Post has shared content
Fun story: I used to work upstairs from Jean'ne Shreeve's lab, which was notable because we would regularly hear suddenly breaking glassware followed by shrieks of delighted laughter.

I should note that piling up giant stacks of nitrogens and then hanging hydrogen peroxide off of them is her sideline, not her main research area. Her main research area is taking fluorine and attaching it to things which would very much not like to be attached to fluorine, the results of which are usually spectacularly exothermic. Especially when you are attaching fluorine to unstable radionuclides in an attempt to separate them by weight.
You may not know that there's a rich tradition of chemists writing, shall we say, rather bluntly about their trade. And one of the kings of it is Derek Lowe, a drug discovery chemist who writes a semi-regular blog titled "Things I Won't Work With," primarily about the research that people in other branches of chemistry do that makes him question their sanity.

While it helps to know some chemistry to follow what's going on (that e.g., most of the molecules in your body use carbon for their superstructure, because nitrogen in the superstructure tends to want to get out of said superstructure rather quickly, which is to say "with an earth-shattering kaboom"), you don't really need to:

"If you or I (’cause we’re sensible, right?) look at a well-known crater-maker like dinitropyrazolopyrazole, we’ll probably decide that it has pretty much all the nitrogens it needs, if not more. But that latest paper builds off the question “How do we cram more nitro groups into this thing?”, and that’s something that wouldn’t have occurred to me to ask. Saying “this compounds doesn’t have enough nitro groups” is, for most chemists, like saying “You know, this lab doesn’t have enough flying glass in it” – pretty much the same observation, in the end."

I should also say that Lowe is the person who introduced me to John D. Clark's classic textbook of the history and practice of liquid rocket propellants, Ignition!, and if this sort of writing at all appeals to you (or if you were just always curious about what kinds of things can cause you to accelerate away from them at remarkable speed), then you should dig up a copy as soon as possible.

Thanks to +Amber Yust for finding and sharing this latest gem in Lowe's collection.

(Bonus: If you go to the homepage of Prof. Shreeve, lead author of the "more nitrogen!!!" paper above, you will find someone who you might mistake for a kindly librarian if you passed them on the street. This is someone who is a distinguished professor of Materials and Fluorine Chemistry, a title which alone will cause most chemists to look for some convenient large object to hide behind.

https://www.uidaho.edu/sci/chem/people/faculty/jshreeve)

Post has attachment
Blurgh, this is an awful, awful article. A few points:

(1) This is somehow derived from a single 1996 paper in the Lancet, which found that several genes which cause intellectual disability are on the X chromosome. Those do, in fact, have huge effects on intelligence -- Fragile X syndrome occurs only in boys. So there's something to it.

(2) More recent research has determined that there are two large clusters of genes -- M1 and M3 -- involved in intelligence, and that those genes are spread across both the autosomes and the X chromosome. M3 has about 150 genes in it; M1 has fewer.

(3) For each of those genes, there are significant numbers of single-nucleotide polymorphisms that produce viable results. Those are equally likely (well, except for those on the X chromosome) to be inherited from mothers or fathers.

(4) If it were the case that many more genes were inherited from the mother than the father, you'd end up with two facts which don't yet seem to be in evidence: (a) that men have more variable intelligence than women, and (b) that women are, on the average, more intelligent than men.

The first is because men would be more likely than women to inherit a full, functional set of positive alleles; the second is because women are more likely to inherit a variable set of alleles, and that rare alleles are more likely to cause intellectual deficits than common ones.


Post has attachment
This is precisely the article that I've been waiting for. A couple of notes:

(1) Letting people know before exposing them to something a reasonable person would find disturbing or upsetting is polite. There's some reasonable quibbling around the edges -- basically, about how sensitive you have to be to context when doing so -- but the arguments against this very basic social rule are essentially bad-faith. If you are going to give a graphic account of a sexual assault, or will be doing a surprise autopsy in front of a class, you should probably warn people that that's what you're going to do.

(2) This doesn't completely absolve the people who are the most avid advocates of trigger warnings. The argument from that corner is that failing to provide trauma triggers creates real, long-term medical harm for people who have PTSD, and that what will trigger breakthrough panic attacks in PTSD is predictable.

In short, no. Most people who suffer substantial trauma will not suffer a trauma disorder; most people who do have those disorders will spontaneously remit; the only thing strongly correlated with failure of trauma disorders to remit is avoidant coping strategies. In other words, telling people that trauma is always (or often) a permanent psychic injury will often result in there being a permanent psychic injury.

