Month: June 2015

Improving Commuting

As a follow up to yesterday, I just want to say that having a thing I look forward to that I can only do on the bus is working out brilliantly as a device for making commuting less miserable and it was totally worth buying the Kindle Fire to make it happen.  If you want to try for yourself, I recommend getting the 16gb, the 8gb could barely hold three HD episodes. I also recommend an anti-glare screen, because its glare is awful and the screen picks up fingerprints like woah.

The next recommendation is a little weirder.  I bike to the bus, which means I’m in biking clothes and have to maneuver my bike onto the cow catcher.  This makes waiting for the bus a little tricky- anything I have in my hands will have to be put away really quickly so I can rack my bike and get on, but I have no pockets.  And for reasons that may not be entirely rational, the time it takes to unsling my backpack and put a toy away makes me really stressed out.  But if I don’t use a toy the time at the bus stop is wasted, and that’s stressful too.  What I wanted was a way to make the whole of commuting feel either productive or like an indulgence.

So I made myself a shirt, out of two shirts and some fabric glue.

IMG_20150609_140424192
Two shirts enter, one shirt leaves
IMG_20150609_141218347
I believe my cat may have been getting high off the fabric glue.

Steps:

  1. Place desired toy on shirt, cut patch around it.  Leave space for depth.
  2. Put fabric glue on single edge of patch and apply to surviving shirt.
  3. Lock cat in other room.
  4. Reapply patch to surviving shirt.  Once you have it aligned, apply glue to other two edges and apply (you leave the top open so it’s a pocket).  Remember to apply it less than completely flat, so the pocket will have depth.
  5. I don’t know if this is a property of the fabric glue, my shirt, or this project in general, but some of the glue filtered down such that the base shirt was glued to itself and the patch ran free.  If you catch this early enough you can separate and try again.

This shirt is not aesthetically pleasing and is weirdly stiff where I overapplied fabric glue, but it gets the job done.

Making Hannibal Make Sense

[explicit spoilers for episode 1 of Hannibal, implied spoilers for 2 and 3.  I haven’t seen past that  but I am making predictions about what will happen]

My need to make movies make sense is most obvious when it comes to animal behavior*, but that’s not its only manifestation.  Take police procedurals.  I made Bones tolerable by saying la la la science fiction.  I’m now watching Hannibal because it’s one of the best shows available on Amazon Prime Streaming, which is the only streaming service that allows offline viewing, which is important because I have a bus commute again.**

Hannibal… does not make a ton of sense.  The premise is that an ex homicide detective is a brilliant criminal profiler but unable to get into the FBI because he couldn’t pass the psych screening, so they just hire him as a “special agent” that does everything a normal agent does, including field work.  You would think that was Hannibal Lecter, but it is not.  Lecter is the psychiatrist the FBI hires to babysit the unstable profiler (who I can only imagine will one day regret copying Lecter’s answers off a psych test).  Lecter also accompanies the agent on field work, which I don’t think even regular FBI agent’s counselors are normally allowed to do.  The actual profiler (Will Graham) comes up with theories more or less out of thin air and they always turn out to be correct.  He insists the evidence makes it obvious.

On the face of it, this makes no sense.  But what if Hannibal is set in a just slightly alternate universe, where psychic powers exist but are considered shameful because they make people mentally unstable?  The FBI can’t officially hire them, and they in fact test for powers before hiring to make sure they don’t hire any psychics.  But if there just happens to be a profiler who makes totally evidence based leaps, enabling them to catch really gruesome serial killers they otherwise couldn’t, well surely you can hire them on a contract basis.***

Why doesn’t the psychic Graham notice Hannibal is murdering and eating people? I’m glad you asked.  Hiring a babysitter for a psychic is tricky business.  You wouldn’t want someone they could manipulate or even read, that would defeat the purpose.  So you hire someone who’s a null field to psychics, or at least your psychic.  As a bonus, this person can lie to them when need be.  Note that Graham says he hates going to psychiatrists- presumably every one he’s been to has been unreadable to him, and he hates that.

