Rothbardian libertarian. The social contract isn’t real.
Boston, MA
Joined November 2009
- Tweets901
- Following743
- Followers163
- Likes1.8K
AM retweeted
Introducing Muse Glimmer, an open-weight 30B-parameter model optimized for local, always-on agent workflows.
Muse Glimmer delivers strong performance on key agentic use cases and benchmarks compared with leading models in its size category, and is designed to run entirely on consumer hardware like a Mac or PCs with performant GPUs.
In keeping with our long tradition of sharing fundamental AI research, we’re releasing model weights under a permissive Apache 2.0 license.
🧵👇
AM retweeted
Every. Single. Time.
They have AC but you can't have AC!
AM retweeted
Was pleased to learn that the "hamster email" featured at a major UK administrative law conference addressing problems with extraterritorial enforcement.
AM retweeted
Prof. Alan Jagolinzer, an accounting prof who runs the censorship shop at Cambridge, says censorship fines don't work and regulators should "name human beings" instead.
Name away, professor. When that doesn't work, either, can I assume you'll be commanding the invasion fleet?
AM retweeted
What happened to my bank account? Absolutely nothing.
You believed a lie.
You believed a lie that was told to you by your political class, and your news media, to keep you from asking uncomfortable questions about how much you are paying in tax, where that money is going, and what quality of care you actually receive for the portion of it they didn't steal.
Don't believe me? Look at the pictures.
Look.
At.
Them.
That's my wife, @acrobatichobbit. Before and after.
That's a five centimeter mass. Stage 4 metastatic melanoma. The worst kind of cancer, the most vicious form of assassin your own body can betray you with. That bright area? Blood.
Ten years ago, anywhere in the world, the scan on the left is a death sentence... an endless gauntlet of painful surgeries, followed by chemotherapy, hair loss, uncontrolled vomiting, wasting away to nothing, death.
In America, today, it's not.
We have things here. Genetic therapies. Tailored viruses that attack tumor cells. Drugs that highlight cancers for your immune system, drag them kicking and screaming into the spotlight to be killed.
I won't tell you about her exact course of treatment, because that's none of your goddamned business, but I will tell you that it cost American drug companies and medical researchers a fortune to discover.
A fortune that your nation cannot afford because you chose socialism instead of progress. And socialism, however fine-sounding in theory, simply does not work.
Were she and I British, living in Britain, relying on the National Health Service, I would be a widower now.
Did saving her cost a ruinous amount of money?
Yes. This technology was expensive to create, and the people who did so deserve to pay their mortgages and feed their kids. So do the oncologists and surgeons.
Many of the men who cared for her were old men, experienced men, long past retirement age, still working because when your profession is clawing souls back from the void, sitting on a beach with a pina colada instead just doesn't hit the same.
They deserve every cent.
Did saving her cost a ruinous amount of money?
Yes.
Did I pay it?
No.
Because believe it or not, when things are ruinously expensive, but vitally necessary, we here in America come up with ways to deal with that.
Ways that don't involve creating a big pot of money and entrusting it to corrupt slimeballs.
We have insurance. And sometimes insurance isn't cheap, but the bite it takes is a hell of a lot less of what we have than the tax man takes from you.
And insurance companies sometimes have to make hard decisions about which spending choices will save the most people. I know about this in detail, because that is my wife's profession. She creates the mathematical models that pay for all this stuff.
The insurance that saved her is the exact same plan that she provides to others.
And at the end of an awful year and a half of treatment, awful because cancer medicines make you far sicker than the cancer itself...
We were left whole.
Battered and wounded in spirit, but financially whole, at least.
The only loss we took was the blow to my career as a novelist, because it turns out you can't write stories while your wife is dying, and you don't automatically recover that ability afterwards. Not right away.
I wondered every day if she was going to live or die. I wondered every day what the hell I was going to with myself without her.
But I never wondered, not for a moment, how the hell we were going to pay for all this.
Your government doesn't solve the problem. It is the problem.
They lie to you.
AM retweeted
Today the U.S. House will vote to cut 0.1% of annual federal spending.
I hope it passes, but that doesn’t make it any less pathetic.
It’s a rounding error.
Don’t let anyone in government get away with pretending this is a serious effort to get spending and debt under control.