(3) One question which I frequently got, working as a DV/SA social worker, was "Will it be like this forever?"

No, maybe. Yes, maybe. What do I know?

The answer I always gave was, "This is the worst thing which has ever happened to you. It is only the worst thing which has ever happened to you. Everyone recovers at their own pace. Some people never recover. Some people need help to recover. Most people eventually do."

What I would frequently hear -- not from coworkers, but from other people in activist spaces, was "Rape destroys the soul." I heard that phrase over and over and over again, and while it captures something about rapists, it tells victims that everything is hopeless, nothing can help them recover, and that they have suffered an irreparable psychic injury which will follow them their entire lives.

There is a very narrow rapid between denying trauma and accepting trauma as an overarching identity; one which instantly overwrites all other possibilities. Supporting the self-efficacy of people in the aftermath of trauma without making rigid demands for healing is critical to the process, and ducking behind victims of trauma in order to smother them with accommodations they (as a group) didn't ask for is ... problematic at best. 

Post has attachment
This result is... unexpected.

Post has attachment
Oh dear. No, this is not good. Some explanations of what's going on here:

This is a trial of chimeric antigen receptor-modified T-cells. The theory behind this is relatively simple: you take a monoclonal antibody which targets a particular cancer; you remove a host's own T-cells; you modify the T-cells to present the monoclonal antibody. Then, so the theory goes, your own immune system targets the cancer. And it seems to work. In previous trials, all of which were small, this therapy has flat-out halted treatment-resisted acute lymphoblastic leukemia.

In other words, it took terminal patients and put them straight into remission. But this latest trial? It took terminal patients and killed them. And this seems to be for one of two possible reasons:

(1) CAR-T modifies T-cells to react to human tissues. Now, cancer is a misbehaving human tissue, but it's still close enough in almost every regard to human that your immune system is likely to cause some collateral damage on the way to killing the cancer that's killing you. Previous trials have shown cytokine storms (i.e., "total, global immune freakout") to be a possible side-effect of CAR-T.

And if that storm of cytokines includes, for instance, IL-1β, your immune system now has a universal passkey to let itself across the blood-brain barrier. Resulting, possibly, in cerebral edema and then death.

This would be extraordinarily bad, because the mechanism of action also causes the side effect.

(2) In order to get your modified T-cells to take root, your oncologist first has to ablate some of your natural immune system. This is especially the case in leukemia, where immune quorum-sensing tends to perversely cause immunosuppression: your body thinks that it has too many immune cells (it does, but broken ones), and so it stops producing good ones.

To get around this, CAR-T first has to burn your immune system to the ground. And this is where things get weird.

Because your body is full of latent viruses. If you're like most people, you're constantly dealing with low-grade viral encephalitis. The JC virus is on a slow burn in your brain, infecting oligodendrocytes through one of their serotonin receptors. Since your body can produce new oligodendrocytes, and your brain's innate immune system can clear out JC virus faster than it can spread, you'll be just fine.

Until you're myelosuppressed.

Once you are, your brain's terrible, low-grade immune system can't contain the JC virus. In a single flash-fire infection, it burns out most of your oligodendrocytes. And you're fine, for a couple weeks, because -- in doing so -- it didn't destroy the infrastructure of your brain. But you're soon going to need myelin. And the cells that make it are dead. Which means that you're dead.

So it's possible, if this isn't an acute immune reaction, that CAR-T would work fine, so long as you don't burn the immune system down in the process. But I'm not quite confident of that. 

Post has attachment
Two interesting facts, when in conjunction:

(1) For the past ten years, it's been believed that the NMDA receptor system, which largely handles central pain response, is deeply involved in depression. Why? Because ketamine, a NMDA agonist, was extraordinarily effective at treating depression.

Very recently, however, irregularities in the dose-response curve for males and females led researchers to a surprising conclusion: one of ketamine's metabolites, an AMPA agonist, is actually responsible for it's antidepressive effect. So it's quite possible -- even probable -- that pain response and depression are only loosely coupled.

(2) Childhood trauma correlates very strongly with chronic back pain after acute back injury. In fact, almost nothing correlates as strongly with chronic back pain as childhood trauma. This seems to indicate that there's some kind of important backdoor from emotional trauma into pain processing.

Ten years down the road, it's going to be interesting to see what shakes out. 