And that’s how I learned to enjoy Hannibal.

*Ask my brother about watching the King Kong with me.

**”So you’re coping with your motion sickness by watching a show about food and gore?” “Yes”

***The same way my team at Microsoft rehired the same contractor for our Apple version over and over again.

Review: Inside Out

First, it’s great, and totally reverses Pixar’s downward slide.  Second, they did an amazing job of making everything work as a character/story and a metaphor for how the brain works and an Aesop.  A good chunk of the time kids’ shows teach lessons I consider abhorrent (looking at you, My Little Friendship Means Never Having Boundaries) , and at the beginning it looked like they were setting up one of those, but in the end in became clear they considered it exactly as problematic as I did.

Also I need to see it at home so I can cry harder. That is all.

Link 6/26/15

Another argument for remote work.  Pixar once accidentally deleted the entirety of Toy Story 2.  When I read the headline I assumed this was some sort of really complicated RAID failure, but no, someone “rm –rf”ed from the wrong directory and the backup’s overflow mode was “overwrite old data.” Translation: someone typed “delete all” from the wrong directory, the backup system didn’t have half the space it needed, and it didn’t tell anyone when it ran out.  The movie was saved because an employee was working from home after giving birth, and she had a two week old copy at her house.

Then Pixar decided they didn’t like the movie and redid the whole thing.  But it was a great story up till then.

How do you think family income predicts children’s college attendance?  This article gives you a chance to predict it before you see the actual results (along with other people’s guesses).  Keep in mind the x-axis is income percentile, not income.  I was unable to find a graph that went by income.

Attempted chlamydia vaccine in the 1960s made people more likely to get the infection (and this was not just because they had riskier sex- they saw the same effects in mice).  New research suggests this is because the vaccine somehow stimulated the part of the immune system that is supposed to recognize your own cells and prevent your immune system from attacking it.  The vaccine taught your immune system to ignore chlamydia infection.  They think they’ve found a way to correct for that.  Bonus: Vaccine works better as a nasal spray than a skin injection because the uterine lining is more like the sinuses.

Putting the bed net problem in perspective.

Diet, Skeptics, and Getting It Wrong:  “Mocking people for wanting to absorb their food and avoid intestinal cancer isn’t going to teach anyone anything except that they should stop listening to you.”

It always makes me happy to see researchers studying who a drug helps, rather than looking at the population average.  This week it’s a suspected link between blood pressure and the efficacy of a particular PTSD treatment.

Many software teams operate on “you were the last person to touch it, you own it.”  Sometimes this makes sense (it prevents people from checking in cheap hacks and leaving others to deal with the consequences), but it also engenders a learned helplessness, because you can’t make one change without becoming responsible for every bug and feature request until someone gets desperate enough to make one little check in, at which they become responsible for every bug and feature.  It engenders a kind of learned helplessness that encourages hacks rather than digging in a fixing a problem.  The Copenhagen Interpretation of Ethics is this problem, but for altruism..

When genes fight

For those of you just tuning in: on a genetic level, all parental investments in their offspring are equally valuable.  From the offspring’s perspective investing in them is two to four times as good as investing in their sibling (depending on if they’re half or full siblings).  The fight to get the amount of resources they think they deserve is parent-offspring conflict

One manifestation of parent offspring conflict is weaning conflict, where offspring would like to keep getting nutrients with no effort and their mom would really like them to go out and get a job.

But it can start before then.  Conflict over exactly how much nutrition a fetus should get may contribute to preeclampsia (high blood pressure during pregnancy) and insulin resistance (PDF).  This is not just about mothers not wanting to give up nutrients- more nutrition leads to larger babies, which is almost always good for the baby but kind of hard on the person forcing it out through their vagina.  There is even speculation that the human custom of making a thick uterine lining (nutritionally expensive) only to flush it away each month evolved as self defense against a placenta that would otherwise invade your uterus like ivy invades bricks.

