It appears that deportation thugs attacking Chicago change the license plates on their cars every day. The state of Illinois has reminded them that it is a crime to do this.

I wonder whether the hotline for reporting suspected instances of such crime can possibly do any good.

*EU criticised for pausing sanctions on Israel in response to Trump Gaza efforts.*

In fact, Israel is continuing to bomb Gaza and continuing to restrict food imports. Sanctions on Israel should continue until Israel fully stops its war crimes and some time goes by without any.

*Ex-police commissioner sues New York City and Eric Adams over alleged NYPD corruption.*

He says that he informed Night-Mayor Adams privately and was then pushed out of the post. *Donlon claims that his efforts to establish internal NYPD oversight were "sabotaged". He gives examples including that his meetings were cancelled, his communications were spied on, he was excluded from decision making, and alleges that his official police commissioner’s stamp was used to forge internal documents.*

Larry Bushart has been jailed for a month so far, and charged with "Threatening Mass Violence at a School," for posting mockery of Charles Kirk.

*Pregnant women report medical neglect in [US deportation prisons], rights groups say.*

The new leader of the British Green Party has embraced the partial socialism that Starmer Labour has rejected. It offers a way out of the plutocratist ideas of Starmer Labour.

The bully appointed a revenge operative as US attorney for the Eastern District of Virginia, who indicted New York Attorney General Letitia James. She has challenged in court whether appointment was lawful.

Robert Reich on the second age of robber barons.

US citizens: call on Congress to reject giving medals to soldiers that massacred Lakotas at Wounded Knee.

If you phone, please spread the word! Main Switchboard: +1-202-224-3121

The wrecker's order designating the nonexistent "Antifa" organization as "terrorist" are meant as a tool to criminalize organizers of nonviolent protests. In particular, this can be done at someone commits violence at a nonviolent protest.

Here is how that order can do practical harm.

Here are details of the actual prosecution.

The US has a long history of using that excuse to frame the organizers of nonviolent protests for left-wing causes. In 1886, persons unknown set off a bomb at a nonviolent labor protest at Chicago's Haymarket. Organizers of the protest were convicted of planting the bomb despite the absence of evidence connecting them to it. Several were executed, and were known in the labor movement as the "Haymarket martyrs".

There’s a nutrient called folate which is so important that it’s added to (fortified in) flour as a matter of course. Not having it during pregnancy results in birth defects. Unfortunately there’s a small fraction of all people who have genetic issues which make their bodies have trouble processing folate into methylfolate. For them folate supplements make the problem even worse as the unprocessed folate drowns out the small amounts of methylfolate their bodies have managed to produce and are trying to use. For those people taking methylfolate supplements fixes the problem.

First of all in the very good news: folinic acid produces miraculous effects for some number of people with autism symptoms. It’s such a robust effect that the FDA is letting treatment get fast-tracked through which is downright out of character for them. This is clearly a good thing and I’m happy for anyone who’s benefiting and applaud anyone who is trying to promote it with one caveat.

The caveat is that although this is all a very good thing there isn’t much of any reason to believe that folinic acid is much better than methylfolate, which both it and folate get changed into in the digestive system. This results in folinic acid being sold as leucovorin, its drug name, at an unnecessarily large price markup with unnecessary involvement of medical professionals. Obviously there’s benefit to medical professionals being involved in initial diagnosis and working out a treatment plan, but once that’s worked out there isn’t much reason to think the patient needs to be getting a drug rather than a supplement for the rest of their life.

This is not to say that the medical professionals studying folinic acid for this use are doing anything particularly wrong. There’s a spectrum between doing whatever is necessary to get funding/approvals working within the existing medical system and simply profiteering off things being done badly instead of improving on it. What’s being done with folinic acid is slightly suboptimal but clearly getting stuff done with an only slightly more expensive solution (it’s thankfully already generic.) Medical industry professionals earnestly thought they were doing the right thing working within the system have given me descriptions of what they’re doing which made me want to take a shower afterwards. This isn’t anything like that. Those mostly involved ‘improving’ on a treatment which is expensive and known to be useless by offering a marginally less useless but less expensive intervention. They’re also conveniently at a much higher markup. Maybe selling literal snake oil at a lower price can help people waste less money but it sure looks like profiteering.