AM retweeted
Keep in mind that I’m currently running against one of the chief architects of FISA 702 warrantless surveillance, @MikeRogersForMI, and one of its biggest backers, @ElissaSlotkin. Let’s make sure neither of these surveillance state sycophants sets foot anywhere near the Senate.
AM retweeted
"If you care about your fucking country, read Ludwig von Mises and the six lessons of the Austrian economic school, motherfuckers."
- @moicanoufc
AM retweeted
My Democratic opponent for U.S. Senate @ElissaSlotkin is so afraid Trump might become a dictator that today she voted to give any person who controls the presidency in the future unprecedented authority and power to spy on Americans in their everyday activities without a warrant.
AM retweeted
This is how the Constitution dies.
By a tie vote, the amendment to require a warrant to spy on Americans goes down in flames.
This is a sad day for America.
The Speaker doesn’t always vote in the House, but he was the tie breaker today. He voted against warrants.
AM retweeted
Any and every federal official—from the White House to Congress—who supports unconstitutional, warrantless FISA 702 surveillance needs to be removed from office.
AM retweeted
Translation: "When I was a member of Judiciary, I saw the abuses of the FBI, hundreds of thousands of abuses. And then when I became speaker, I got the confidential briefing from a different perspective: They said if I didn't support FISA, I wouldn't be speaker for much longer."
AM retweeted
NEWS: Possible Room Temperature Superconductor Replicated by Two Independent Labs
A paper was published this morning by a collaboration of two different teams looking to confirm or invalidate the original LK-99 superconductivity result.
Here's everything you need to know:
- A room temperature superconductor would be an invention on-par with the transistor, or perhaps even greater, for its ability to revolutionize every technology that uses electricity.
- LK99 was a breaking news story last August of a proposed room temperature ambient-pressure superconductor but key measurements failed to be replicated by other laboratories.
- It was determined that LK-99 was a more mundane diamagnet in a low-strength magnetic field, and developed ferromagnetic properties in a higher magnetic field - it didn't seem to be a superconductor after all.
Here's how this current publication stacks up to the previous work on LK99.
Is it the same material? Maybe.
The publication today has a slightly different chemical formulation by the inclusion of Sulfur, whereas previously LK99 did not incorporate this - although the starting ingredients in producing LK99 did include sulfur, and so, sulfur contamination may have been present in the original LK99 sample, but if the original authors thought it wasn't present, then higher purity replications would've missed the mark by cooking up the wrong material.
In other words, that this current material includes sulfur doesn't rule out that the effect is from the same crystal structure as the original LK99 authors reported (the general name for these materials is 'lead apatite crystal' which you will see floating around, like tiny rocks are want to do).
For both the current paper and the original LK99 and its other attempted replications, getting a high purity of the right crystal has always been hard, because the 'right spot' for the copper atoms in the crystal is hard for them to reach naturally, i.e. the lowest energy configuration which is the most likely was not though to get the right material properties.
So, the materials produced are low purity, might contain small specks of 'the right stuff', and its not completely known which is the right composition or how it works. There have however been simulation studies that don't rule out such a crystal structure has the right properties believed to explain superconductivity at higher than ultra-low temperatures.
How did the researchers measure superconductivity?
A defining property of superconductors is their ability to perfectly expel an applied magnetic field, up to a limit. This is called the Meissner Effect. If the magnetic field is too strong, the material will fall out of superconducting state and back into a normal conducting state. The field strength at which this happens is called the Critical Field.
Developing superconductors that have a high critical field has been the material science breakthrough which enables the majority of current approaches nuclear fusion, so, it's a big deal, although the critical field of this sample is fairly small - about 60x the strength of Earths field, or 0.2% that of an MRI.
The current publication measurement does the following:
- Cools the sample down to a low temperature
- Applies a DC magnetic field
- Measures the ability to expel this magnetic field, i.e. measures the Meissner effect
- Sweeps the field strength to find out the critical field
The susceptibility of a magnetic material is how easy it is to become magnetized in the same direction as an applied field. Since a superconductor expels magnetic fields, the susceptibility is negative.
The measurement technique they used is extremely sensitive as is based off a commercial device called an MPMS3 SQUID or Superconducting Quantum Interference Device. This is an extremely mature device built on 40 years of SQUID Magnetometry and can measure down to one part in 10e-8 in terms of sensitivity.
