After all the work they've put in, they now have control of enough levers for the patient's various bodily processes that nothing is going to go too wrong in the next few hours. 

Somewhere in the background, while all this is going on, a minor horde of high-ranking programmers and prediction market experts are frantically scurrying around, trying to get out ahead of testing the new untested prediction-market trading backend, turning up bugs before they explode - now with 70% less statistical learning, but still enough statistical learning that advice propagates in three seconds instead of ten seconds.

Among the acerbic comments being exchanged on backend, where nobody in the operating theatre can hear, are comments to the effect that statistical learning is sort of like summoning a deranged nonreductionist alien-spirit into your system: it will serve you faithfully until it stabs you in the back.  And the entire invention of statistical learning was a bad idea with bad brain-incentives, because it often initially seems to work great and your brain is happy about that.

They're sad about how miserable their fates as programmers are, that they have to get involved with these statistical learning methods that are not always perfectly legible and sometimes produce bugs that are hard to understand even when you're looking directly at their causal antecedents.

They, too, lack context.

Actually in this case, three-quarters of the people on this team have context; but everyone who has context is trying to act exactly like someone who doesn't.

The previous Personnel has now gone off-shift - temporarily, she's left instructions that she's to be called back into the office once this is over or Merrin gets rotated out, even if she's asleep then, because she wants to be the one to congratulate Merrin and also she wants to see the look on Merrin's face about a lot of things.  The new Personnel is named Villar, doesn't know Merrin quite as well, and has been left with a lot of extremely detailed instructions about Merrin-related contingencies; many of them counterarguments Villar is to use if anybody suggests rotating Merrin out before Merrin herself says she's feeling tired or actual performance degradation has been measured.

(The previous Personnel doesn't have a name she uses in-hospital besides "Personnel".  She's an adherent of that subculture which holds that people in real life ought to go by names that correspond tightly to what they are in a social context, like LARPs where everyone is named 'Acerbic Cook' or 'Frustrated Ruler's Daughter' so that players can actually hold all the identities in their head instead of dropping them while they try to associate names to unfamiliar people.  Or as she puts it for hospital purposes:  New employees are going to have a hard enough time remembering new coworkers without asking them to remember that Personnel is called anything except Personnel.)

The Venture Capitalist - whose personal name would be Ashre, if he was asking anybody in this social context already juggling complicated problems to remember him as anything except The Venture Capitalist - has done a lot of organizing and quietyelling and spent really a lot of money, for a plan that prediction markets currently give only a 20% chance of working.

He got his initial set of 20 custom proteins hours ago.  And then his people sacrificed a number of rats, testing the new protocol on them and checking results in lethal detail.  After which they asked for 8 of the custom proteins and 7 more of the off-the-shelf proteins to get replaced - this being when the patient's survival probability dropped from 30% to 20% - and now they've tried the second round on rats and a few precious monkeys.

The basic story seems to be that whether you get a flood of damaged-neuron cell signals (followed in fact by actual brain damage) depends on how much the monkey got cold-damaged beforehand, and it's hard to tell in advance how much damage happened beforehand.  They have debugged every bug about this protocol that is scrutable on a few hours of work by very smart people, and which Exception Handling's mysterious protein-design capabilities think they can possibly fix in hours rather than weeks.

It is obviously not the best possible protocol.  It is not even a very good protocol.  But the patient is not incredibly stable, right now, and doing another design-observation iteration on the protocol would add another 4 hours to timeline and probably not be worth it, all of which is playing out pretty much as the prediction markets expected.  The entire research timeline, at this point, looked like cooling and rewarming some rats while applying 80 proteins and peptides and a handful of lesser chemicals already known in the literature, the rats getting brain damage, writing back for 20 more proteins, getting those, new round of testing, asking to fix remaining errors if possible; there is an old saying in dath ilan that what you can't solve by killing two groups of rats you are unlikely to solve by killing three.  It's usually taken as a metaphor for correcting first-order errors and then correcting second-order errors, but it's also often literally true.