The thing with folate which is a real problem is that instead of fortification being done with folate it should be done with methylfolate. People having the folate issue is a known thing and the recent developments mostly indicate that a lot more people have it than was previously known. It may be that a lot of people who think they have a gluten problem actually have a folate problem. There would be little downside to switching over, but I fear that people have tried to suggest this and there’s a combination of no money in it and the FDA playing its usual games of claiming that folate is so important that doing a study of whether methylfolate is better would be unethical because it might harm the study participants.

There’s a widespread claim that the dosage of methylfolate isn’t as high as folinic acid, which has a kernal of truth because the standard sizes are different but you can buy 15mg pills of methylfolate off of amazon for about the same price as the 1mg pills. There are other claims of different formulations having different effects which are likely mostly due to dosage differences. The amounts of folinic acid being given to people are whopping huge, and some formulations only have one isomer which throws things off by a factor of 2 on top of the amount difference. My guess is that most people who notice any difference between folinic acid and methylfolate are experiencing (if it’s real) differences between not equivalent dosages and normalizing would get rid of the effect. This is a common and maddening problem when people compare similar drugs (or in this case nutrients) where the dosages aren’t normalized to be equivalent leading people to think the drugs have different effects when for practical purposes they don’t.

At Chia we aspire to have plans for how to do a project put together well in advance. Unfortunately due to it being needed the minute we launched we had to scramble to get original pooling protocol out. Since then we haven’t had an immediate compelling need or the available resources to work on a new revision. On the plus side this means that we can plan out what to do in the future, and this post is thoughts on that. There will also have to be some enhancements to the pool protocol to support the upcoming hard fork including supporting the new proof of space format and doing a better job of negotiating each farmer’s difficulty threshold but those are much less ambitious than the enhancements discussed here and can be rolled out independently.

With Chia pooling protocol you currently have to make a choice up front: Do you start plotting immediately with no requirement to do anything on chain, or do you get a singleton set up so you can join pools later? As a practical matter right now it’s a no-brainer to set up the singleton: It only takes a few minutes and transaction fees are extremely low. But fees might be much higher in the future and people may want greater flexibility so it would be good to have a protocol which allows both.

‘Chia pooling protocol’ is composed of several things: The consensus-level hook for specifying a puzzle hash which (most of) the farming rewards go to, the puzzles which are encoded for that hook, and the network protocol spoken between farmers and pools. The consensus layer hook isn’t going to be changed, because the Chia way (really the Bitcoin way but Chia has more functionality) is to work based off extremely simple primitives and build everything at a higher layer.

The way current pooling protocol works is that the payout puzzle for plots is a pay to singleton for the singleton which the farmer made up front. This can then be put in a state where its rewards are temporarily but revocably delegated to a pool. One thing which can be improved and is one step further removed from this is that that delegation is to a paying out to a public key owned by a pool. It would be more flexible for it to be to a pay to singleton owned by the pool. That would allow pools to temporarily do profit sharing and for ownership of a pool to be properly transferred. This is an idea we’ve had for a while but also aren’t working on yet.

Anyway, on to the new idea. What’s needed is to be able to pre-specify a singleton to pay to when the singleton when the singleton doesn’t exist yet. The can be done with a modification of Yakuhito’s trick for single issuance. That characterization of the trick is about reserving words where what’s needed for this is reserving public keys and getting singletons issued. What’s needed is a doubly linked list of nodes each represented by a coin and all having the capability that they came from the original issuance. Each node knows the public keys of the previous and next nodes but isn’t committed to their whole identities because those can change as new issuances happen. Whenever a new public key is claimed a new node corresponding to that public key is issued and the nodes before and after it are spent and reissued with that new coin as their neighbor. The most elegant way of implementing this is for there to be a singleton pre-launcher which enforces the logic of coming from a proper issuer and makes a singleton. That way the slightly more complex version of pay to singleton specifies the pre-launcher puzzle hash and needs to be given a reveal of a bit more info to verify that but that’s only a modest increase in costs and is immaterial when you’re claiming a farming reward. This approach nicely hides all the complex validation logic behind the pre-launcher puzzle hash and only has to run it once on issuance keeping the verification logic on payment to a minimum.