Post has attachment
A good 90% of my success has been the result of a genetic accident, no fairer than if I'd inherited an enormous trust fund from my parents. Intelligence is useful. It's a tool. It's not a measure of human worth.

And I'm unsure why, at this point in history, people are starting to believe this. "I deserve the lion's share of our society's resources because certain things which are easy for me are difficult for others" has never been a particularly compelling argument. If this was believed to be true five hundred years ago, then oxen would be entitled to most of the agricultural output of Europe.

Post has attachment
I've been saying this for years: we do not have to worry as hard about storing energy as we think we do.

Most of the discussion I've seen about how to deal with intermittent renewable power storage has been as resolutely high-tech as the renewable energy sources themselves. Massive flow batteries. New types of lithium-ion cells, and where are we going to get the lithium to do it? Self-driving electric cars being used as on-grid storage. Exotic molten fluoride salts being pumped into huge thermoses. Massive vacuum flywheels running at fractions of the speed of light.

That's all nice. Maybe some of those technologies will turn out. But we don't need them. The solution isn't to think smarter. The solution is to think dumber.

We already deal with storing renewable power to normalize energy supply to meet energy demand. In hydroelectric power, the store of energy is literally the reservoir: we take the water we have, and run it through at a consistent rate. For other types of renewable energy, we take water and pump it uphill, then run it through a dam to get the power back out.

Here, we think even dumber than that. This energy storage is literally just moving rocks uphill. You can't get dumber or more scalable than that.

Some Things I Noticed About Echopraxia, in No Particular Order, But With Spoilers

First: it is a tough read. The book makes very clear that the narrator does not understand what's going on. Ever. Things happen, alliances change, or don't, and people die, or don't, and nothing is wrapped up in a neat little bow. You are expected to read the references at the end, and if you don't, you are likely to be lost.

Second: it displays enough fine structure, on initial reading, that it doesn't seem to be lazy authorial nonsense. It's making an argument, but you've got to slog through three hundred-odd pages of academic reading in a half-dozen fields in order to understand what it is. Because it's a very complicated, very weird argument that I've never heard before: it's an atheist argument about the existence of "God" in a way that's sort of Spinoza-with-indirection; it's an argument that consciousness does not exist in the way we think it does, but maybe it still does in a way which we can value; it's an argument about the iron mathematical laws of game theory and evolution, the inability of any living being to outrun them, and a glimmer of hope that they can still coexist with what makes us human.

It's an argument that humanity can't leave biology behind. The best we can do is to create biology which contains us in some sense.

On to the spoilers:

* One of the main plot points in Echopraxia is the discovery of a signal from Siri, the main character of Blindsight. The signal appears to be a narration of the events of Blindsight.

Given the events of Echopraxia, here's what's probably the case: Siri is dead, and the events of Blindsight are a lie. What remains of Siri is running in one of the severed lobes of the MPD linguist, who is the real "survivor" of the events of Blindsight. Remember, the voice coming from "Siri's" coffin is, after decryption, probably female.

And there's only one female member of the Theseus team that's specialized in simulating the cognitive processes of other sophonts -- the "imagine you are" language is likely that person trying to simulate Siri's cognition for some unclear purpose rather than Siri himself trying to induce a third party to imagine his cognition.

* The name of the ship from Echopraxia -- "Crown of Thorns" -- is not a religious reference. It's probably a reference to the Crown of Thorns starfish, which has an appearance very close to the Scramblers from Blindsight. This is either (a) a clever non-diegetic wink from the author, which I find unlikely, or (b) a punny implication by the Bicamerals that they know about Rorschach, and by further implication what they're likely to find on Icarus.

* As the book makes clear, Bicamerals cannot envision their own death. This is possibly because they cannot actually die; every individual is backed up to the hive-mind before the death of their individual body. They do not lack individuality. They simply lack individuality at the level of abstraction which all material things in this universe can detect.

When the book talks about Heaven, and God, and a number of other things, it is implying that this universe -- the one we live in -- is itself a delegated sub-process of a greater computational whole. Intelligences can be conscious at this level of abstraction by being integrated by an abstract non-sentient net at a higher level of abstraction, or by emergently achieving consciousness at this level of abstraction.

* Portia is covered in little pimples which Brueks thinks are sporangia. They are. Each Scrambler grows in stacks, like jellyfish medusae, and they pile up until the ones at the end fall off and start scrambling. If you look at the slime-mold-like structure of Portia and the basket-star-like structure of the Scramblers, they're both just different elaborations on a single ramified quasisymmetric body plan.

