100ideas

what a non-answer answer!

You just answered my question:

Is it the case that UnitedHealth and Cigna each own (or control) one of the "big three" PBMs? If so, that is a just crazy - the control insurance premium pricing, benefit decisions, AND the pricing of covered medications?

yadaebo wrote below "Medical Loss Ratio (MLR) is capped at 85% in the US which means 85% of revenue must go to patients". Does controlling a big PBM allow an insurance company a loophole?

Very interesting comments and moderation discussion on this article.

I think you both make valid points, but I also get the sense that the article is articulating insights gained from pure math explorations into the theoretical limitations of learning, which in the article can sound "turboencabulator-speak" when compressed into words.

Maybe I should have just linked to the research paper:

[B'MOJO: Hybrid state space realizations of foundation models with eidetic and fading memory](https://www.arxiv.org/abs/2407.06324)

Oops, thanks. I changed it.

Basically, as LLMs scale up, the author (Soatto, VP at AWS) suggests they're beginning to resemble Solomonoff inference: hypothetically optimal but computationally infinite approach that executes all possible programs to match observed data. Repeating this approach for any given question by definition gives the best answer, yet requires no learning, since the entire process can be repeated for any query (thanks to infinite computation).

The article develops a theoretical framework contrasting traditional inductive learning (which emphasizes generalization over memorization) with transductive inference (which embraces memorization and reasoning). Here's a quote:

"What matters is that LLMs are inductively trained transductive-inference engines and can therefore support both forms of inference.[2] They are capable of performing inference by inductive learning, like any trained classifier, akin to Daniel Kahneman’s “system 1” behavior — the fast thinking of his book title Thinking Fast and Slow. But LLMs are also capable of rudimentary forms of transduction, such as in-context-learning and chain of thought, which we may call system 2 — slow-thinking — behavior. The more sophisticated among us have even taught LLMs to do deduction — the ultimate test for their emergent abilities."

Sadly, the opening quote is not elucidated.

Yes, I should have made that clear in my first comment. Thanks for doing so. I used the quote in my title because I found it a fascinating way to start a technical blog post, and it made me want to read the article to understand what the author was planning to write from such a beginning.

I found the opening quote of this article to be intriguing, especially since it was from a 1992 research lab:

“One year of research in neural networks is sufficient to believe in God.” The writing on the wall of John Hopfield’s lab at Caltech made no sense to me in 1992. Three decades later, and after years of building large language models, I see its sense if one replaces sufficiency with necessity: understanding neural networks as we teach them today requires believing in an immanent entity.

Who are the lobbyists pushing this?

So the control was the previous GILEAD drug (presumably a weekly injection?) for the same condition (HIV+)?

reminds me of the anthropic's recent work on identifying the neuron sets that correlate to various semantic concepts in Claude: https://news.ycombinator.com/item?id=40429540 "Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet"

Ditto. The larger question is how does the linear sequence of DNA (primary sequence) ultimately drive the spatio-temporal development program that leads to mature differentiated cells working together at the organoid / organ / organism scale. How do cells know what to do in time and space as the organism grows? How is that logic encoded in the genome?

Eric Davidson did a bunch of pioneering work meticulously "debugging" this spatiotemporal genomic logic in the sea urchin. Pretty amazing. Eukaryotes like us have control elements directly upstream of our genes (trans-acting aka close acting) and also 100,000's of base pairs distant (cis-acting). The region of DNA directly preceding the beginning of an open reading frame at the start of a gene usually has a sequence of DNA motifs that bind proteins that can increase or decrease expression of the gene. Davidson and others showed that the transcription factor proteins that bind to these control motifs actually have additional other proteins that bind to them, in literally a layer on top, and that the sequences of proteins in this second layer recruit a tertiary layer of proteins that conditionally cause more or less gene expression, depending on their identities. You could say the secondary and tertiary layers are a form of "abstraction" in a literal sense, since they encode a hierarchy of logical operations.

Here's an open-access overview of Davidson's work which incidentally illuminates a lot of these concepts in more detail for a lay audience: "ERIC DAVIDSON: STEPS TO A GENE REGULATORY NETWORK FOR DEVELOPMENT" by Ellen Rothenberg, 2016; doi: 10.1016/j.ydbio.2016.01.020 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4828313/

to see the decoded logic in pseudocode and with a diagram, see "cis-Regulatory control circuits in development", Howard and Davidson, 2004, Developmental Biology, vol 271, https://doi.org/10.1016/j.ydbio.2004.03.031 (open access)

one of my favorite books since high school. I remember mentioning it to my HS English teacher before class one day (20 years ago) and she amazed and surprised me by replying "oh that's one of my favorites!" It seemed like an esoteric book to me at the time, but her generation also knew of it. maybe the OP can share what they find interesting about it...

Dear god, how is that an advantage? Are we all here just rooting for techno-dictator supremacy?

This comment is tone-deaf to the unique (and effective? TBD) arrangement of the board OpenAI 501(c)3 without compensation and the company they regulate. Your comment strikes me as not appreciating the unusually civic-minded arrangement, at least superficially, that is enabling the current power play. Maybe read the boards letter more carefully and provide your reaction. You castigate them as “non-techies” - meaning… what?

what's the deal with the host url?