It works by measuring how the expelled magnetic field of a potential superconducting sample interacts with the current traveling in another superconductor circuit called a Josephson Junction (Josephson Junctions are also a fundamental building block of some designs of quantum computers).
Here is probably the most important plot from the paper, showing the 'Magnetization' is negative for small field strengths, and this effect is slightly stronger at lower temperatures but still present up to 250K, or about -23C.
An important consideration here is that there is a 'memory effect' involved, where as you sweep the applied field to a low strength, and then back up to a higher strength, you don't perfectly re-trace the result. This is called Hystersis and is why there are two lines for a given temperature in the plot.
To understand this, there are actually two types of superconductors, Type-I and Type-II. Type-I will just perfectly expel a magnetic field until you hit the critical field, and then it all falls out of superconducting at once.
Type-II are different and more interesting. When you exceed the critical field strength, the magnetic field penetrates the superconducting material and forms a tiny current loop - the magnetic field aka flux is thought to be 'pinned' through the material, and around this 'pinned flux' circulates a tiny loop of current known as a vortex. Flux pinning is why Type-II superconductors levitate over a magnet, and the current vortices can influence the measured susceptibility of the sample and generate the hystersis effect, or, memory of past magnetic fields.
Type-I superconductors therefore always have a susceptibility of -1, while Type-II can have susceptibilities that are negative fractions, i.e. -0.3, as seen in the graph.
An oddity I will point out is their device, the MPMS3 SQUID Magnetometer, is also capable of performing a measurement technique called AC Susceptibility which is usually better at picking out superconducting from small grains in an impure sample, but the authors instead used DC Susceptibility. I'm not sure why they didn't also run the AC Susceptibility test.
Whats the Net-Net?
The authors are careful to say this is a 'Possible Meissner effect' and that it is 'near room temperature' - which is wise. Last August, there was a lot of hype and a flurry of scientific interest, which didn't pan out.
Scientists have to be very cautious to get things right and not over-state their claims, since their industry largely runs on reputation of credibility to be taken seriously by others in the scientific community.
It therefore makes sense that after last August, most people will treat this with a lot of caution, perhaps dismiss it entirely, or at least not engage with as much interest and fervor as last time.
The likeliest outcome here is that Lead Apatite Crystals just have some weird magnetic properties at low field strengths, and experience a transition in properties at near-room temperatures.
A much more exciting but lower-odds outcome is that Lead Apatite Crystals end up providing a fruitful line of research that develops room temperature superconductors, which means we don't need to invade that planet from Avatar for unobtanium - we can produce it in bulk from some of the most common elements on Earth - Copper, Oxygen, Lead, and Sulfur.
AM retweeted
A further thought on the ramifications of the way capitalism places the points of decision next to the relevant information:
It's not just about proximity, or indeed expertise. It's about the tightness of the feedback loop.
One of the things I learned very quickly when I became a professional engineer was that the faster a mistake is detected, and the closer to the point at which it was made, the faster and cheaper it was to fix.
When I caught by own bugs, I could usually fix the code in minutes. If Test/QA engineers caught them, there had to be emails back and forth, multiple people involved, and it took hours. If somebody let one slip into production... well, in the embedded systems world, pushing out a patch is horrifyingly difficult and expensive.
So SMART engineers, the guys fresh out of college, would try to think very hard and be really smart, and not make mistakes. They would create vast architectures of code with sophisticated directed graphs on whiteboards, and try to code them up perfectly.
WISE engineers, the guys with some years on them, wrote code a little at a time, complied every few lines, and had a suite of unit tests they'd run almost reflexively. Hell, sometimes we'd compile and run things we KNEW were incomplete, that we knew would not work, just to get more feedback.... like SpaceX launching rockets they know will blow up.
Tightening your feedback loop, getting empirical information sooner rather than later, almost always beats smart and sophisticated thinking.
In capitalism, tight feedback loops are often automatic, and when they are not, they are often easy to design. Decision makers have domain expertise, and are right next to the flow of information, so they know very quickly when they are wrong. They are also motivated to fix the problem, because they are not only the visible source of it, but they also share the suffering it causes.
In a communist system, when a central planning committee in Moscow makes a dumbass ideological call about agriculture, and starves millions of people in Ukraine, not only can they not detect the problem early (before people start eating each other), but they also are personally better served by covering up the problem rather than fixing it.