This is your uterus on babies.
This is your uterus on babies.

Creepy, yes.  Abhorrent, at the level of individuals.  But totally logical and predictable from the level of a gene.

Kin Selection

The idea of kin selection was implicit in my post on haplodiploidy but let’s make it explicit.  The unit of selection is the gene, not the individual.  The individual is a co-operative venture by many genes to reproduce themselves [I attempted to explain why this was and it became 4 paragraphs on the origin of DNA, so let’s just take it as a given].  There’s no reason for genes to prefer directly reproducing themselves: if you can get more copies of yourself by helping someone else’s co-operative venture than your own, that’s a better investment.  Doing so is known as kin selection.

The most obvious example of this is parental investment in offspring.  Offspring aren’t you, but they have your genes.  For a diploid sexually reproducing organism, a given allele of yours has a 50% chance of being in your offspring via common descent (sharing an allele via coincidence doesn’t count for reasons we will get to).  But a full sibling is just as related to you as your child, so raising them is just as good.  The technical term for this is helpers at the nest*.  It’s especially likely when raising offspring is exceptionally costly and resources (e.g. territory) are very limited, so the choice is between raising siblings or nothing, rather than raise siblings or raise your own offspring.

As you might guess from the name, helping at the nest is most common in birds, but you do see it elsewhere. Golden Lion Tamarins live in groups of 2-8, but will usually have one, with a maximum of two, breeding females.  Females are unable to provide sufficiently for their offspring on their own.  They’re helped out by other group members, which are likely to be their own children, siblings, or sibling’s children.

Honestly I have no idea if that's a helper or the parent but it's adorable.
Honestly I have no idea if that’s a helper or a parent but it’s adorable.

Nest-helping may be at intermediate step between the “good luck, fuckers” school of parenting (technical term: r-strategist) and eusociality.  For example Carpenter Bees (carpenter is a genus, there are 500+ species within it) are usually solitary, but some build nests near each other in trypophobia-inducing pattern,. show specialization and cooperation between nesting adults (e.g. one guards all nests while others forage) , and daughters sometimes share a single nest with their mothers.

Don't trust pictures in nature articles. I couldn't find a picture of cooperation so I used one of fighting.
Don’t trust pictures in nature articles. I couldn’t find a picture of cooperation so I used one of fighting.

One way to help your own siblings is to raise them, like helpers at the nest do.  Another is to free up parental energy by taking less for yourself.  From a genetic perspective, you should stop asking for things from your parents if the energy would benefit a full sibling twice as much, or a half sibling four times as much.  But from the parents perspective children are all equally valuable, so they will want to switch giving as soon as resources benefit one child more than another.  This is parent offspring conflict, as explained by noted evolutionary biologist Dylan Moran at 30:00 in this lecture:

More on parent offspring conflict tomorrow.

*Note: kin selection is not necessarily the only reward for helping to raise siblings; individuals may also learn parenting skills or give themselves a leg up claiming their parents’ stuff when they die.

Haplodiploidy

Let’s start way at the beginning.  All eukaryotic cells (which includes any multicellular organism) carry their DNA wrapped up in a chromosome

2000px-Chromosome.svg

In any given species the chromosomes are numbered, and chromsome N contains predictable information.  For example, the human chromosome 16 contains the DNA to alpha-globin, a component of hemoglobin, the thing that lets your red blood cells carry oxygen.  Some people have a variant in their alpha-globin genes that leads to sickle cell anemia.  It’s still chromosome 16.  The general location and form of the DNA that produces a protein is the gene, different variations are called alleles.  So technically it’s wrong to say “the gene for sickle cell”, you need to say “allele for sickle cell”, but everyone knows what you mean.