They've killed two rounds of rats and observed half of a third group end up with unacceptable levels of brain damage, they've already run out all their most promising nitwit-ideas and improved-nitwit-ideas on the first two rounds, it's time to roll.

It's not a very good protocol, but it's possible at all because it leans heavily on decades worth of rats sacrificed to narrow in on many - though certainly not all - of the links in the cascade that they're already too late to block at the source, and need to fall back on interrupting at every possible point. 

Reviving patients from temperatures as low as 18 C is actually fairly safe, IF the cooling process is done under controlled conditions - for example, to perform a complicated and risky surgical repair on an aortic aneurism, with an initially stable patient. This is no longer the standard process, since there are now simpler and safer bypass methods to avoid having to disrupt circulation at all, but originally, a patient would be placed on a heart-lung bypass machine while at a normal temperature, and rapidly cooled - with fully- supported circulation and oxygenation throughout, controlled to a high level of precision - until their heart stopped beating, all electrical activity in the brain ceased, and the patient's circulation could be paused for up to an hour, the oxygen already in their blood sufficient to cover any minimal remaining metabolic activity. The survival rate for the procedure itself was 95%, and this was in patients already badly off enough to need major surgery. (Infection risk in the days afterward, due to the suppressed immune activity on top of multiple invasive procedures, was the main motive for figuring out systems that only required cooling to 28 C, if that.) 

Using the exact same rewarming procedure on patients brought in with accidental profound hypothermia to the point of cardiac arrest had, at the time, a survival rate of around 25%. (And those who had dealt with a lack of oxygen or disrupted circulation before they were cold enough for circulation to spontaneously cease still faced a high likelihood of brain damage.) The raw survival rates were predicted to be much higher now, largely because of new techniques that gave finer control of all the other affected variables - electrolytes, liver function, clotting, etc, but this was an estimate based on very few case studies, because most people's medical testaments preferred not to take the risk of further irreversible damage. 

The rats in recent experiments analogous to the patient's history now had decent chances! In a dozen different studies (all repeated by separate research teams using their own methods to check the findings), around 95% survived - even after 10-15 minutes of submersion in ice water before treatment was initiated, and 50% had no measurable signs of brain damage - but, of course, it would be far harder to detect any subtle cognitive deterioration in 'recovered' rats.

None of the individual studies addressed more than fifteen of the pathways now being targeted, and many had been intended to isolate a single enzyme path and measure the effects in isolation of either blocking or amplifying it. The interactions and side effects involved in combining them were as-yet unmeasured and unstudied.  And rats weren't humans. Their small size and resulting high surface-area-to-volume ratio meant that their internal organs, including the brain, would cool much faster than those of a much larger human, and the metabolic suspension likely kicked in before their blood oxygen was depleted.

The team designing the current process was leaning rather hard on the premise that combining a dozen different protocols that had worked okay, in rats not humans, would block a wider range of the cascade, while not interfering with the effect. This premise had never been tested at all, let alone confirmed to a reasonable standard of proof. 

These treatments have never been combined. Fifty-three of the custom proteins have never been administered experimentally to mammals, though sixteen of those have been tried on nematodes.

(C. elegans, with its fully-mapped proteome and conveniently deterministic arrangement of exactly 302 neurons, is an excellent case study for fundamental research - and a surprising number of the metabolic pathways that play out in a human brain are ancient, shared by approximately all animals that have nervous systems. But for all the shared pathways, the nervous system of a millimeter-long invertebrate worm faces a very different set of constraints; biomolecules may be passed on and reused for billions of years, but that doesn't mean that their exact usage stays the same.) 

And, of course, twenty of the proteins had never been synthesized before today; they're for the cases where a given metabolic feedback loop is described in the literature (usually in gene-knockout studies with mice showing how overexpressing or underexpressing a particular enzyme pathway rendered them more or less susceptible to anoxic brain damage.) They were rapidly tested in vitro on a neural cell culture, before the actual rat trials, but there was never going to be enough time to unpack the details of how they behaved differently in a rat brain. 