The decision-making powers that be, in communist countries, are huge, powerful, clumsy, half-blind giants, with great power to smash things, but insufficient agility to fix or fine tune them.
By contrast, the decision makers of capitalism are small, weak, but nimble squirrels, finding a nut here, finding a nut there, storing them for the winter. Some may fail, and starve, but there will always be plenty of squirrels, and wherever there is a nut, one of the swarm will eventually find it.
Giants may rage and stomp and build huge nuclear arsenals, pile up mountains of skulls, but the squirrels will seek and dig and find and survive the winter, and they will store the next generation's acorns in the giant's bleached skull.
Spread out your decision-making. Trust the experience of the man on the ground. Tighten your feedback loop.
This is a fair question, so I'm not going to give you the Emmett treatment, but the answer is "no".
See, most people think that the problem with communism is that people aren't perfectly altruistic, or central planners aren't smart enough, but neither of these is the real issue.
The real bottleneck is data, not intelligence.
See, economics is the study of choice. Specifically, it's about the choice of how you allocated limited resources that have alternate uses.
Do you make desks out of wood, or steel? How much water do you allocate to filling swimming pools vs growing almonds?
In a communist system, there tends to be some kind of central planning entity that makes these kinds of decisions. In capitalism, the decisions are spread out, with many individual actors making small decisions about what THEY will do, based on what effects THEM.
The advantage of capitalist distributed and emergent decision making is not only that it makes far more neurons (and lately CPUs) available to this decision making process. I mean, that's a big deal. It really has orders magnitude more processing power.
But the real, insurmountable advantage of capitalist distributed decision making is that it brings those decisions close to the information they are based on.
A communist central planning committee is many layers removed from the guys in the forest cutting trees, and the guys smelting steel, and the guys making and selling furniture.
They truly can't figure out how many desks to make out of steel, and how many of wood, no matter how smart they are. They just don't have the data.
Capitalist decision makers are closest to the data that corresponds to their specialty, and they get an aggregate of other data from prices, which emerge from other people's decision making.
The dudes cutting trees know fucking everything about wood, but nothing about furniture. So they set the price of wood (based on their abilities and needs and expenses), and the guys who make furniture can work out how many wooden desks to make based on that price, because they know how many desks people buy, whether they prefer wood or steel, how much it will cost them to make each based on prices, etc.
In other words, capitalist systems farm out decisions to domain experts, while communism relies on bunch of isolated generalists.
This is why communism not only doesn't work, but can't work in any possible universe.
AM retweeted
This is a fair question, so I'm not going to give you the Emmett treatment, but the answer is "no".
See, most people think that the problem with communism is that people aren't perfectly altruistic, or central planners aren't smart enough, but neither of these is the real issue.
The real bottleneck is data, not intelligence.
See, economics is the study of choice. Specifically, it's about the choice of how you allocated limited resources that have alternate uses.
Do you make desks out of wood, or steel? How much water do you allocate to filling swimming pools vs growing almonds?
In a communist system, there tends to be some kind of central planning entity that makes these kinds of decisions. In capitalism, the decisions are spread out, with many individual actors making small decisions about what THEY will do, based on what effects THEM.
The advantage of capitalist distributed and emergent decision making is not only that it makes far more neurons (and lately CPUs) available to this decision making process. I mean, that's a big deal. It really has orders magnitude more processing power.
But the real, insurmountable advantage of capitalist distributed decision making is that it brings those decisions close to the information they are based on.
A communist central planning committee is many layers removed from the guys in the forest cutting trees, and the guys smelting steel, and the guys making and selling furniture.
They truly can't figure out how many desks to make out of steel, and how many of wood, no matter how smart they are. They just don't have the data.
Capitalist decision makers are closest to the data that corresponds to their specialty, and they get an aggregate of other data from prices, which emerge from other people's decision making.
The dudes cutting trees know fucking everything about wood, but nothing about furniture. So they set the price of wood (based on their abilities and needs and expenses), and the guys who make furniture can work out how many wooden desks to make based on that price, because they know how many desks people buy, whether they prefer wood or steel, how much it will cost them to make each based on prices, etc.
In other words, capitalist systems farm out decisions to domain experts, while communism relies on bunch of isolated generalists.
This is why communism not only doesn't work, but can't work in any possible universe.