Many organisms contain more than one version of their chromosomes: the second (or more) chromosome has the same genes but different alleles (unless something goes weird, which does happen but we don’t have time to get into).  You are probably most familiar with the human chromosomes: 2 versions of 22 normal chromosomes, an X chromosome, and either an X or a Y chromosome, with Y chromosomes conferring maleness.    Having two versions of each chromosome is referred to as being diploid, and it’s not the only choice.  Certain sugar cane hybrids have as many as 12 copies of the same chromosome.  Some species show variation in ploidy between individuals.  This is more common in plants, which can self-fertilize, but is also seen in insects.  In humans individual chromosomes are occasionally doubled when they shouldn’t be: this causes death if it’s a big chromosome and things like Down’s Syndrome if it’s not.

Note that the Y chromosome does not contain all the information you need to be male: it releases the signals to be male, and numerous genes on multiple chromosomes respond accordingly.*   That’s not the only way to determine sex.  Birds, some fish, some reptiles, and a few others species use the ZW system, which is the same as XY except females are the one with the Y chromosome.  Many reptiles sex depends on the temperatures their egg experienced (and not every species responds to temperatures the same way).   We don’t know how sex is determined in the platypus, which, yeah, that’s about what I expect from the platypus.

Then it gets weird.  Fruit flies, and have the equivalent of X and Y chromosomes- but they have anywhere from one to four versons of sex chromosomes, and 2-4 versions of each non-sex chromosome (autosome). All autosomes have the same number of versions, but there may be a different number of sex chromosomes.  The sex of a fruit fly depends on the ratio between the # of autosome copies and sex chrosomsome copies.

But what I really want to talk about (600 words in) is haplodiploidy.  A haploid cell has only one version of each chromosome, a diploid two.  In haplodiploidy, females are diploid, and both male and female produce haploid gametes (egg and sperm).  Unfertilized eggs grown into males, and fertilized eggs grown into females- so females have twice as many chromosomes as their brothers.  Males have one grandfather and no father at all.

This has a couple of implications.  One, most deleterious recessive genes get weeded out right quick, because no male will produce them.  On the other hand, such a gene could persist if it boosted female reproduction sufficiently.  Second, females can reproduce without males, although they will produce only sons.  Which they can then mate with (and inbreeding isn’t nearly as dangerous as it is in diploid animals, because of the aforementioned filter on negative recessive genes), and produce daughters, so a single female can repopulate the planet.

Then there’s relatedness.  A human is 25% related to a half sibling (for any given gene there’s a 50% chance it came from the shared parent, and a 50% chance the parent passed it on to the sibling).  But a haplodiploid father passes on the same genes to every child, so each sibling is 50% related to each other through him.  If the siblings also share the same mother, they are 75% related to each other.  That is more related than they could be to their own offspring (50%).**

It was initially thought that this was why/how eusociality developed: it was literally more advantageous to raise sisters than daughters.  That’s not strictly true:  There are haplodiploids without eusociality, and strict diploids with eusociality.  Some  eusocial haplodiploid queens breed with multiple males, so their daughters are raising sisters only 25% related to them.  But eusociality is heavily overrepresented in haplodiploid animals, so it clearly affects the math.

*Fun fact: the signal for male development in utero is not exactly the same as the signal for male development at puberty, and it’s possible to be unable to produce the fetal signal but successfully produce the puberty signal, producing babies that are born with female external genitalia but grow a penis and testicles at puberty (nearly all of these children identify as men as adults).  This was common enough in certain villages in the Dominican Republic that they have a name for it, which translates to “penis at 12.”  It’s considered a joyous thing because sons are more valued than daughters.

**But aren’t their sons 100% related to them?  Yes.  But relatedness is not necessarily reciprocal.  A gene in a female has only a 50% chance of being in a given son, so she is only 50% related to him.

