[Elizabeth Tawanda Tubbs->me] Pierogi Quiche!

Pierogi Quiche!
(Recipe can also be found in my upcoming book, “Eating for Two- Me & My Brain Tumor”)

“Ingedients”

-Enough leftover frozen waffle fries to cover bottom of pan
-Frozen pierogi, same quantity as above
-8 eggs + more
-leftover frozen sofrito
-spinach
-“Yo Life brand fake cheddar shreds” if you’re on an maoi and cannot have cheese or soy. Any shredded cheese that meets your abled desires and dietary restrictions if you arent

“Directions”

-DO NOT GATHER INGREDIENTS UNTIL TOLD TO
-grab whatever smallish round/square/triangle pan you can reach out of the top cabinet (that your cohabitent put all the pans in to dissuade you from using the stove/oven unsupervised given your hx of kitchen fires 🔥)
-take the wafflefries out of the freezer and layer the bottom of pan with them
-preheat oven to 450 (or dont, because preheating is a timewasting scam that does nothing to change the outcome)
-stick pan in oven WHILE it’s still preheating (ok, in this ONE instance preheating is beneficial)
-NOW you can gather rest of ingredients
-take pan out of oven and smush fries with a fork to form an uneven bottom layer
-layer perogis
-put in oven and start whisking (forking?) beating. 8 eggs 🥚 with a fork in a bowl
-realize you forgot to take the frozen sofrito out of the freezer and get it
-fail at combining frozen sofrito with beaten eggs until your doggo insists she cant hold it any longer
-important take pan out of oven. (Ha! Daniel)
-walk doggo
-finish combining eggs with now slightly thawed frozen sofrito, pour over pan DO NOT PUT BOWL IN SINK.
-eye pan and pretend to gauge but just wildly guess how many more eggs are needed to fill it (my guess was 4. Not at all influenced by there being four eggs left in the carton. Last time my “guess” was 3.)
-beat guessed amount of eggs for however long you have the patience for after just beating 8 eggs already
-Top pan with spinach
-pour guessed eggs over spinach, realize you guessed wrong but keep pouring because youre committed now and Never wrong
-tediously poke spinach into eggs, wishing you had stirred the spinach Into the beat egg and then poured them together. (If You read all the recipe steps beforehand like You are supposed to then You will stir the spinach into the egg. And get 2 bonus points on your test. Congrats, you ruined the curve.)
-top with “cheese”
-bake at 450, set the oven timer 45min (NOT 45 hrs. It goes 00hrs:00min there are no seconds. Good job catching that. Too bad you didnt catch yourself forgetting to put foil over the top. Oops, Daniel)
-after 45 min, let the timer continue to beep while you do whatever you got involved with, such as refilling you pill case, until your cohabitent comes out and checks the oven. (Heh.)
-Ask your cohabitant to turn off the oven (pretty please)
-forget about quiche and let it sit in the oven until you get hungry and go into the kitchen and are reminded of it when you see the refuse left on the counter.
-take quiche out of the oven and use a Heavy knife to cut since the bottom and top are crunchier than they should be.
-Enjoy with a side of caprese salad (you can google that) and lemon ginger blueberry loaf (recipe in photo description)

See 2 Photos

Jun 02 12:00AM

Continuing the saga of United being terrible (though DTW takes some of the blame on this one) my flight changed gates without any announcement

Continuing the saga of United being terrible (though DTW takes some of the blame on this one) my flight changed gates without any announcement. I was suspicious of its "on time" status with no plane at the gate and the big screen broken, so I checked flightaware which has the new gate. New gate had the flight listed and the Departures screen showed the new gate.

I went to the old gate and made the announcement as loud as I could and a couple of people got up, including one in an airline uniform (though not crew on our flight -- I asked)

I award myself one Actually Doing Stuff point (and am now boarding the correct plane)

Privacy: Public

Jun 01 12:00AM

United has hit a new low in price discrimination. If you buy the cheapest class of ticket, checking in requires a person. The person doesn't need to do anything: just swipe a badge and go

United has hit a new low in price discrimination. If you buy the cheapest class of ticket, checking in requires a person. The person doesn't need to do anything: just swipe a badge and go. They exist so you have to wait for them. Ordinary check in agents can't do this: it needs to be a special one. It's implied that a more expensive class of ticket would avoid this. They are paying a person just to make thrifty customers wait and stress.

It's possible the agent was checking that I wasn't sneaking an overhead-bin-sized bag on board, but I could have evaded such a check by stashing the bag (especially if I was flying with a confederate) and the gate agent will check for real.

Privacy: Public

Jun 01 12:00AM

I'm more mundane news, I did vacate the old place on time and turn in my keys. Even got it reasonably clean

I'm more mundane news, I did vacate the old place on time and turn in my keys. Even got it reasonably clean. Am mostly unpacked in the new place, though the living room is still a staying area, and the bedroom, which had been coming along well, is now covered in books.

Privacy: Public

Jun 04 12:00AM

This metal tube was straight. It was also a key load bearing component of my new bookcase (the old one didn't survive the move).

I'm the "bookcase" maker's defense, I did stack the modular units taller than spec (7 verticals instead of 5) but I also reinforced it against bowing and twisting with nylon cord. I think it just didn't occur to them that someone would load their creation entirely with books.

(I am not injured. I don't think any books are either.)

((Not sure where I'll sleep tonight, though, since my bed is covered in books and I have no place to put them))

See Link

Jun 04 12:00AM

[Elizabeth Tawanda TubbsOh Canada!🇨🇦. tagged me] Untitled

See 5 Photos

Apr 17 10:23PM

Photos from our canada trip (more comments on individual photos)

Photos from our canada trip (more comments on individual photos):

See 5 Photos

Apr 15 10:27PM

Untitled

See Link

Apr 08 3:30PM

Why are dependency management systems so terrible?

Whether it's autoconf, virtualenv or npm, they're always a constant source of headaches. And there's always another layer you can put on top to try to manage them, adding more complexity, and not actually solving the problem.

I see two fundamental tensions, and some unforced errors.

First tension: where do canonical libraries live? If it's on the local machine, code may lose portability. If it's on some server, external dependencies can cause flakiness. If it's in the codebase, the codebase gets huge and you have responsibility for maintaining other people's code.

Second tension: versions. If a library gets a bugfix, it's probably good to automatically upgrade, but if it gets a breaking change, it's bad. Unix tried to solve this with major and minor versions, but there are very few library authors that I'd truly trust to not put breaking changes into minor version numbers.

Unforced error: trying to pretend they're not interacting with a filesystem. They invariably are interacting with a filesystem, so the pretense turns into awkward interactions full of magic. Maybe you can't import from certain directories. Maybe you can't go up a directory because .. is syntactic. Maybe you can't kebab-case a filename...

Unforced error: import . It's really useful to be able to trace where any symbol comes from, and remarkably little work to keep a list of what you're importing at the top of the file. At least, so long as no one makes it extra work (e.g by requiring complex sorting, or barring unused imports at compilation-time). Extra dishonerable mention for Go and its habit of automatically* importing everything in the same directory.

Unforced error: leaving most of it for later. Often, half the import system isn't part of the language itself, but an afterthought add-on.

Unforced error: making the whole thing dependent on a single server which has the potential to fail and burdens somebody with the task of quality-controlling-or-not every library ever published.

