.  This had the effect of disproportionately excluding Black applicants, who apparently didn't do as well on the test for reasons that are not germane to the holding.  

Held: If a facially neutral job requirement has the effect of discriminating based on a protected category, the employer has to show that the requirement is "a reasonable measure of job performance"--with a strong undertone of "How stupid did you think we were?" from Chief Justice Burger, writing for a unanimous court.

Different kinds of language proficiency

English natively, then learned Spanish in school, all throughout primary, secondary, and university.  I often dreamed in Spanish and could communicate fluently.

The emotional valence you're talking about is something I never experienced, probably because I never trained it.  All of my Spanish was academic, even the literature classes I was taking in college.  I had very few organic Spanish conversations with Spanish speakers, and I never needed (or had the opportunity) to use Spanish to build a friendship or court anyone.  

The few times I was purely socializing with Spanish speakers for (say) an hour or more, the Spanish part of my brain kicked into overdrive, and I was even more fluent than normal--like the English part of my brain was just gone, and I was meta-aware of that feeling, but meta-aware in Spanish.  It was similar to a flow state.  The chasm was definitely gone in those situations.  Or at least it was for me--maybe my Spanish was just bad enough for the other person that I was in an uncanny valley (uncanny chasm??).

The Comprehension Curve

The part about habitual speed rings true to me. I am slowly working my way through the Bermúdez textbook on cognitive science and find that my brain "wants" to read at a customary pace, which is too fast for comprehension of a subject I'm relatively unfamiliar with (haven't touched a science textbook since high school).

Forcing myself to slow down gut-feels like going under the speed limit in the far-left lane on the interstate, like wasting time, even though I conscious-know it's approximately infinity times more important to understand the text than it is to be able to truthfully say that I did "reading" on all the words in it.

Overconfidence is Deceit

When a presidential candidate promises to do something that would require congressional approval but ends up unable to get it, how is that not an instance of apparent overconfidence? And neither president seems to have suffered electorally due to that particular failure. So they both seem like evidence for the claim that apparent overconfidence (within reason) isn't punished.

The map and territory of NFT art

This comment captures both the substance and the style of my reaction to the story.

Covid 1/7: The Fire of a Thousand Suns

You deserve to be paid for these covid updates. Are you getting paid? How does one pay you for this?  This must've taken at least 15 hours.

Load More