* The Bicamerals wanted to be infected by Portia, but not because they wanted Portia to win. They wanted to be infected by Portia because integrating with Portia is the sole effective method of understanding Portia, and by extension Rorschach and the entire alien biosphere. In other words, they were not self-annihilating. They were descending into a common level of abstraction shared by any hypothetical recursive (that is, conscious) subprocesses within both nonsentient nets.

This allows them to do something that all parties intend. Something extremely complicated, which I'll get to later.

* It's not made explicit, but what happens when Valerie and the Bicamerals agree to fight a third force? Bayesianism. Aumann's Agreement Theorem implies that rational entities cannot disagree. If rational entities disagree, it can only be because of incomplete information or defects in their rationality.

Valerie begins by attempting to destroy the Bicamerals. The arrival of a third force -- Rorschach or some extension thereof -- is the information which the Bicamerals had but which Valerie was lacking. They both update their priors and decide to attack the third force. Later, the Bicamerals and Valerie seem to become compromised by Rorschach and start trying to move Portia back to Earth, despite having destroyed the station it lives on.

Everything which happens aboard Crown of Thorns and Icarus is a bit of Kabuki. The three involved parties do not hate each other, and they are not opposed. What's going on is actually extremely complex:

(1) In Blindsight, the crew of the Theseus comes to the conclusion that the Rorschach biosphere regards communication as war, because attempts to communicate waste its energy on listening to what we regard as an attempt to exchange thoughts, and what it regards as a malformed control flow. It does. But this isn't the end of it. Because of Aumann's Agreement Theorem, the Rorschach biosphere regards war as communication, and trade as war, and communication as trade, in perfect symmetry. We do not, but only because we lack perfect isomorphisms between information, matter, and energy.

Once it has recognized a fitness-optimizing process (i.e., extended post-humanity), it will attempt to come to the resolution which optimizes the mutual fitness function. This will very seldom entail the utter destruction of the other party. And in this case, it probably doesn't.

It is not conscious. It is extremely, extremely smart, on a scale not comprehensible by humans. If it wanted to destroy posthumanity, it could, but it would be a tremendous waste. It wants to understand what our fitness function entails, and then mutually optimize both fitness functions, first by assimilating enough sophonts to make it clear what's going on, then by coming to some mutually agreeable solution or destroying us.

To anthropomorphize, it wants to destroy humanity and intends us no harm, and considers there to be no incompatibility between the two.

(2) They all want to control Dan Brueks. There's an argument between people about who, at the end, actually does. The question, I think, would be incoherent to the parties involved.

Valerie wants Dan Brueks to bring back a Portia variant which cures vampires' genetic inability to cooperate. Rorschach wants to integrate all of posthumanity into an energy-minimizing superorganism. The Bicamerals want to maximize understanding of the universe. Like bacteria exchanging plasmids, they all exchange fitness functions through an obfuscated channel.

They do this by predicting each others' moves. But why can they do this, despite the fact that they seem to be superintelligences on vastly different scales? Because being utterly transparent to other rational entities is the most effective means of minimizing computation time, and thus energy.

Dan Brueks, at the end of the book, is not a discrete agent of any single one of them. He's an agent of the joint fitness function of all three actors, on a mission to integrate lower posthumanity despite its irrationality. He is the same thing to Rorschach that Lianna was to the Bicamerals: a hyper-jargonaut, able to interact directly with lower posthumans on behalf of superintelligences in the same way that Lianna could deal with humans on behalf of lower posthumans.

* Similarly, and this is not made explicit, what's going on with Dan Brueks? This is a relatively simple answer too, and it's a result in computational game theory: any player with a sufficiently long decision horizon and the capacity to effectively model the internal processes of its opponent, can spoof any particular fitness landscape it prefers. More generally, any player which is omniscient relative to any other player, in game theory, utterly abrogates the free will of that player.

Dan Brueks believes that he is outside the system which is making decisions. This is incorrect. He's had his entire fitness landscape replaced by a set of omniscient players, and is, for all intents and purposes, a strict subprocess of those omniscient players. He does not have free will.

When he commits suicide, even that is not of his own free will. It's the decision that the omniscient players made. In order to spread Portia. In order to subsume the entire human race into the Rorschach biosphere. In order to prevent a war which would waste the energy of both, which would not maximize Rorschach's fitness function. Which it is incapable of doing. 
Wait while more posts are being loaded