(The sacrificed rat brains weren't thrown away, of course; some are frozen, some are being plasticized in resin to section and examine under a microscope, and there's more information to be gleaned in future weeks and months - it's just that even with every lab in the world willing to work under a grade three secrecy oath, you can't do that in six hours.) 

Some of the proteins are ones normally expressed in the human body, like the family of superoxide dismutases that break down the highly reactive superoxide radical; an O2 molecule with an extra electron, roughly; back to O2 and a hydrogen peroxide – which is still not a substance one wants in one's brain, and is in turn processed by the ancient catalase pathway into oxygen and water. 

(So many of the biological pathways in all oxygen-metabolizing organisms all the way back to aerobic bacteria are focused on mitigating the damage from the various inevitable byproducts of using oxygen's chemical reactivity for fuel. You start to wonder sometimes if the whole oxygen-breathing evolutionary track was a mistake.) 

And it doesn't stop with oxygen. Nitrogen also likes to form highly reactive radicals, calling for an entirely different set of antioxidant enzymes. These radicals damage lipid cell membranes, dumping an entirely new metabolically-reactive class of peroxidized fatty acid bits, which float in the bloodstream and scream out their own inflammatory alarm.

The RNA-to-protein transcription of proteases (protein-digesting enzymes, which is sort of like cannibalism if you think about it) is upregulated, for some reason that the cellular code leaves entirely undocumented, and start eating the carefully-arranged tight junctions and cellular "feet" that form the blood-brain barrier, letting panicked immune-system cells from the bloodstream ooze through and, in their desperation at the vandalized state of the city, overreact drastically. This process is complicated enough that no one is sure which steps are actually critical bottlenecks – the metabolism likes to have eight different ways to accomplish the same thing, and often, like with anaerobic metabolism of glucose, the fallback ones are even lossier and messier – so they're throwing a dozen wrenches into those cogs, putting rather a lot of hope in 'it didn't kill a C elegans in one study ten years ago.' 

Caspases, sometimes called “Executioner” molecules for their key role in noticing damaged cells and convincing them to commit seppuku, are a very robust pathway - the body is under strong selection pressures to catch DNA-damaged cells before they go rogue and found a tumor colony, and...rather less selection pressure to be conservative about preserving neurons that can only probably recover with further heroic interventions. 

Thirty of the proteins aren't human proteins at all (six were isolated from a single species of hibernating frog.) None of them seem to be actually toxic to rats, but that's about all that's known. 

(Presumably this, as well as the CT scans, is going to end up significantly increasing the patient's later cancer risk, but requiring more frequent and thorough cancer screening for the rest of his life is a relatively tolerable price to pay.) 

Surprisingly quickly in the reperfusion process, it's known that (in rats, at least) the changes ripple up to the level of directly altering gene expression. Among many, many changes, cluster of genes known mainly for their tumor-suppressing effects are upregulated - an excellent idea in theory, but more appropriate in the context of organs that have more redundancy and regeneration ability than the central nervous system. Directly regulating gene expression is a science in its early stages, even in dath ilan, but according to the simplified biochemical simulations, they need to at least try to block it that far upstream to prevent the system, noticing that its usual response is being suppressed, from desperately exploring other side pathways to compensate. 

(The proteins used for this aim are likely to do all sorts of other poorly-understood things, because there is nothing biology loves more than to reuse bits and pieces of four-billion-year-old code in new and creative ways.) 

Most of the uncertainty on success vs failure is already determined, not the result of the team performing worse or better. In the world where the patient was without circulation for ten minutes or less, the success estimate on “no brain damage” rises to 50%, and the remaining probability mass is on outcomes that wouldn’t be acceptable for this patient specifically, but would still include leaving the hospital walking and talking. 

On the other hand, if the interval was more than 30 minutes, the predicted success probability drops to less than 10%. 

They don’t know which world they’re in, yet - in fact, they aren’t likely to find out, unless the investigation into how this happened turns up an exact time for when the patient fell into the river. The outcome of the treatment will give them some evidence, maybe, but not conclusively.

The plan is to have very extensive monitoring in place, including via actual brain surgery, to add a far more thorough suite of sensors across the entire brain. (Some regions, especially the cortex, are a lot more vulnerable to anoxic and reperfusion damage than others, and they want every chance lined up to notice failure as soon as possible.)

