Sorted by New


2014 Survey Results

I'd expect a Pareto distribution for charitable donations, not log-normal, and that's exactly what the histogram looks like:

Looks like alpha >> 2, so the variance is infinite.

Happiness interventions

It's good to see Alan Sokal is still doing God's work.

Autism, or early isolation?

Why these things? They largely involve plenty of "analytical engine" skill. I think I'm a pretty good singer, I was varsity basketball, had good enough balance and coordination to climb V6 before injuries, won the district-wide art show in high school three years in a row, fix all my own plumbing and fixed my lawn mower engine. My wife literally rebuilt her car's circuit board, which is maybe more up the typical geek alley, but if you can do that, or build a gaming platform from parts, you can rebuild a lawn mower engine. You got me on social skills, but I don't think that's universal for smart people so much as universal for people who use most of their socializing resources on the Internet. General intelligence doesn't have to mean "super focused on one thing." You might have to give 100% attention if you ever want to be Kobe Bryant or something, but you can be really good at a lot of things without being among the top two or three in the world at any of them.

Anecdata, but just as reference to get away from bragging, the guy who got the second highest SAT score at my high school is now a pro rugby player. My best friend from college, who scored pretty close to us, just won an Emmy for writing comedy television.

When does heritable low fitness need to be explained?

And traditional behavior gives us an imperfect window into the economics of the past, which is what's under discussion when we talk about historical selective fitness.

I think we should keep in mind just how far back we're talking. I'm not saying we inherited homosexuality from our common ancestor with the modern fruit fly, but at least our common ancestor with other great apes. Framing the question as why would it be selected for in the context of human societies is probably wrong, when what we want to know is why it wasn't sufficiently selected against given it already existed (I doubt we'll ever figure what advantage it gave the proto-ape whose social structures we'll never know). Once a trait already manifests in 3% of the population, it takes work to get rid of it, and even within that 3%, it was doubtful the case that 0% of them reproduced while 100% of heterosexual men reproduced. I'm sure it wasn't exactly parity, but it's possible there is no explanation in terms of the organization of human societies except for we're really optimized to enjoy sex, sometimes that wire gets flipped, and it doesn't provide an advantage, but it also doesn't give enough of a disadvantage to completely disappear within 300,000 years.

Don't forget also, that if some gene combo is necessary but not sufficient, and requires other developmental factors to manifest that don't manifest in your brothers and cousins (which seems to be the case if it's only 20% between twins), then when they reproduce, even if you don't, the gene still gets passed on. Take me, for example. I'm not gay, but I am sterile and don't want kids anyway. Nonetheless, I have 3 sisters and 13 cousins that have had kids so far. Without doing the exact math, off the top of my head I'm guessing at least 80-90% of whatever I'm carrying made it to the next generation.

Edit: Also, one last thing is we don't know the prevalence in the ancestral population. Given it's roughly 100% bisexual in such a closely related other species, it could have been fairly high in the common ancestor, obviously not 100% obligate, but more than 3%, and it actually has been selected against, a lot, just not enough to get us to zero yet.

[Link] Robots Program People

Did you know about this?


The SUBNETS vision is distinct from current therapeutic approaches in that it seeks to create an implanted, closed-loop diagnostic and therapeutic system for treating, and possibly even curing, neuropsychological illness. That vision is premised on the understanding that brain function—and dysfunction, in the case of neuropsychological illness—plays out across distributed neural systems, as opposed to being strictly relegated to distinct anatomical regions of the brain. The program also aims to take advantage of neural plasticity, a feature of the brain by which the organ’s anatomy and physiology alter over time to support normal brain function.

Sounds pretty straightforwardly like programming a brain.

[Link] Word-vector based DL system achieves human parity in verbal IQ tests

Just to pimp my school, Georgia Tech offers a free course through Udacity in Knowledge-Based AI that involves programming an agent to take the Raven's progressive matrices test. I never took the course, but I wanna say from hearing other students that somewhere around 80 is the current state of the art (that's not an IQ and I'm not sure how to translate a Raven's score to an IQ).

In praise of gullibility?

This seems like a decent explanation of why I change my own mind as frequently as I do. If you're just tracking my history of Internet comments, I probably sound all over the place, but it's really me going from 54% certain of position X to 52% certain of not X, and it's hard to properly express that in an environment prone to rhetorical flourish and a debate atmosphere where you feel like you really really can't back down or you'll look weak. Most of the interesting things out there are very hard to legitimately be certain of. Factor in availability bias and it's easy to find yourself arguing for something you're really on the fence about just because you read a good argument for it a few hours ago (but not really any better than the argument for the opposite position a few days ago), then you make a good argument because you're good at arguing, and you just convinced yourself without actually introducing any new evidence.

And now I'm trapped in an infinite meta-regress wondering if I actually believe what I just wrote or it just sounds plausible.