Eusociality/How to Train Your Dragon in Behavioral Ecology

Like many people who know things, I often find inaccuracies in movies frustrating.  I’ve learned to let this go in most instances, but I still have a weakness around behavioral ecology, possibly because I spent so long studying it and use that knowledge so little now.  This week’s victim is How to Train Your Dragon (1 and 2).  The questions I want to answer: how/why are they so many different types of dragons, and what is their social structure?  The ecosystem as described by characters makes so little sense I’ll never come up with a plausible system that makes their statements true, so I’m going to focus on generating a system that could generate their observations without being constrained to make them actually true.  For example: real species always have variation between individuals, even if they’re all clones (because of environmental variation). Vikings indicate all dragons of a given type are identical.  So I will design dragons that have little enough variation that Vikings could plausibly mis-measure them as having none, but not actually none, because that would be dumb.

I’m also just going to ignore the fact that that ecosystem couldn’t support that many predators of that size, and the whole fire breathing thing, because those are just things you accept when you watch a movie.

Here are the observations I need to explain (enormous spoilers for both movies, although mostly not the parts anyone else cares about):

  1. There are a lot of different types of dragons that live together and appear to work cooperatively.dragons_crowd
  2. Nests with lots of dragons tend to be controlled/led by a single enormous dragon.
  3. Movie 1 had a queen dragon and Movie 2 had a king dragon.  Valka says that kings outrank queens.
  4. The nest with the queen (a Red Death) was in a volcano, lesser dragons hunted for her, and she ate them if they failed.reddeathfacereddeath-full
  5. The nest with the king (a Bewilderbeast) was located in an iceberg, and he fed the other dragons.bewilderbeast
  6. There was a second collection of dragons led by a different Bewilderbeast under the control of a human that we don’t know much about.
  7. Dragons react to human speech at a level that indicates understanding, not just keyword matching.  They pick this up without formal instruction.
  8. The kings are shown giving sophisticated commands to lesser dragons.  The queen did not do so on screen, but that may be how she compelled the lesser dragons to hunt for her.
  9. The bewilderbeasts breathe ice and can stay underwater indefinitely.  They can survive on land but do not fly.
  10. A number of underwater dragons are mentioned.
  11. The Vikings believe very strongly that dragons are groupable into distinct species where members have identical stats.  They refer to these as species, although it’s never mentioned how Vikings define that.
  12. Dragons vary enormously in size.  Bewilderbeasts are 520′ x 160′.  Terrible Terrors are 1.5 feet tall, and judging by their behavior, adults.
  13. For all their morphoological diversity, we see dragons doing mostly the same things.  This could be an artifact of their interaction with humans.
  14. Dragons appear to congregate in massive nests.
  15. There is some variation in morphology even among juvenilesDragon_hiddenability_baby

The mass nesting and single head sound a lot like bees/ants/termites, where thousands of sterile workers coordinate to support a single breeding individual.  The technical term for this is eusocial.  Eusociality was something of a puzzle to evolutionary biologists for a while because wait, individuals are not breeding?  How could that ever be selected for?

The answer is kin selection: workers bees are able to pass more of their genes onto the next generation by caring for eggs the queen lays than they would by raising their own offspring. In the particular case of Hymenoptera (bees, wasps, ants), this may be helped by the fact that females are more related to full sisters than they are to their own offspring (I’ll explain this tomorrow), but this is neither necessary nor sufficient to generate eusociality.   Eusocial animals have some variation in their breeding structure: ant queens mate once and store sperm for life, males die shortly after mating.  Termites, certain shrimp, and  Damaraland mole rats have a single breeding pair.  Naked mole rats and some bees have a queen and several breeding males.  Bees and wasps can have either either the first or third structure (and not all bees or wasps are eusocial).  However, and this is important, there’s no known (non-fictional) eusocial species with one male and multiple breeding females.  There are animals with a harem structure (e.g. gorillas, lions), and females may have some cooperative care (e.g. lionesses will nurse others’ cubs), but children are cared for primarily by their mother and immature siblings, and there are no permanently non-breeding females.