From Merrin's perspective, though, this whole desperate Plan of Proteins is going to infringe on her personal reality in the form of an overhead voice waiting until the next time Merrin has a half-minute free, and then telling Merrin to prep for shift handoff to her replacement.  Rewarming protocol is incoming, Merrin's performance has started to degrade a little as they approach the 12-hour mark, rewarming is a new regime that'll take a while... in other words, this is Merrin's best chance to escape.

On the one hand: finally!!! Merrin is definitely starting to feel— not exactly tired, but definitely mentally slower, emotionally out of energy, and frazzled enough that she’s been building up more and more quiet dread at the possibility of anything else going wrong before she manages to hand off. Now seems like an excellent time to escape, and she can even feel like she’s handing over a reasonably stable patient.

On the downside, she’s definitely going to be handing off to someone five times as qualified as herself, and she’s worked for this long before but never given a report on something this complicated while already this low on mental and emotional energy. What if they think she’s stupid.

- Eh, they wouldn’t be wrong, but Merrin still feels pretty good about how this went overall. She signals back that she’s ready. 

The woman who comes in to take over looks to be in her early thirties, as is plausibly where you'd find the optimal combination of stamina, reflexes, and having done a whole lot of weird sims plus a large chunk of actual weird medical experience.

(Avarris, rank six medtech at Default Hospital.  Merrin is unlikely to recognize her from Exception Handling scenario TV; Avarris sticks to real patients for the most part.  Merrin might recognize her if Merrin is familiar with the most famous medical oppers on the planet as such.)

It’s probably a good thing for Merrin’s state of mind that she isn’t, at this point in her life, familiar with the most famous non-TV-scenario medical oppers. (She is going to look up Avarris and freak out AFTER.)

…She does catch herself four seconds into the usual “dropped by surprise on a pile of steaming superheated nuclear waste” sim report with all the background context and scene-setting, before remembering that this continues not to be a sim, and obviously Avarris has been watching the proceedings and will be up to speed on everything she can follow from outside the room.

The remaining handover mostly concerns Merrin trying to convey as much as possible of the procedural information involved in handling the machines.

This unfortunately is embarrassing, because Merrin is notably worse than most dath ilanis at converting her intuitions into a legible-to-others verbal explanation at the best of times, and mental fatigue hits her verbal fluency significantly sooner and harder than her ability to keep reacting intuitively. Fortunately she’s had them in a fairly predictable state for a couple of hours, so can at least hold everything else together while Avarris has a chance to incrementally take over one machine at a time and get used to it.

….Wow she is actually pretty exhausted. It was less noticeable when she had been in a half-trance flow state doing the exact same things for five hours, but Merrin is so ready to be done.

Though not to go home. She’s too wired to be sleepy and too emotionally invested to walk away before she finds out what happens next.

Avarris will do very well, very quickly; and only do a few things, while operating an increasing number of machines, that strike Merrin as clearly not what she'd do for this patient.

Wow! That’s impressive. Merrin wants to be that cool when she grows up which is never going to happen, a lot of it has to be raw thinkoomph, that’s so fast. Being sad about that is a waste of mental bandwidth she doesn’t have.

- Well, maybe the thing Avarrisdid instead is a totally reasonable thing to do and Merrin is just missing the reasoning for why! Merrin totally does not do anything mortifying like try to slap her hand away from the console, even though she is very tempted and apparently less inhibited than usual! She will politely point this out while immediately tabbing across all her sensor-data screens to make sure that didn’t break anything else.

They at least manage not to cause any new and exciting deterioration while doing the handover.

And now Merrin can sneak out and find a quiet corner somewhere else on the unit to sit down and watch the next proceedings? 

When Merrin leaves the operating room, she'll find somebody in the (overt) uniform of Exception Handling, waiting for her.

He can call her a helicopter straight home, accompany her home and give her updates, accompany her to a massage parlour and give her updates while somebody works on her muscles, or she can head over to the hospital command center as has been commandeered by Very Serious People to watch how the final rewarming ends.