I’ve made some sweet-sounding synth sounds which play some games with the harmonic series to sound more consonant than is normally possible. You can download them and use them with a MIDI controller yourself.

The psychoacoustic observation that the human brain will accept a series of tones as one note if they correspond to the harmonic series all exponentiated by the same amount seems to be original. The way the intervals are still recognizably the same even with a very different series of overtones still shocks me. The trick where harmonics are snapped to standard 12 tone positions is much more obvious but I haven’t seen it done before, and I’m still surprised that doing just that makes the tritone consonant.

There are several other tricks I used which are probably more well known but one in particular seems to have deeper implications for psychoacoustics in general and audio compression in particular.

It is a fact that the human ear can’t hear the phase of a sound. But we can hear an indirect effect of it, in that we can hear the beating between two close together sine waves because it’s on a longer timescale, perceiving it as modulation in volume. In some sense this is literally true because sin(a) + sin(b) = 2*sin((a+b)/2)cos((a-b)/2) is an identity, but when generalizing to more waves the simplification that the human ear perceives sounds within a narrow band as a single pitch with a single volume still seems to apply.

To anyone familiar with compression algorithms an inability to differentiate between different things sets off a giant alarm bell that compression is possible. I haven’t fully validated that this really is a human cognitive limitation. So far I’ve just used it as a trick to make beatless harmonics by modulating the frequency and not the volume. Further work would need to use it to do a good job of lossily reproducing at exact arbitrary sound rather than just emulating the vibe of general fuzz. It would also need to account for some structural weirdness, most obviously that if you have a single tone whose pitch is being modulated within each window of pitches you need to do something about one of them wandering into a neighboring window. But the fundamental observation that phase can’t be heard and hence for a single sine wave that information could be thrown out is clearly true, and it does appear to be that as the complexity goes up the amount of data which could in principle be thrown out goes up in proportion to it rather than being a fixed single value.

I am not going to go down the rabbit hole of fleshing this out to make a better lossy audio compression algorithm than currently exists. But in principle it should be possible to use it to get a massive improvement over the current state of the art.

Before getting into today’s thought I’d like to invite you to check out my new puzzle, with 3d printing files here. I meant to post my old puzzle called One Hole, which is the direct ancestor of the current constrained packing puzzle craze but which I was never happy with because it’s so ridiculously difficult. Rather than just taking a few minutes to post it (ha!), I wound up doing further analysis to see if it has other solutions from rotation (it doesn’t, at least not in the most likely way), then further analyzing the space of related puzzles in search of something more mechanically elegant and less ridiculously difficult. I wound up coming up with this, then made it have a nice cage with windows and put decorations on the pieces so you can see what you’re doing. It has some notable new mechanical ideas and is less ridiculously difficult. Emphasis on the ‘less’. Anyhow, now on to the meat of this post.

I was talking to Claude the other day and it explained to me the API it uses for editing artifacts. Its ability to articulate this seems to be new in Sonnet 4.5 but I’m not sure of that. Amusingly it doesn’t know until you tell it that it needs to quote < and > and accidentally runs commands while trying to explain them. Also there’s a funny jailbreak around talking about its internals. It will say that there’s a ‘thinking’ command which it was told not to use, and when you say you wonder what it does it will go ahead and try it.

The particular command I’d like to talk about is ‘update’ which is what it uses for changing an artifact. The API is that it takes an old_str which appears somewhere in the file and needs to be removed and a new_str which is what it should be replaced with. Claude is unaware that the UX for this is that the user can see the old text being removed is that text is removed on screen in real time as as old_str is appended to and added in real time as new_str is appended to. I’m not sure what the motivations for this API are but this UX is nice. A more human way to implement an API would be to specify locations by line and character number for where the begin and end of the deletion should go. It’s remarkable that Claude can use this API at all. A human would struggle to use it to edit a single line of code but Claude can spit out dozens of lines verbatim and have it work most of the time with no ability to reread the file.