So, what would I do? I think the best way to deal with the tensions is head-on, using tools and concepts from the language proper. So it might look like this:

// local file in this directory
const foo = import('foo.js')

// local file in parent directory
const pfoo = import('../foo.js')

// raw absolute path, don't actually do this
const bar = import('/usr/share/lib/bar.js')

// SYSLIBS is a core language constant with the absolute path the system libraries are installed in
// libtools is a system library with tools for finding other libraries
// Here we use object deconstruction, which every language should have
const {unTarredCacheOf, findVersion, findModuleTag} = import(SYSLIBS+'libtools.js')

// unTarredCacheOf knows how to download and untar a library, and does so
// iff it doesn't already have it cached. The hash is optional, but protects you if the
// online version changes, or if the cache gets corrupted.
const np = import(unTarredCacheOf('https://github.com/numpy/release4.2.1.tgz', {hash:54213423}))

// findVersion returns the URL for the best version of the library listed on the releases page
// that meets the restriction. Any library that wishes to be importable this way will need a
// releases.json that conforms to the standard.
const cv = import(unTarredCacheOf(findVersion('https://github.com/cv/releases.json', '2.1.*')));

// findModuleTag looks in the current directory for a file named MODULE with the
// contents ExampleProgram, then in the parent directory, then onwards toward root.
// It returns the name of the directory that has it, or throws an exception if it's not there.
const myUtils = import(findModuleTag('ExampleProgram')+'common/utils.js')

The key thing is that unTarredCacheOf, findVersion and findModuleTag are just functions. You can invoke them for other purposes. You can invoke them on the repl to see what they do. You can replace them with your own versions. (Rewriting findModuleTag will require the ability to find the calling stack frame's file, which can be done in js with a clumsy hack and Error, but should be elegantly available through some sort of Stack core builtin.)

In the comments for findVersion I said “best”. This means newest, unless it's tagged as beta or known-buggy, or unless a different best has been specified in the libtools state (yes, this means libtools is a stateful library). This operates on a stack and is useful for transitive dependencies. For example, if cv invokes findVersion for numpy '4.*' and you want to use cv without having your program's behavior change with each numpy release:

const {unTarredCacheOf, findVersion, pushVersion, popVersion} = import(SYSLIBS+'libtools.js')
pushVersion('https://github.com/numpy/releases.json', '4.2.1')
const cv = import(unTarredCacheOf(findVersion('https://github.com/cv/releases.json', '2.1.0')));
popVersion('https://github.com/numpy/releases.json')

Similarly, if a canonical version has gone offline or become corrupted, you can use pushReplacement(url, newUrl) where url may end in a wildcard to cover all children, and newUrl may be http:// or file://. The unTarredCacheOf and findVersions functions will then apply any replacements before fetching URLs.

These two features will sometimes be annoying, because they mean you cannot say exactly what your code does without knowing where it was called from. But I think this is unavoidable. Sometimes URLs go dead. At least this way the redirect is specified in the code and you know what to search for (unlike gazelle's barely-textual state files that specify how to import from 404 urls in go). Still, they should be used sparingly and for mirrors, and using this as a way of monkey-patching a meta-dependency is strongly discouraged.

One thing worth noting is that you can totally have different versions of a library imported by different files in the same project. Or, if you want, by the same file. Usually this is fine, but it becomes a potential footgun if the library has mutable state. A library that wishes to avoid this may call the exclusiveLibrary(moduleName) builtin, and if this builtin is called twice with the same name, it throws an exception. A library with mutable state but no exclusiveLibrary call triggers a warning (suppressible with an argument to import, or by declaring the variable as unimportant).

One more thing we'll need is a way to run all of a program's imports without running the program itself. Python's if __name__ == “__main__” will work here. Otherwise it'll be hard (occasionally impossible) for a static analyzer to find the full list of dependencies.

Privacy: Public

See 5 Photos

Apr 04 4:57AM

Thinking about this again in the context of the latest SSC article.

The BANAL-52 thing appears to have been genuinely unknown in May of 2021, but the PRRAR and frameshift couldn't have been. Those are pretty strong zoonosis evidence that took three years to reach me, and only because a “physics student, programmer, and mountaineer” who “obsessively researches random topics” picked this one.

It seems pretty clear that there were people who knew about PRARR and frameshift who chose not to get loud about it. And part of the reason was probably because they didn't want to legitimize lab-leak theory by engaging with it directly, either as their own judgement or because they expected their community to react negatively to such engagement.

I still don't find the lab-leak vs spontaneous species-jump question itself to be very interesting, but I do find the epistemic issues surrounding it to be interesting.

People arguing the lab-leak side have put together some very well written arguments. They get down into the details, sometimes into specific nucleotides, and they explain why it is evidence and how strong.

People arguing against lab-leak generally don't. They tend to be dismissive and insulting, using terms like "conspiracy theory" and "racist dog-whistle". Sometimes they attempt censorship on the grounds that it's "covid misinformation".

It would be very nice to say "If there were real counter-arguments they would make them; since they're resorting to this, it must be unassailably true." But this does not accurately reflect the tactical landscape. It is entirely likely that they have concluded mockery is more effective than argument when one has the opportunity to use it, and they may even be correct (at least in the short term).

So I could form an opinion based on the evidence I've seen, which is that lab-leak looks pretty convincing, or I could hold off on the basis that I've only seen selected and unchallenged evidence. From a first-order perspective, I favor the latter. I know how easy it is to make an unreasonably convincing argument by cherry-picking and spinning and not getting called on it. One side of a dispute is inherently weak evidence.

But in an iterated sense, this seems like a huge mistake. It advertises a strategy any bullshit-peddler can use to keep me on the fence.

The scientists who reject the lab-leak hypothesis have derelicted their duty to human discourse, and it is unjust to reward them for it.

This isn't the first time I've run into a dynamic like this. I doubt it'll be the last.

Does anyone have a good way to resolve such problems?

Mar 28 2:40PM

Getting an early start on this year's solstice composition.

I probably want to make the true verses more different from the intro, since things being similar-but-not-the-same can lead to confusion in singing. And maybe polish up some lyrics. I also keep wondering if the end should drop a full octave instead of just dropping to the relative minor, but that's probably too much range for easy singability.

See One Shot

Mar 13 12:43AM

I put up all my colormap stuff on github so you don't need to go through facebook's ui to browse it: https://github.com/dspeyer/colormaps?tab=readme-ov-file#colormaps

Not much new there. A few things I didn't previously generate, and a few fixes to indirect labels. Plus the (slightly cleaned up) code, in case you want to play with it.

See github.com

Mar 13 12:35AM

This was the Hudson yesterday. Lowest I've ever seen it. Low tide plus strong winds from the north, I guess. Apparently it does not drop off steeply after the shore, which is nice to know.

See Link

Mar 13 12:29AM

Maxim 11: Any baking dish bottom is removable at least once

See Link

Apr 26 11:35PM

[Elizabeth Tawanda Tubbs12 photos and a videoOh Canada!🇨🇦. tagged me] Untitled

See 4 Photos

Apr 18 12:18AM

[Elizabeth Tawanda TubbsOh Canada!🇨🇦Nika Patenaand. tagged me] Untitled

See 5 Photos

Apr 17 11:04PM

(The Team Formerly Known As The Team Formerly Known As)* The Team To Be Named Later has finally gotten the 2024 hunt archive up, albeit not linked from the main hunt page

(The Team Formerly Known As The Team Formerly Known As)* The Team To Be Named Later has finally gotten the 2024 hunt archive up, albeit not linked from the main hunt page, so I can ramble about puzzles I worked on...

## Transformations

https://mythstoryhunt.world/puzzles/transformations

Extracting the text here was really fun, even if some of the transforms were pretty iffy. And the myths were a nice way of getting more clues at things.

## 🤞📝🧩

https://mythstoryhunt.world/solutions/crossed-fingers-memo-puzzle
Variant crosswords are always fun. Much more fun than standard crosswords, a genre so played out that it's disappeared into its own lore. Emoji are a nice way to do this.