The sheer number of dragons, the size differential, the fact that they cooperate to feed the alpha (sometimes), and that they can be compelled to take orders points to eusociality.  However eusocial animals are usually pretty dumb.  Dragons are smart, and have relationships outside that with their alpha (e.g. Toothless trying so hard to impress Valka’s dragon).  This suggests they have a more complex social structure than a beehive.  Moreover, they actively work to understand what humans want and to do it: this suggests they’re something like wolves, where there is a complex hierarchy beyond a simple alpha, and that like wolves dragons have applied their ability to interpret intentions and respond to them to us.*  This doesn’t necessarily mean they view humans as alpha, they may view us as a senior pack member to be listened to.

If they are eusocial, they’re going to be more like naked mole rats (where any individual can theoretically become alpha) than insects (where alpha-ness is designated at birth).  I believe this because we see three different dragon types as alpha (Red Death in the first movie, two Bewilderbeasts and Night Fury in the third), and because alphas obviously retain the capacity to be ordered around (by both humans and other dragons).

That’s social structure.  What about speciation?  We see an enormous variation in dragon morphology and behavior in the movies.  There’s a few ways to generate that:

  1. There are many closely related species, like whales.  This does not gel with the fact that dragons are clearly cooperative. While different species sometimes cooperate (e.g. humans and dogs), there’s nothing approaching interspecies eusociality.
  2. There are very different species with tight, possibly obligate, mutualisms (humans and dogs were probably this once, and gibbons and gazelles have some mutualism as well).  But those species tend to be good at very different things (dogs and gazelles have smell and we and gibbons have eyesight), while the dragons seem to mostly do the same thing.  This could be sufficient to explain them nesting together, but not following the alpha.
  3. They could all be the same species (meaning they can interbreed), but have diverged into semi-distinct genetic pools, like dogs.  Rare types could be hybrids. If hybrids are frequent and distinct enough to be recognized as their own type, it suggests morphology is controlled by only a few genes, or that all the genes are located on the same chromosome.  Otherwise you would just have a bunch of dragons that were intermediate between their two parents.
  4. Dragons are born having the ability to take many forms, and move through multiple forms in their lifespan or settle on one form based on the environment. This would be a good fit with eusociality, where drones often specialize in a single task.
  5. Dragons vary a lot and the Vikings more or less made up categories to shove them into

I suspect there’s more than one base form, because the babies in HTTYD2 were already bigger than the smallest adult dragons we’ve seen. But the different types must be able to interbreed, because different species don’t cooperate that extensively.  This points to 3 and possibly 4.

I also have to explain the feeding.  Eusocial alphas get fed, they do not feed their drones, which doesn’t match the behavior in movie 2.  The easy answer is that it depends on environmental conditions: when food is plentiful the alpha feeds their minions, when it’s not their minions feed them.  Since the sex making the larger investment in offspring is usually the one fed by the other, this suggests alphas are male in the water and female in volcanos.  Normally that would make no sense, but I’m about to get to a really satisfying explanation.

Let’s talk about reproduction.  Given the babies in movie 2, alphas must reproduce with members of their own nest, not an alpha from another nest.  All the examples I can think of with a single alpha that have that level of control (as opposed to a dominance structure with one member at the top) are eusocial and have a female leader.  This gels with the minions feeding the alpha in movie 1.  Or they could be like gorillas, with a single male keeping a harem of females.  That matches well with the fighting between the two alphas and the loyalty shifts, plus Valka would probably notice if her king laid hundreds of eggs.  It seems to me to be incredibly unlikely a species would have the ability to have alphas of either sex, so we have to choose one or choose hermaphroditism.  There’s only one reptile that could be called hermaphroditic, and that’s really more like being intersex than a true hermaphrodite.

But… hermaphroditism is really common in fish.  Fish also feature more morphological variation through their life than reptiles, and have more and more complex social/pack behavior.  And we know there are underwater dragons.  What if those came first, and they moved out of the sea later?  The intermediate step would look a lot like flying fish, which totally exist.  I’m not sure about the fire breathing thing, but it’s not like assuming they’re reptiles makes that so much easier to explain.  Dragons eat fish a lot, even the terrestrial ones.  While I said I was going to ignore the inability of that environment to sustain that many predators, the ocean comes closer to meeting their requirements than land. And it explains the variation in feeding behavior.