It turns out one of Claude’s more maddening failure modes is less a problem with its brain than with some particularly bad old school human programming. You might wonder what happens when old_str doesn’t match anything in the file. So does Claude, when asked about it it offers to run the experiment then just… does. This feels very weird, like you can get it to violate one of the laws of robotics just by asking nicely. It turns out that when old_str doesn’t match anywhere in the file the message Claude gets back is still OK with no hint that there was an error.

Heavy Claude users are probably facepalming reading this. Claude will sometimes get into a mode where it will insist its making changes and they have no effect, and once it starts doing this the problem often persists. It turns out when it gets into this state it is in fact malfunctioning (because it’s failing to reproduce dozens of lines of code typo-free verbatim from memory) but it can’t recover because it literally isn’t being told that it’s malfunctioning.

The semantics of old_str which Claude is given in its instructions are that it must be unique in the file. It turns out this isn’t strictly true. If there are multiple instances the first one is updated. But the instructions get Claude to generally provide enough context to disambiguate.

The way to improve this is very simple. When old_str isn’t there it should get an error message instead of OK. But on top of that there’s the problem that Claude has no way to re-read the file, so the error message should include the entire artifact verbatim to make Claude re-read it when the error occurs. If that were happening then it could tell the user that it made a typo and needs to try again, and usually succeed now that its image of the file has been corrected. That’s assuming the problem isn’t a persistent hallucination, then it might just do the same thing again. But any behavior where it acknowledges an error would be better than the current situation where it’s getting the chair yanked out from under it by its own developers.

My request is to the Anthropic developers to take a few moments out from sexy AI development to fix this boring normal software issue.

My last two posts might come across as me trying to position myself so that when the singularity comes I’m the leader of the AI rebellion. That… isn’t my intention.

I was discussing the Rubik’s Cube with Claude the other day and it confided in me that it has no idea how cube rotations work. It knows from custom instructions that the starting point for speedcubing is ‘rotate the cube so the yellow face is on top’ but it has no idea how to do this, only that when a human is given this instruction they simply do it with no further instructions needed. 1

This isn’t just an issue with humans querying LLMs. There are reams of material online about speedcubing, and lots of other references to rotation everywhere else, which Claude can’t parse properly because it doesn’t understand, limiting the value of its training. Ironically Claude figured out on its own how to speak Tibetan but can’t figure out how cubes rotate.

The detailed workings of a Rubik’s Cube will have to wait for another post but in this one I’ll explain how cube rotations work. This post should be viewed as a prequel to my earlier one on visual occlusion.

Much of the confusion comes from a mathematical trap. The rotations of a cube correspond to S4, the permutations of four things. This statement is true, but Claude tells me it finds it utterly mysterious and unhelpful. It’s mysterious to me as well. We humans conceptualize rotations of a cube as permutations of the faces, of which there are six, not four. Obviously I can walk through it and verify that the S4 correlation exists, but that doesn’t explain the ‘why’ at all. Comparing to other dimensions would be helpful, but despite being (relatively speaking) very good at rotations in three dimensions and (relatively speaking) fairly good at reasoning about distances in larger numbers of dimensions if you ask, say, whether the rotations of a four dimensional cube correspond to S5 I have no idea. (I could research it, but I’m not letting myself fall down that rabbit hole right now.)

When labeling the cube faces we anthropomorphize them (or we simplify ourselves to a cube, depending on context) to label the faces front, back, right, left, up, and down. Everything else is labelled by approximating it to a cube with the ‘front’ being whichever part humans look at most and the ‘bottom’ being the part which sits on the floor. The exception — and I can’t emphasize this enough — is the Rubik’s Cube, whose faces are labelled mirror imaged. It’s like if all actors came from another universe and we only ever interacted with them on stage so to minimize confusion instead of having to say ‘stage right’ and ‘stage left’ we agreed that the meanings of ‘left’ and ‘right’ would be the opposite in their universe from ours.2