## Don't Feed the Trolls

https://mythstoryhunt.world/puzzles/dont-feed-the-trolls

The troll images in this puzzle are considerably more clever than the ones facebook keeps showing me. Not a great puzzle, since it was too dependent on subjective aspects of the quotes, but worth glancing and chuckling at (and then never sharing images like this again unless they're at least this clever).

## Connecting Connections

https://mythstoryhunt.world/puzzles/connecting-connections

Came into this one at the end, and it just seemed really elegant. Those who'd been solving from the beginning agreed it was particularly well-constructed. If people who didn't do hunt want to try a puzzle, I'd suggest this one.

## Inaugural Funky Stickperson Contest

https://mythstoryhunt.world/puzzles/inaugural-funky-stickperson-contest

I avoided this puzzle despite teammates' efforts to drag more people in. Mystery hunt should not involve this much staring at butts.

## This Space Intentionally Left... Well, You Know

https://mythstoryhunt.world/solutions/this-space-intentionally-left-well-you-know

I immediately pulled all the attributes into the spreadsheet via Download as HTML and some python. Glad I did. After which... It's really remarkable how many ways binary can encode text if you want to get tricky about it.

## The Everglades Round

https://mythstoryhunt.world/rounds/everglades

I did not work on this meta, but wanted to link the round page so you can all appreciate the art. Does the Everglades contain a portal to the underworld? Makes as much sense as some of the overworld rounds.

## Make a Winning Hand

https://mythstoryhunt.world/solutions/make-a-winning-hand

This is the puzzle I was beating my head against when we gave up for the final night. I wrote a minmax program to win the actual games. It won the first two rounds fine, but missed the third thanks to a bug in copying data off the spreadsheet. It's a case where using enums instead of strings would have saved nontrivial aggravation -- I guess that is a thing that can happen with sufficiently sleep-deprived programmers.

Looking at the solution... I was right about ancient/historical/modern! I figured I was approximating, and I'd had to fill in a bunch of those by testing.

Then the cards left in our hand were the next extraction step. Didn't think of that. And... am I seeing that the R/G/B mapping depended on the game and therefore could be different across sessions? That's a really bad design if you're rebooting your server to close all websockets every five minutes

See 8 Photos

Feb 21 12:44AM

New version of https://dspeyer.github.io/subwaymap/ , featuring

New version of https://dspeyer.github.io/subwaymap/ , featuring:

Fixed elevator data
Rotate the map so uptown is up
Squish Staten Island so there isn't a ton of white in the upper left
If you tap an arrival, all arrivals of that train are highlighted, useful for ETAs, train-change planning and noticing when a route is running in pieces
Colorful station headers
If stations are right next to each other, split one of them into a little box showing location, a bigger box showing everything else, and a dashed line connecting them

See Realtime Realspace Subway Map

Feb 28 7:52AM

[View edit history->me] This post it is only a test

This post it is only a test
It's not meant to keep up with the rest
So if judged in the way
Of all else that I say
It will at most come out second-best

And again as a quote:

This post it is only a test
It's not meant to keep up with the rest
So if judged in the way
Of all else that I say
It will at most come out second-best

-- Me, just now

Privacy: Public

Mar 02 12:00AM

One more color post. I can stop any time I like.

This time, I subtracted a blurred version for the votes to get the names that most specifically referred to specific shades.

See 3 Photos

Mar 01 12:00AM

A while back I tried to write an all-music solstice program and ran into trouble replacing No Royal Road with a song

A while back I tried to write an all-music solstice program and ran into trouble replacing No Royal Road with a song. I tried one entitled "Sometimes You Just Lose", but it didn't really come together, I think because it was too abstract. So I wrote this one, and I think it mostly works. A song isn't going to out-speech a speech, but given that constraint...

See Toast To The Fallen

Feb 15 12:00AM

I've been getting frustrated with the MTA's realtime subway info page. It won't display multiple stations at once (often relevant for decision-making) and it often requires multiple network

I've been getting frustrated with the MTA's realtime subway info page. It won't display multiple stations at once (often relevant for decision-making) and it often requires multiple network round-trips to get new data (problematic in tunnels).

Then I talked to Elizabeth Tawanda Tubbs, who had a different set of frustrations. Like how it doesn't integrate elevator/escalator or geographic data.

Then I checked out the MTA's APIs, and while I wouldn't exactly call them well-designed or well-documented, they do exist and provide a pretty good set of data.

Then I spent most of a week poking at this more intensely than I probably should, and voila: https://dspeyer.github.io/subwaymap/

Things to Note...

The actual code is pretty bulky, but loads all at the start. I recommend leaving the tab open on your phone. Though hopefully local caching will do good things.

The counting down is automatic. Fetching new data isn't. The buttons on the left control that, as well as reporting how stale the data is. Old data is often quite accurate, but obvious caveats apply.

If you don't allow it GPS access (don't worry, nothing gets uploaded), you may start in the upper left corner where nothing is. Just pan over.

Tapping on a station both brings it to the front and enlarges it and gives details. The button in the upper right shrinks it back down, but it remains brought-to-front.

If you zoom in, the relative size of text shrinks. So if stations are overlapping, this can help.

I've tested in Chrome and Opera on android and Chrome and Firefox on desktop (ubuntu). So it probably works on all modern browsers. Probably.

Please don't spread it around really widely. MTA asks devs to put in caching layers and I haven't. I figure so long as traffic is low they won't care.

Though if a bunch of you like it, maybe I'll find someplace to host a simple cache cheaply.

See Realtime Realspace Subway Map

Feb 09 7:00AM

Proposed Python feature: parameter-specific enums.

Example:

def drawLine(img, p1, p2, style % {“AA”:True, “NO_AA”:False}):

# These all do the same:
drawLine(img, p1, p2, True)
drawLine(img, p1, p2, %AA)
drawLine(img, p1, p2, style=%AA)

When evaluating the style parameter (positional or named), the unitary % operator comes into existence and means “treat what follows as a string and look it up in this parameter's % map. Keys in said map must be valid python identifiers, to avoid ambiguities in expressions. And you can use this with expressions:

# These do the same:
chmod(fn, 0o600)
chmod(fn, %OW | %OR)

I call unitary % an “operator”, but it acts on the name of the following token, rather than the value. This lets it look and feel more like an enum. The other behavior can be achieved with parenthesis, analogously to brackets in javascript map literals, but I predict it will rarely be used:

chosenStyle=”AA”
drawLine(img, p1, p2, %(chosenStyle))

Just for thoroughness, you can also get the map from the function using the binary % operator:

# These both return {“AA”:True, “NO_AA”:False}
drawLine % 3
drawLine % “style”

(Though if drawLine is an object with a __call__ member, you'll need to do drawLine.call % “style”).

You can also combine % with default values, but if so the % must come first. If so, unitary % becomes available for the default value:

def drawLine(img, p1, p2, style % {“AA”:True, “NO_AA”:False} = %AA):

The restriction is to avoid ambiguities in

def badFunction(x = “foo %s bar” % {“FOO”:1, “BAR”:2})

BadFunction does not have parameter-specific enum, just a really weird default value.