In conclusion:

  •  Dragons are descended from fish, and have a highly cooperative harem structure when food is plentiful and a eusocial one when it is not.
  • The alpha is a hermaphrodite who leans male when food is plentiful and female when it is not.   Either minion dragons are also hermaphroditic, or male and female forms exist and whichever complements the alphas goals gets to reproduce.
  • There may be other dragons that get an occasional shot at reproduction, who may or may not be considered part of the pack (a la the side-blotched lizard or marine isopods).
  • There are probably a lot of non-breeders of either sex, or dragons stay sexually immature until promoted by an alpha.
  • A given baby dragon can turn into many (but not all) kinds of dragons, and may be more than one over its lifetime.  Minion dragons are extremely intelligent and can understand complex instructions, but have limited ability to talk back.  They can be compelled to follow these against their will but also have a strong desire to follow them without compulsion.
  • Not all dragon types can be alphas, but more than one can.

As for the movie plots… yeah, they were pretty good.

*Wolves are obviously not dogs, but they’re much more trainable than other animals.

Links 6/19/15

Gross things.  What creeps me out is the branching.

Against tulip subsidies.  This covers two of my biggest fears with charity: that I’m reinforcing a bad system, and I’m redistributing wealth rather than generating it.  Note that an RCT won’t catch those unless you think to look for them.

How they genetically engineered bacteria to make cheese vegan.

For a change this week we will have an “Ugly theory killed by a beautiful gang of facts” award.  Anyone who’s taken psych 101 knows about the Zimbardo experiment (aka the Standford prison experiment, which proved all humans will descend into savagery given a modicum of power) and Asch’s conformity experiment (the line length experiment, which proved people will always choose agreeing with a group over the actual correct answer). Except… Zimbardo deliberately induced sadism in the guards, and 2/3 resisted it.  And more people in Asch’s study showed total independence than total conformity.  Somehow over the years these got severely exaggerated- not just in the popular media, but in textbooks.

I noticed this when I started checking citations in pop science books.  “People with trait A do B and people with trait X do Y” usually turns out to be “there is a 10 percentage point difference in how often people with A do B, relative to X.”

For another example of how to lie with statistics: “80% of Americans support labeling food with DNA”.  Which if you actually look at it shows that 80% of people believe in labeling and weren’t too hung on the specifics by the time they got to that part of the form.

Effectiveness Updates

Suicide Hotlines: One of the reasons I estimated crisis chat/suicide hotlines as having as high an impact as I did was visitors’ self-reports.  Since then I saw a discussion of leafleting on FB, where several people said they had high priors for it working because when they leafleted, some people seemed really interested and said they were going to change.  My reading of the quantitative research is that there’s no proof leafleting has any effect, so I should discount my own estimates of crisis chat.  I also completely failed to account for the damage a bad counselor can do.  I found one metastudy on the effect of suicide hotlines, and it appears debatable they’re accomplishing anything, much less have a good cost:benefit ratio.

I may also be pessimistic because I had two people initiate attempts while they were talking to me in a month.

Blood donation: I was hoping to talk to the actual blood bank, but they’re no longer returning my e-mails and this has gone on too long. I stand by the calculations for average effectiveness, but after a discussion with Alexander Berger I am retracting the claim for marginal effectiveness.  New donors can apparently be recruited rather cheaply.  There are concerns that the blood is more likely to carry infection (not all of which can be caught by testing), or that incentives will crowd out more altruistic donations, or that the marginal cost will grow with time, but these have the whiff of valuing moral purity over results.  You could convince me otherwise, but at this point it needs data.  On the other hand, blood donation may have health benefits, especially if you don’t menstruate (thanks to Kate Donovan for pointing this out).