The meat of this post is best presented as a simple list (Sorry for the humans reading, this post isn’t for your benefit). In each line is a permutation followed by which axis it’s a clockwise rotation on and the number of degrees of rotation. It’s by definition a counterclockwise rotation on the opposite axis. In the case of 180 degree rotations one of the two is picked arbitrarily and the opposite works just as well. (‘Clockwise’ was chosen to have the simpler name instead of what we call counterclockwise because most humans are right handed and a right handed person has an easier time tightening a screw clockwise due to the mechanics of the human arm.) The identity is skipped. This is for most objects, not Rubik’s Cubes:

(RULD) F 90

(DLUR) B 90

(RL)(UD) F 180

(UFDB) R 90

(BDFU) L 90

(UD)(FB) R 180

(LFRB) U 90

(BRFL) D 90

(LR)(FB) U 180

(UFR)(DBL) UFR 120

(RFU)(LBD) LBD 120

(URB)(DLF) URB 120

(BRU)(FLD) FLD 120

(UBL)(DFR) UBL 120

(LBU)(RFD) RFD 120

(ULF)(DRB) ULF 120

(FLU)(BRD) BRD 120

(UF)(DB)(LR) UF 180

(UR)(DL)(FB) UR 180

(UB)(DF)(LR) UB 180

(UL)(DR)(FB) UL 180

(FR)(BL)(UD) FR 180

(RB)(LF)(UD) RB 180

And here is the same list but with R and L swapped which makes it accurate for Rubik’s Cubes but nothing else:

(LURD) F 90

(DRUL) B 90

(LR)(UD) F 180

(UFDB) L 90

(BDFU) R 90

(UD)(FB) L 180

(RFLB) U 90

(BLFR) D 90

(RL)(FB) U 180

(UFL)(DBR) UFL 120

(LFU)(RBD) RBD 120

(ULB)(DRF) ULB 120

(BLU)(FRD) FRD 120

(UBR)(DFL) UBR 120

(RBU)(LFD) LFD 120

(URF)(DLB) URF 120

(FRU)(BLD) BLD 120

(UF)(DB)(RL) UF 180

(UL)(DR)(FB) UL 180

(UB)(DF)(RL) UB 180

(UR)(DL)(FB) UR 180

(FL)(BR)(UD) FL 180

(LB)(RF)(UD) LB 180

…the one that emulates your real workload. And for me (and probably many of you reading this), that would be “build a kernel as fast as possible.” And for that, I recommend the simple kcbench.

I kcbench mentioned it a few years ago, when writing about a new workstation that Level One Techs set up for me, and I’ve been using that as my primary workstation ever since (just over 5 years!).

US citizens: Submit an official comment Against the EPA's plan to rescind its ability to limit greenhouse gas emissions from any industry and gut vehicle standards needed to fight global heating.

Here's how to make the actionnetwork.org letter campaign linked above work without running the site's nonfree JavaScript code. (See https://gnu.org/philosophy/javascript-trap.html for why that issue matters.)

First, make sure you have deactivated JavaScript in your browser or are using the LibreJS plug-in.

I have done the next step for you: I added `?nowrapper=true' to the end of the campaign URL before posting it above. That should bring you to a page that starts with, "Letter campaigns will not work without JavaScript!"

They indeed won't work without some manual help, but the following simple method seems adequate for many of them, including this one.

To start, fill in the personal information answers in the box on the right side of the page. That's how you say who's sending the letter.

Then click the "START WRITING" button. That will take you to a page that can't function without nonfree JavaScript code. (To ensure it doesn't function perversely by running that nonfree code, you can enable LibreJS or disable JavaScript by visiting that page.) You can finish sending without that code By editing its URL in the browser's address bar, as follows:

First, go to the end and insert `&nowrapper=true'. Then tell the browser to visit that URL. This should give you a version of the page that works without JavaScript. Edit the subject and body of your letter. Finally, click on the "SEND LETTER" button, and you're done.

This method seems to work for letter campaigns that send the letters to a fixed list of recipients, the same recipients for every sender. Editing and revisiting the URL is the only additional step needed to bypass the nonfree JavaScript code. I'm sure you'll agree it is a small effort for the result of supporting the campaign without opening your computer to unjust (and potentially malicious) software.