Option maps don't have to be literals, which is handy if multiple functions use the same map:

fontStyles = {
“HERSHEY_SIMPLEX”: 1,
“HERSHEY_COMPLEX”: 2,
“NESTLE_SAN_SERIF”: 3
}

def putText(img, x, y, txt, font % fontStyle)
def getTextLength(txt, font % fontStyle)

Option maps are also useful for signal values in parameters that are only sometimes flags:

def drawCircle(img, center, radius, thickness % {“FILLED”: -1})

drawCircle(img, p, r, 5) # thick circle
drawCircle(img, p, r//2, %FILLED) # filled-in circle

Without this, you get lots of magic numbers in code or symbols cluttering up the module namespace. The namespace thing is important. For example, we currently have:

MARKER_CROSS = 0
MARKER_TILTED_CROSS = 1
MARKER_STAR = 2
def drawMarker(img, pt, markerType) # markerType should be one of the preceeding

MORPH_RECT = 0
MORPH_CROSS = 1
MORPH_ELLIPSE = 2
def getStructuringElement(shape, size) # shape should be one of the preceeding

So you invoke getStructuringElement(MORPH_CROSS, (3,3)) or more likely cv2.getStructuringElement(cv2.MORPH_CROSS, (3,3)) and that MORPH is verbose and unhelpful, but if you got rid of it you'd conflict with drawMarker's flags. It'd be much nicer to write cv2.getStructuringElement(%CROSS, (3,3)) and be done with it.

(And even with prefixes everywhere, I suspect a library as big as cv2 has had difficulty avoiding collisions when different devs weren't aware of each other's constants. Maybe not cv2 specifically, since c++ enums give errors for reusing symbols.)

As you can probably tell, I've been using cv2 lately and it would really benefit from this feature. But others would too. Take subprocess.run. I'd much rather write stdin=%PIPE than stdin=subprocess.PIPE (so verbose!) or from subprocess import PIPE (what a generic name for such a context-specific constant).

Feb 04 12:00AM

Two weeks ago we had enough snow to attempt sledding. The hill north of the Civil War Memorial is steeper than it looks

Two weeks ago we had enough snow to attempt sledding. The hill north of the Civil War Memorial is steeper than it looks. And the hay bails at the bottom, while vastly better than the chain link fence we'd otherwise hit, wasn't exactly soft.

Videography and audio commentary by Elizabeth Tawanda Tubbs

See Link

Feb 01 12:52AM

Also, yet more colorspace visualizations. This time drawing topographical style maps of what percentage of answers were a certain color name.

I tried not to overlap a lot on any single image, so you may need to flip back and forth to see how color names relate.

I was thinking of using this data to look for colors that are consistently named differently, which would be good for use in board games, but with how wide these spaces are, I don't think that'll work.

See 5 Photos

Jan 21 12:00AM

You might think covering the entire cube leaves nothing left to visualize. But all of my visualizations thus far have been based on plurality name, and hide significant disagreement.

I've re-rendered the full-sat and pastel facesets this time only marking sections where a majority of respondents gave a particular answer. Much of the space is empty -- including all variations of pale green.

I've also taken three particularly interesting bits, draw lines across them, and "zoomed in" to graph the fraction of responses giving each name along that line.

See 5 Photos

Jan 24 12:35AM

I'm amazed how much there is to do with the xkcd color data. I'll probably find a new obsession soon. For now, have the entire cube, as a video

I'm amazed how much there is to do with the xkcd color data. I'll probably find a new obsession soon. For now, have the entire cube, as a video:

Play Video

See Link

Jan 02 12:00AM

Three more color planes, since the first batch proved popular

See 3 Photos

Jan 03 12:00AM

A few flavor moments from Hunt that are easy to write up (thoughts on puzzles will take more effort)

A few flavor moments from Hunt that are easy to write up (thoughts on puzzles will take more effort):

Neptune declared: "My younger brother and I move in the same circles. He may seem cold and distant, but he has a big heart."

Tiresias visited our team and offered a hint to a pair of identical twins. So David and I spoke up and Tiresias agreed that our voices marked us as identical twins.

Dionysus visited our team having lost his voice to Hera. He would wager one puzzle answer against one of our voices and a simple roll of dice. His dice were extremely weighted, always coming up with fives or sixes. So we rolled a pair of d20s.

Privacy: Public

Jan 21 12:00AM

Remember the xkcd color survey? Where they showed a ton of random people colors and then asked them to name them? It was a large-scale descriptive attempt to define color terms.

But it only provided a visualization for the fully saturated faces. I was looking back on that and thinking it should at least have the pastels. And maybe some fixed-hue slices so we can see the vicinity of grey...

So I did it. And also regenerated the original full-sat to validate my methodology (which was basically treating each answer as a vote for that name Gaussianly distributed around that color, then summing votes by color and winner-takes-all.)

For the non-three-face ones, I omitted the most obvious colors from the running to get more subtle shade-naming.

See 5 Photos

Jan 03 12:00AM

Congratulations to Death and Mayhem for preventing the world from falling into the collapsing underworld. I always knew Death and Mayhem were good at solving things.

Privacy: Public

Jan 15 2:25PM

Well, my post-a-day thing finally fizzled, but I only have three fb posts left in my drafts folder, so that's pretty good. I did say "reasonable size", not "empty".

Privacy: Public

Jan 11 2:50PM

What can be done to fix a stroad surrounded by giant parking lots?

For those of you not in urbanist circles, a stroad splits the difference between a street (which provides access to a multitude of things along it) and a road (which provides rapid transit between its ends) doing neither job well. Stroads are notoriously unpleasant for anyone not surrounded by a meter of steel crumple zones and an isolated atmosphere, and still quite dangerous to those who are. Every driveway either has a traffic light (frustrating those who just want to get somewhere) or doesn't have one (resulting in surprising high-speed merges). The particular one I pictured is Hartford/SE Road between Westfarms Mall and Costco, but I realize very few of you have spent much time in central Connecticut, so feel free to visualize your own.

The first thing to do is recalculate the parking lots. Parking lot sizes are set by law on the principle that no driver should ever struggle to find parking, even on Black Friday before Amazon was founded. On an ordinary day, there's enough empty pavement for a full size street hockey court (though anyone using it as such would be at risk from fast-moving cars headed for better parking). Recalculate based on 95th percentile usage and pledge 5 minute headways on buses the other 5 percent of the time. This gets us some land to play with.

We can get a little more land by formally abandoning the sidewalks on the stroad. They were never desirable places to walk: full of noise and exhaust at the best of times. And they were explicitly designed without anything strong enough to stop an out of control car, lest the vehicle be damaged (pedestrians were either deemed less valuable than cars, or expected not to be there).

Some of that land becomes “on/off ramps” for people making right turns between the stroad (soon to be a road) and the parking lots. Some of it becomes streets on the other side of the parking lots, right up next to the stores. And some is returned to the nominative owners, who may do as they like with it (most likely build more retail).

The remaining parking, now a municipal service, is organized into rows that run perpendicular to the road, with alternating one way lanes. Barriers between parking lots that had been “owned” by different establishments are removed.

We've mentioned that drivers making right turns in or out of parking get “on/off ramps” for smooth, lightless merging. But what about left turns, or crossing the road? At some reasonable spacing, there will be intersections where two lanes cross the road at a single traffic light, such that someone travelling along the road will experience right-to-left cross traffic and then left-to-right – the opposite of crossing a two-way street. To make this less counter-intuitive, all signage will treat the two one-way lanes as separate, and they will maintain thirty five feet of space between them, with a pedestrian crosswalk in the middle of that space. Each intersection will allow left turns either into the parking lot (from dedicated left turn lanes) or out of it. Either way, the left turns happen on the same cycle as the lanes go.

The crosswalk will have an island in the middle with real concrete-and-steel protective barriers. This protects the pedestrians, and any out of control left-turn-from-road drivers will still be better off striking a stationary barrier than oncoming traffic in the opposing left turn lane.