(She's also got a fairly unlimited number of other options, like getting flown by private jet to Nandville on a comped ticket.  Like, one hears secondhand reports that Merrin personally is unlikely to go for that.  But she should be aware that this, and many other things, are among her options if she wishes.  Merrin has worked hard today, and Civilization recognizes that.)

Merrin is sufficiently nonplussed by this unexpected attention that she just stares blankly at him for ten seconds.

“…I’d like to watch from the command center,” she says, sort of shyly because aaaaaaaaaah Exception Handling is important and scary and she feels unusually stupid even for her. “I. Um. If there’s space and I won’t be in the way or distracting anyone?”

"I expect you'll validly distract some people who would on reflection endorse their desire to pay attention to you," he says, heading off in almost the right direction inside the hospital.  "Oh, one quick word though - don't say or do anything that would contradict the theory that you're a real Sparashki entering her aquatic phase.  We passed word to the Fake Conspiracy section of Exception Handling, and they've spent the last few hours quickly planting evidence consistent with how Civilization should look if the Sparashki are real.  The notion being that their apparent fictional status and licensing is just a cover, so Sparashki can walk around if they have to and just get compliments on their incredible cosplay.  Since this event is medium-secret, the CEO of Yattel's Less Expensive Tunneling Machines has been photographed by surprise through a window, looking like a Sparashki, to explain why conspiracy-theoretic research is suddenly focusing there and turning up the evidence we've planted."

People are going to pay attention to her???? Why this??????? Merrin wants NOT THAT. Unless maybe the attention is a hug. From her mom specifically. If someone from Exception Handling tries to hug her she has no idea how she would react but she might cry. 

“I, um, oka…ay,” she manages. It’s fine. Everything’s fine. She’s not even that much of an expert on fictional Sparashki, mostly her friend convinced her to pick that cosplay as an excuse to show off her swimming skills. Merrin is terrible at bluffing (by dath ilan standards) under normal circumstances. Maybe she just won’t talk???

Also who even had time to go plant a fake conspiracy in the last twelve hours? Sure, it’s useful cognitive training for someone somewhere probably, but weren’t people BUSY? Were they somehow still bored despite the absurd level of THINGS HAPPENING and needed to add in even more fictional complications to keep themselves amused? Sometimes Merrin feels like she does live among people of a different species, and this is one of those times. 

She will follow the Exception Handling person and try not to look as socially uncomfortable as she feels.

(She doesn’t really succeed at that, but it’s notable that even now, she doesn’t look physically especially tired. It was actually an unusually light shift for physical exertion, and she’s so caffeinated.)

Correctly reading that Merrin looks more fried by her twelve-hour medical emergency than like she's following along with all this, he'll mention that it's not important for Merrin to execute her Sparashki masquerade perfectly.  Exception Handling didn't do all that for a reason.  It's just that, if Exception Handling only did things that made sense, it would then be far too easy to infer the true states of affairs in Civilization by looking at Exception Handling's actions and asking what background truths would make their actions sensible ones.

…Sure. That makes sense. Merrin can at least avoid denying being a Sparashki, though if someone straight-up asks her to her face, she will probably react by bursting out laughing and then possibly having to flee the room before she cries (wow this was way more emotionally and socially intense and draining than anything she’s done before, huh. Merrin feels sort of vaguely capable of going to the sim room, but definitely not of having a normal conversation with anyone smart and famous aaaaaaah.)

This person happens to be unusually good at reading people and has been briefed on Merrin in particular.

So before they get to the command center (once he's correctly figured out how to get there inside the hospital, if Merrin doesn't correct him earlier), he'll ask Merrin to wait a quarter-minute, and duck in ahead of her.

Then he'll inform everybody present (physically or remotely) that he infers medium-confidence that Merrin is pretty fried from her twelve-hour emergency shift, and possibly nobody should talk to Merrin directly for a few minutes unless she initiates conversation.

Very understandable!

Governance's Chief of Medical Oversight will get out of one of the comfier chairs, in case Merrin happens to want to sit down once she comes in.