Meanwhile the new streets next to the actual stores will have a single narrow lane in each direction, frequent turns into parking lanes, and sidewalk-grade crosswalks, all of which should ensure slow traffic. There will be a bike lane on the store side of the street, and an assortment of services storeward of the bike lane (handicap parking, bus stops, bike racks, bikeshare stanchions, lockers, benches, artwork...). This does mean that busses and handicapped drivers will cross the bike lane, but they'll be stopping or starting, and so should be moving at sub-bicycle speeds, and some of those travelers might have trouble crossing a bike lane on foot.

This combination, I believe, benefits everyone:

Through drivers get fewer traffic lights and fewer phases per traffic light cycle. Also less chaos, less danger, and less chance of being delayed by someone else crashing.

Drivers going to stores get a slower, safer end-of-trip where they can take the time to orient themselves.

Pedestrians and bicyclists going along the streets get a much safer and more pleasant environment.

Pedestrians and bicyclists going across the road get a light cycle with no crossing turn traffic. Granted, they travel in the middle of parallel car traffic, which some find uncomfortable, but with 15 feet of dead pavement between themselves and the cars (and the cars themselves serving as ablative armor against any opposing traffic which neglects the lights).

Mass transit users get stops that are close to their actual destinations.

Existing stores get the possibility of walk-in customers.

Landlords get some of their land back, with the potential for more tenants, and get out of the parking-lot-maintenance business.

The city gets more taxable commerce and fewer emergency room visits.

The environment gets people to start looking seriously at non-car transit options.

This is the ideal form. Looking back at Westfarms-Costco, do I see some complications.

There are nearby residential neighborhoods with strangely limited access to the commercial area. I suspect they have done this deliberately to avoid traffic. They might reconsider when connecting to a street rather than a stroad, but I suspect they best answer is mode-limited connections. Perhaps just a narrow bike path, connecting a quiet residential neighborhood to the nearby bike lane through the commerce but not bringing any car traffic. In some cases, a bus-only lane may be worth building, enforced with metal posts that retract into the ground when sent a signed radio message from a bus (the crypto is easy, but building an all-weather retractable metal post may be difficult. (Rising bollards exist, but I'm not sure how all-weather they are).

At one end, route 9 passes under hartford road. We could just end the western street there, though a no-private-cars bridge to Autumn Lake Medical and Normandy Heights appts sounds useful. On the eastern side, we'll definitely need a bridge to connect Target and Costco to the rest. We're far enough away from the road that the existing ramps shouldn't be an issue, but building two bridges still bumps up the cost a lot.

The existing commerce is mostly set back a thousand feet from the road, but there are exceptions. Some (Chili's, Burger King) are up against the road. We can make sure they have good left turn access and leave it at that. Some (Seoul BBQ, Barnes and Nobles) are smaller overall. For them, I think we run the street around their backs (they may wish to partially re-orient). This opens up the forest behind them as either new commercial space or recreational nature that can actually be accessed. Most aggravatingly there is TJ Maxx sitting perpendicular to everything and positioned right across where the street should go. I think it's viable to bulldoze the westmost 30 feet of the store, grant them more space to make up for it, and pay the construction costs. It raises the total cost, but not all that much.

There's also a residential neighborhood on the east side that comes up to the road. And it's very clearly designed to prevent through traffic. The actual street will have to end there, but the bicycle path can continue on what's currently public lawn, protected by a steel or concrete barrier (and possibly something higher that muffles sound and limits the wafting of exhaust? And is prettier than a road? Maybe see what the residents of the closest houses think.

The attached diagram is of the theoretical form, with parking in grey, street in pink, sidewalk in yellow and grassland in green. The attached annotate satellite photo shows westfarms/costco in particular, with pink for streets and pale blue for mode-filtered routes.

See 2 Photos

Jan 03 12:00AM

Two recent posts using market mechanisms. How about one on the intrinsic flaws of markets?

Let's define a few symbols: q is the quantity of the good being produced S(q) is the marginal cost to produce an additional unit after q have already been produced (the “supply”) D(q) is the price the marginal customer is willing to pay for the qth unit (the “demand”) (I'm not sure if these are quite standard econ, so I'm being specific.)

Assuming no price descrimination, the producer(s)' revenue is qD(q) and their costs are ∫S (integral is from zero to q, but unicode isn't good at putting letters above integral signs). So the profit is qD(q)-∫S. To maximize profit take the derivative, so D(q)+qD'(q)-S(q)=0.

Depending on the shapes of the curves, there could be multiple solutions. The q such that D(q)=S(q) which maximizes total social value is only a profit maximizer if D'(q)=0, that is, if the price people will pay is completely independent of how many they already have (this never happens). So long as both D' and (D-S)' are negative (approximately always), the profit maximizing quantity will be less than the value maximizing.

Producers don't have a perfect ability to set quantity. They may have to deal with their own dis-unity. There's room for another entire essay there, but suffice to say they will attack that problem with all the creativity they possess.

Another option is price discrimination, which allows much higher revenues, higher production levels and higher total value generated. Assuming more value doesn't get destroyed by negative sum implementations: such as the effortful creation of real or perceived flaws in the cheaper SKU.

I don't have a solution to propose. The best option may be to live with it. But the problem runs deep.

Privacy: Public

Jan 03 12:00AM

Apropos of nothing in particular: Georgism...

For those of you not familiar, this is the idea that government revenues should be primarily derived from a land value tax. An extremely high one: approximating the use-value. Not a general property tax, but one specifically on land. Unlike other taxes, this does not disincentivize creation, as land creation is so impractical (or is it? save that thought). Indeed, such a tax prevents land ownership from competing with actual investment for investment dollars. And legitimate acquisition of land ownership is so rare that the value might as well flow to the state, which in theory will spend the money on behalf of the public. There can also be negative externality taxes (e.g. on pollution), but these exist to align incentives, and if they never raise money, mission accomplished.

I would definitely want to see one other large tax: on inheritance. There can be a sizeable deductable ($100k or so) to support items of sentimental value and to save the probate courts' time, but then 80% for illiquid and 100% for liquid. No one should be rich just because their ancestors were. Professional descendants are a toxin on any society. Obviously, this would need an army of lawyers to block large inheritance-like gifts and other loopholes. And if the loopholes are closed, large estates will end up willed to 501c3s, which will probably need a bit more of an eye kept on them too. So this doesn't generate a lot of state revenue (though it might send serious money to charities and relieve some state burdens).

But back to Georgism. There are two ways to structure it. One in which nominative private land ownership remains but is taxed at the use-value. This is the capitalist framing and is popular among hypercapitalists. The other is that the state owns the land and people rent it, which is classic socialism. They aren't very different, but I favor the socialist framing for three reasons.

First, it's less vulnerable to accounting weirdness. Many have argued that we do assess land value distinct from building value now, but those assessments are largely fictional, and that's before anyone got an incentive to mess with them. Something that's a basis for taxation needs to be robust to determined attacks (like renting a similar lot to a subsidiary at arbitrary rates in order to establish a “market rate”).

Second, it's more honest. Taxing at 100% of use value sounds like a way to substantively deprive people of property without due process of law. Let's treat the constitution and the rule-of-law principle with more respect.

Finally, if the assessment is too high, the property becomes a poison pill. Can people renounce ownership? Sell it to a corporation and dissolve the corporation without distributing the land? The entire system isn't built for land with negative effective value.

So what might the socialist framing of Georgism look like in the steady state? The state owns the land, and people rent it. If the previous renter leaves, the land goes up for rental auction (second price). If you wish to displace an existing renter, you initiate a special auction. It's your responsibility to ensure the existing owner is aware, and their bid is counted as 5% higher for purposes of winning the auction. Also, they receive half the difference in the first years' rents as compensation for the unexpected move. There are reasonable rules about timing.

Does this mean you could be minding your own business and suddenly need to get in a bidding war to keep your home? Yes. But only if your neighborhood is considerably nicer than you've been paying for, and you've been enjoying that for some time.

(Note: when it comes to renting land from the government, there is no application process. You bid; you pay; you get. If you then fail to continue paying, the government is well-positioned to track you down and seize your bank accounts. And if you have no money, the government is well-positioned to absorb the loss.)

In the absence of a special auction, the state may raise the rent at up to 5%/year. If the tenant refuses this, the lot goes up for general auction. If the state overestimated the land's worth, the tenant will win the auction at the old rent.

Is there any way to securely obtain housing and stop worrying about the real estate market? You can purchase bonds whose interest covers your rent plus rent insurance. If insurance companies don't come forward, the state will provide it and try to make a small profit while doing so.

If land changing tenants contains objects that cannot be practically detached (e.g. large buildings), the new tenant must purchase them from the old tenant. If they cannot agree on a price, the court will find one (and charge both sides a fee, to encourage out-of-court settlements, waived for parties that made offers more generous than the final).

This does bring appraisers back into the system, but they'll have a much easier time. First, it's easier to find very similar buildings than land in very similar locations. Second, buildings separate from land are often getting sold in this system while land-without-its-buildings hardly ever is the other way. Finally, there's always construction cost to anchor on.

What about mineral rights? Those are tracked separately. We don't want to create a powerful incentive to dig as fast as possible. The state owns all valuable minerals in their natural state, and decides as a matter of policy what quantity to sell each year. Then they auction off those extraction rights. (This could be thought of as another tax, but it's better considered a sale. For one thing, that reminds everyone that the reserve of underground minerals is finite.)

Another nice feature of making state ownership of land open is that it can make exceptions. If someone actually does create land (high, Netherlands!) or seizes it in a declared war under a letter of marque (could, in theory, happen) then they can own it, and it's an entirely proportional reward for the extraordinary achievement.

The land still ends up with the state in the long run, as inheritance tax when the owner dies. If the state comes to control a corporation that owns land as a result of inheritance tax, it will sell the land to itself, then slowly divest from the corporation.

We could even allow people who plan to increase land's value to buy it. They would need to convince a state agent that the land isn't going to increase in value on its own, and that they have a credible plan to do so. A credible plan might involve building high-speed rail from there to a popular city, or discovering where in the land one can efficiently reach valuable minerals. Then they pay 20x the current rent and buy the land.

Selling the land back to the state is simple. The state conducts an auction and pay 20x the winning bid. The owner gets Current Tenant status in the auction. If the owner wins the auction, they may choose to cancel the sale, so this allows sell-back with a price floor, so to speak.

Bidding in a sale auction if you do not intend to win is illegal (as this could be used to make the state overpay for the property). If the rents go up for auction again a year later, and the original second bidder does not rebid at the same level or close, this will attract investigative interest.

If privately owned land is gaining value quickly due to factors besides the owner's efforts, the state can compel a sale (in which case the cancellation option does not apply).

But that's steady state. How to get there from here?

In a sense, this is where the capitalist framing shines. You can create a tax and then gradually increase it. But this is still very harsh on land owners, who were never the true villains, just people holding the wrong hot potato when the music stopped. True, most property entered the private domain by either violence or corruption, but most current owners bought it honestly.

But back to the socialist framing. The state can buy the land using the 20x auction rule above, or as it comes up for offer naturally, at whatever rate is convenient. But this requires a large one-time expenditure.

(So the capitalist variant of land reform involves just taking it while the socialist involves fair compensation. Figures.)

I'm inclined to go back to the inheritance tax. Normally, an inheritance tax will produce a lot of people spending their money while they're alive (and hedging against long life with appropriate financial instruments) or leaving their money to charities. But an ex post facto inheritance tax can collect a great deal of money, once. Including, in a very crude way, the value of the land back when it was first stolen, however many transformations and investments its passed through since. This is not as bad as an ex post facto law, and in a sense is only taking from people what they've been enjoying unearned for some time. But it's still worryingly ex post facto. Also, someone who inherited and wasted a great deal of wealth is now on the hook for a debt they cannot realistically repay.

I suppose the more practical-minded would also prefer a route to georgism that doesn't require defeating the most powerful political power-block in a straight up conflict.

I wonder what would happen if you just printed the money. It's a giant one-time infusion of cash coupled with other economic changes. Hard to say. Inflation's weird.

Privacy: Public

Jan 10 3:00AM

Speaking of TV, let me pull my She Hulk commentary out of the drafts folder...

Before the final episode aired, I was thinking about a Nando-style One Small Change. And it's this: it's not a partner at GLK&H who approaches She-Hulk about representing Abomination; it's one of his disciples.

He doesn't do a great job of selling the case. He and the other disciples have scraped up some money, but not really enough. He's unsympathetic; Abomination's very unsympathetic, and there's the whole tried-to-kill-her-cousin thing. But he makes one argument she can't ignore: no one else will do it.

It's a parole hearing, so he's not guaranteed a public defender (some states do have public defenders for parole hearings, but some don't, and there's no canon on what state we're in). Some lawyers offered to take the case by VC, but it'll look terrible to the board that his own lawyer isn't willing to get within arms' reach of him. Sure, his lawyer being indestructible has similar implications, but they're less emotionally salient. The parole system here is basically unnavigable without lawyer, so she is his only hope for justice.

Despite her discomfort, she accepts the case.

(Note that she is not working for GLK&H here. In fact, she's been an employee her entire professional life and this is her first time in private practice, causing her various entertaining annoyance about billing, taxes, etc.)

The case goes on exactly as in the show. After winning, she starts thinking about a Superhuman Law Department. She realizes that she doesn't want to operate as a single practitioner, nor to run a small firm, so she needs an employer. She shops the idea around, and GLK&H is the only one to accept. From here, proceed as before.

But now every conflict with GLK&H isn't just her being easy to push around. It's what she has to put up with to get the department she wants. And wants for principled reasons. The same re-contextualization applies to everything else life throws at her.

Agency + Morality = Heroism.

Because apart from the title character, it's a good show. I really enjoyed the world building and the chance to explore at low stakes. And I definitely want Wong to get his own show now. (His own movie? Not so much. A movie will have a really big villain who will force Strange, Chavez, Chi, Walters, or even Madisynn to get their acts together. A show can let him semi-stomp lesser threats (still with hundreds of lives on the line) while trying to get his organization to grow up. We can explore the consequences of approximately everyone learning magic is real and do character-driven stories about Wong growing into a leader while the rest of the cast grow into real adults.)

But the title character, for all her power, doesn't act like a hero. Her role in the show is as a victim. She may be a highly capable professional with awesome superpowers and none of the usual drawbacks, but her status as a victim is too intrinsic to be touched by this, and her role in the plot is “person bad things happen to”. Sometimes by Diabolus ex Machina; sometimes by her own passivity.

And in the final episode, she says it straight to the camera: “This isn't even a reluctant superhero story; I'm just getting screwed over.”

In fact, that's pretty much the moral of the show. If you're a woman, it doesn't matter how awesome you are, you're a victim. And there's no point trying to prevail via brains or brawn, your only outlet is to find the man in charge and complain until he saves you. Which he will, regardless of his own interests, because Taking Care Of Women is just What Men Do.

It's a really toxic, sexist message. And terrible storytelling.

From a show with impeccable feminist credentials. Right down to the straw-mra villains. I'd call it “ironic”, but... it's not.

But back to storytelling. The confrontation with Kevin was kind of fun, and I was glad to see Marvel call themselves out for too many night scenes, but retconning away all the secondary villains was unfulfilling. We spent a lot of episodes wondering what was going on, and that needs to lead somewhere. So let She-Hulk throw that in Kevin's face. And Kevin promises to have an actual vision for the conspiracy and whose powers do what by season two (post-credits: Kevin turns directly to the audience and asks for fan theories). But until then, she needs to defeat them all in a rapid-fire montage of both violence and legalism.

Ending with the most satisfying: her boss at GLK&H. That's the conflict we've actually gotten invested in. And it's one she can neither punch nor sue. But somehow, in the process of dealing with all the lesser villains, she obtains leverage. Some connection or reputation that the firm can't do without. And she renegotiates a stronger position. Satisfying conclusion.

It could have worked. So many good elements. Just a little more, and it could have been a good show.

Privacy: Public

Jan 02 12:00AM

How would I have done an Ahsoka show?

(I was disappointed in the one we got, in large part because I'm so fond of the character in general. I was going to write an analysis of that, but it was boring, so have a pitch instead.)

We start with the final scene in rebels, recreated in live action (assuming we're committed to the live action part – given the choice, I'd probably keep her animated: why change what works so well?).

Ahsoka: I'm ready to go after Ezra. Are you?

Sabine: I've been ready since the day he disappeared. It's been three months. What took you so long?

Ahsoka: I wear a lot of hats for the rebellion. Had to reliably hand them all off to other people.

Sabine: I suppose

Ahsoka: Also, the purrgil don't owe me a favor like they did Ezra. He got a bunch to break from their pod and migrate early, but we get to go with the main pod on their schedule.

Sabine: Right. Purrgil schedules.

Ahsoka: Look on the bright side. They want to find their podmates, which means they'll figure out where those podmates took Ezra and take us to the same place. Could have been an awful lot of search area otherwise.

We establish the time (pre-ANH), the relationship to the purrgil, and the mission.

The actual ride goes smoothly, but Ahsoka has a bad feeling, and tells her nav computer to investigate. It takes a bunch of readings and identifies which galaxy they're in, and also that four years have passed.

Ahsoka: I guess it makes sense. There's no fundamental reason that the duration experienced in hyperspace has to match the time difference between entering and exiting.

Sabine: Every hyperdrive I've ever worked on uses fijian pathes which limit the ratio to the seventh root ---

Ahsoka: But the purrgil aren't drives. They have their own idea of when and where they want to arrive.

Sabine: But isn't it always better to arrive sooner rather than later?

Ahsoka: If you want to be prompt for the people at your destination. But everyone the purrgil care about travels with them. So why not hop down the timeline a bit? Give trouble a chance to subside, food sources time to regrow.

They find the wreckage of Thrawn's star destroyer. Ahsoka knows the purrgil well enough to know it wasn't their doing.

As this thought sinks in, a squadron of unfamiliar fighter craft drop out of hyperspace. They look almost organic, and rather than force fields, they use physical shields of ultra-dense material for protection. These shields need to swing out of the way for them to fire, which would seem like a convenient vulnerability for Ahsoka's force-enhanced timing, but she can't get a feel for them in the force. So Sabine takes over weapons, and gets some hits, but there are just too many of them. Ahsoka dives for a nearby planet, where thick upper-atmosphere clouds might provide some cover.

It's not looking like they're going to make it, but as they approach the clouds a fleet of Tie fighters comes up to meet them, spreads out, and fires on the alien ships from multiple angles, destroying them. The radio crackles to life.

Ezra: Ahsoka? Is that you

Ahsoka: It's me

Sabine: And me. Since when do you pilot a tie fighter?

Ezra: Since I needed to

Thrawn: And since I needed a pilot with super-human reflexes

::end episode one::

(Hopefully characters will be recognizable by voice despite recasting. Also, Ezra's pretty clear by context and Thrawn has distinctive word-habits.)

Episode two opens on the ground, with Ezra, Thrawn, Ahsoka and Sabine seated around a campfire, with a handful of stormtroopers camped nearby. Thrawn explains that the attackers seem to call themselves Yuuzhan Vong, and seem to practice a particularly brutal form of conquest. He hasn't been able to fully learn their language, but he believes they are preparing an invasion of the main galaxy. It'll be ready to go in only five more tethlits (units of time are hard to translate). They have two plans for getting there, a small fleet of fast scouts, and a large number of brain parasites to control purrgil with. The Empire is in no position to repel such an invasion, so he wants to stop it here. Ezra is working with him because Empire or no Empire, the invasion would be a humanitarian disaster. Also, the plan to mass-mind-control the purrgil is vile.

They go on various missions to gather intel and prepare for the big sabotage. Since it's four of them against a galaxy, they all get to be total badasses and we don't need to hold back at all. Amidst this, there are side-plots:

First, Thrawn gets in the habit of packing everyone's supplies. They're suspicious about it at first, and do a lot of double-checking, but he's always honest about it and often packs backups they turn out to need, so eventually they get in the habit of trusting him.

Also, there's an attempted romance between Sabine and Ezra. They're not a particularly good match, but once they are literally the only members of each others species in this galaxy (excepting, maybe, some storm troopers of unknown species) adrenaline and biology do their thing. And it goes badly. Eventually a mission fails because Ezra is too busy watching out for Sabine to get his part done. So Sabine breaks up with him.

Ahsoka comforts Ezra in the aftermath. Tells him about Anikin and Padme, and that she's starting to think the council had a point about the whole attachments thing.

This leaves Thrawn to comfort Sabine. They bond a bit over being mundanes in a world that seems to only take force users seriously. Thrawn goes so far as to say that the empire'd be in better shape if he were the emporer's right hand and Vader were demoted to Chief Assassin, and he hopes if the rebellion wins that Sabine is given the high place she deserves in what follows. Which is enough to make her ask “Are you trying to be my rebound relationship?”. “Perhaps if I were twenty years younger, and you were my species, and I had never tried to kill you... I do admire you. But things are not otherwise, and any romance between us would be doomed from the start.” (This totally will not deter fanfic shippers, but it's the sort of maturity the character should have and I don't think I've ever seen represented in a visual medium.)

Another thing that happens at some point is that our heroes learn the YV perspective. As they see it, the Force has a will, neither light nor dark (those are its tools) but chaotic: always keeping the galaxy in turmoil. This means that building a civilization that can last requires cutting off all tools of the force, i.e. force-sensitive people. This is done by genetic engineering if possible, but genocide when not. Ahsoka thinks this is crazy, but Thrawn thinks they may have a point: the Jedi started holding up the Republic and their power waned, then Palpatine secured the empire and immediately lost half his intelligence, while new Jedi like Ezra are suspiciously powerful for their level of training. He's not tempted by their plan, because he's against genocide and adamantly against genocide of his own people, but he wonders if they have a point about the rest of it.

Eventually the big day comes. They blow up the Purrgil Mind Control Parasites in a big satisfying explosion. They destroy all but one of the scout ships, too. And they run together to the final scout ship –

When Thrawn pulls a remote-detonator out of his pocket and pushes the button. Ahsoka, Sabine and Ezra's backpacks all explode.

They're small explosions. He couldn't hide big explosives without risking getting caught. Also, two of them are force-sensitive and the third is heavily armored. So they're not much injured, but they are thrown from the elevated walkway (which, of course, has no banisters). Sabine uses her jetpack to rescue the other two, but it's too late. Thrawn has launched the scout-ship.

Our heroes now have to fight their way to the other exit and leave in their original shuttle. They make there way back to the Purrgil, and Ezra explains that they need to get back to their own galaxy quickly. The Purrgil recognize that they owe a huge favor here, but they're not actually good at finding hyperspace routes that arrive shortly after they depart. Ezra estimates two years.

While attached to the Purrgil traveling through hyperspace, Ahsoka tries to reach out in the force to any jedi back in the main galaxy. The Purrgil are themselves highly force-sensitive, which makes them force-rich locations. And she doesn't know for a fact that either Obi-Wan or Yoda is dead. It's a long shot, but she's not giving up.

Mid-credits scene: The Vong commander surveys the wreckage of the explosion. Underling: “It will take hundreds of tethlits to rebuild everything”. Overling: “Perhaps you should hurry.”

Post-credits scene: Luke Skywalker is meditating. Suddenly he opens his eyes and instructs his computer “Call General Syndulla”. Her hologram appears promptly. “General Skywalker, what's wrong?” “Have you ever heard the name 'Thrawn'?”

Cut to black.

What did we accomplish here? Besides keeping the characters in character and providing a backbone on which fun action scenes could be hung?

We've explained where Ahsoka was during the original trilogy in a way that doesn't massively count against her. She had no reasonable chance of anticipating the time-dilation thing.

We've explained why Ahsoka is so down on attachments when she meets Grogu. She's still not considering herself a Jedi nor trusting the council's traditions, but she thinks they were on to something on this one.

We've allowed Ahsoka to join the post-RotJ world without having to imply a ton of experiences. We don't really want to shove six years of character development for a beloved character into an offscreen time-skip.

We've also explained why Ahsoka is hunting Thrawn. Presumably when he returned to find the Empire destroyed, he went into hiding and started building his forces.

We've allowed Thrawn to win. A lot. Against impressive odds. If he's going to be our next big bad, we need him to be threatening. Especially since he now has a two year head start and some Yuuzhan Vong tech to reverse engineer.

We've foreshadowed the Yuuzhan Vong, explained why they're not showing up immediately, and left it wonderfully vague how long it'll be until they do show up.

I'm pretty happy with this.

Privacy: Public

Jan 01 12:00AM

[View edit history->me] One possible design for market-based healthcare, continued from a discussion a (long) while back with Alexander Rapp

One possible design for market-based healthcare, continued from a discussion a (long) while back with Alexander Rapp

First, we simply accept near-total transparency, and trust that the flood of information causes people to chill out. We also assume that payments and account-transfers are minimal friction (though understanding and making decisions can be far higher). Finally, no patient ever lies to governance about their symptoms. These things work by authorial fiat.

When you're born, governance declares an annuity that tracks your health: perhaps $100k/y for good health, with stated fractions for “full thinking capability”, “can walk”, “not in chronic pain (but conscious)”, “breathing”, and even “intact brain”, which continues to pay out for the cryo-preserved. Once you qualify as an adult, you can adjust these fractions, or add new targets, but they must always sum to one.

The annuity is sold to the highest bidder. Given net-present value formulas and risks, the revenue brought in here covers most but not all of the annuity's expenses. The remainder represents society's concern for the wellbeing of all its members.

The buyer, probably a specialist in children's health, is now strongly incentivized to support your health, especially anything that affects you in the long term. They might offer your parents limo rides to vaccinations (note that this is in addition to herd-immunity-based incentives offered by the Public Health Department) or, as you got older, offer you monetary incentives to exercise or avoid dangerous hobbies.

Now suppose you suffer an injury. You (or whoever finds you, if you're unconscious) append your knowledge of the injury to your public health record . The owner immediately sells to the highest bidder. Most bidders are nervous about the health of someone whose recent injury hasn't been professionally examined, so the highest bidder is likely a specialist in this sort of diagnosis.

This specialist decides whether the issue is time-sensitive enough to justify an ambulance. Since they are holding the net-present value of all your future health payments, they are incentive-aligned, or something close to it. They then send an ambulance or an Uber at their expense, bringing you to a diagnostic center. They then perform the tests that will be most relevant to treatment-planners in judging whether they want your case, and sell your account to the new highest bidder. The expected value associated with your case hasn't improved, but treatment specialists are no longer worrying about types of uncertainty they are ill-equipped to price. The diagnostician captures this difference.

Next your case is bought by a treatment provider. They pay based on their confidence that they can restore you to health, minus what the procedure will cost.

Non-intervening investors might also bid at this point, based on their expectation of how well you'll recover on your own. This protects the system from both hypochondriacs and from spending vast amounts of money on lost causes.

Eventually, your health stabilizes and your case is sold to an appropriate investment firm.

All these organizations would need to swear to having no hidden motives. For example, if you suffer from something that might be gender disphoria, an organization with an interest in adjusting world gender ratios would be forbidden to bid on your case using normal procedures. This isn't done as economics, just plain old law enforcement.

But there are special procedures for organizations with additional motives, which are mostly researchers. They are required to be extremely open about what they do and why, and to follow additional procedures. To interact with such an organization (wlog we'll just say “researcher”) you specify a Risk Appetite (default 10%) and a compensation ratio (default 1).

If your health state is such that a researcher wants you, they can bid on your case. If they win, the highest non-research bid will be recorded. The researcher immediately sells a derivative that will pay out when they sell your case for 1% of what they sell the case for. If 100x the price of the derivative ever drops below the non-research bid minus your risk appetite, they must immediately sell your case (this could happen at initial sale). They are free to continue monitoring your health, and even to offer uncompensated care. When they sell your health, the price is compared to the non-research bid again. If the new price is lower, the researcher pays you the difference times your compensation ratio.

Note that if you set your risk appetite to zero, you will be in the hands of researchers if and only if the overall market expects the research team to outperform all ordinary treatment planners at treating you, and value-of-information is a nice bonus. If you set your risk appetite to 100%, you are essentially saying there is a price you will accept for the right to kill you.

Note also that as the hope for your recovery via conventional treatment decreases, the likelihood that you'll end up in a research study increases without your needing to tweak anything in your policies.

All this does a pretty good job of aligning incentives. But it doesn't cover stupidity. Sure, a stupid case-management company will lose money, but if it's attached to a profitable business and there's principle-agent stuff going on inside, it could lose money for a long time, and hurt a lot of people on the way down. If the profitable business is also case-management, but specializing in a somewhat different condition, this might be hard to detect or make rules about.

You could require that businesses that haven't been around long or are running at a loss operate under additional-motive rules. But the standard for “long” needs to be short enough that there are plenty of “ordinary” bidders in every market. And nothing here covers the “case management company overconfidently extends to an adjacent sort of medice” scenario.

And any such regulation runs into a complimentary problem: thin markets. These aren't commodities. Each market is on a specific individual's health. No single one has a lot of money to be made. Which means there's a definite risk that some auctions will have very few bidders, and all the desirable properties of a market go away. We might hope that an ill-served edge case will attract the (possibly part-time) attention of some ambitious or greedy specialist, but if the total size isn't very large, and the fixed barriers are nontrivial, it might not.

So we've now listed five issues with this: three fiated away at the beginning and two down here. But there's one more that strikes me as especially critical. This entire system treats the patient as passive.

Sometimes the patient is passive. Sometimes the patient is comatose. But this isn't actually the common case. This is a system in which Medicine is done by the investor, and it's the investor's problem to get the patient to go along with it, while Lifestyle is done by the patient, and the investor is mostly uninvolved. And that's just not how life is divided. This system gets extremely clunky when a patient wants to take agency over their own health or when something is ambiguous about whether it or not it is medical.

So I don't endorse this system. It was fun to think about, though.

Privacy: Public

See Link

Jan 11 9:00AM

One possible design for market-based healthcare, continued from a discussion a (long) while back with @Alex Rapp.