In principle, these costs could be 1000x less. Much like minicomputer -> dell price shift.

VSI thermodynamics is my favorite so far, it covers statistical mechanics of gasses, Boltzmann distribution defn of temperature… if you don’t know what these concepts are, I am confident you can get a good intro from the VSI thermodynamics book.

The author referenced a paper from 2009 about the microRNAs and BK subunits: "Post-transcriptional regulation of BK channel splice variant stability by miR-9 underlies neuroadaptation to alcohol" https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2714263/

"Tolerance represents a critical component of addiction. The large conductance calcium-and voltage-activated potassium channel (BK) is a well-established alcohol target, and an important element in behavioral and molecular alcohol tolerance. We tested whether microRNA, a newly-discovered class of gene expression regulators, plays a role in the development of tolerance. We show that in adult mammalian brain alcohol upregulates microRNA (miR-9) and mediates post-transcriptional reorganization in BK mRNA splice variants by miR-9-dependent destabilization of BK mRNAs containing 3’UTRs with a miR-9 Recognition Element (MRE). Different splice variants encode BK isoforms with different alcohol sensitivities. Computational modeling indicates that this miR-9 dependent mechanism contributes to alcohol tolerance. Moreover, this mechanism can be extended to regulation of additional miR-9 targets relevant to alcohol abuse. Our results describe a novel mechanism of multiplex regulation of stability of alternatively spliced mRNA by miRNA in drug adaptation and neuronal plasticity."

And here is the Nature article: "Hidden order across online extremist movements can be disrupted by nudging collective chemistry" https://www.nature.com/articles/s41598-021-89349-3

In my experience it’s easier to measure and thus control the temperature of a thermal mass that is much much larger than both the sensors (thermocouples) attached to it and the samples inside of it.

The large thermal mass of the heating block in a traditional thermocycler vs the relatively tiny masses of the samples inside the pcr tubes contained within means that the samples are basically always at the temperature of the block, which means the control loop for the heating/cooling system can be modeled and controlled with just a PID feedback loop.

Most excellent. After systematically scraping the nyt homepage and exploring some (undocumented) apis, the author quantifies “how much of the homepage is straight news vs puff/opinion/other?”

Answer: ~48%

Logically, the next step was to create a “news-only” version of the nyt homepage: https://nyt.tjcx.me/?tone=NEWS

Thanks for mentioning muon-catalyzed fusion, it’s an interesting phenomena that should come up more often in these discussions.

There appears to be a newer and more extensive paper that encompasses and expands on Lentz' work, while also claiming Lentz's paper was mathematically incomplete or un-reproducible in some way:

Bobrick A, Martire G. Introducing physical warp drives. Classical and Quantum Gravity. 2021 Jan 25. https://arxiv.org/abs/2102.06824

from the abstract:

...We present the first general model for subluminal positive-energy, spherically symmetric warp drives; construct superluminal warp- drive solutions which satisfy quantum inequalities; provide optimizations for the Alcubierre metric that decrease the negative energy requirements by two orders of magnitude; and introduce a warp drive spacetime in which space capacity and the rate of time can be chosen in a controlled manner. Conceptually, we demonstrate that any warp drive, including the Alcubierre drive, is a shell of regular or exotic material moving inertially with a certain velocity. Therefore, any warp drive requires propulsion. We show that a class of subluminal, spherically symmetric warp drive spacetimes, at least in principle, can be constructed based on the physical principles known to humanity today.

from pg. 3:

...(Natario 2002) constructed a warp drive solution without space contraction or expansion, contrary to the earlier assumption that it facilitated the movement of warp drives. (Natario 2006) constructed a new subluminal warp drive solution in the weak-field regime, which required negative energies. (Loup et al. 2001) had previously introduced a modified version of the Alcubierre drive intended to alter the rate of time for the observers inside the bubble. However, their modification reduces to the original Alcubierre metric, as we also show in Appendix A. Finally, (Lentz 2020) has recently proposed a warp drive metric claiming to have purely positive energy everywhere in both subluminal and superluminal regimes, although without providing means to reproduce the study.

The works above, to our knowledge, summarize all the modifications of the Alcubierre drive available in the literature. Superluminal travel had also been studied by (Krasnikov 1998) and (Everett & Roman 1997). In these studies, the authors introduced Krasnikov tubes...

---

edit- adding quote from pg 4 for context:

In Section 4, we introduce a simple method to construct warp drive metrics by choosing the state of spacetime inside them. Additionally, we show that the Alcubierre metric is only a particular member of a much more general class of warp drive solutions admitting negative energy densities. Subsequently, we construct several spacetimes with a range of new properties. We introduce a macroscopic negative- energy drive which satisfies quantum inequalities, as well as several metrics with negative-energies which allow one to control the rate of time and spatial contraction inside the drive. In Section 5, we review the interpretation of warp drives in light of this study.

This may (?) be the same article, unpaywalled at msn https://www.msn.com/en-gb/money/technology/facebook-bans-aus...

Reminds me of the "Kind of Bloop" fiasco 10 years ago: https://waxy.org/2011/06/kind_of_screwed/

Remember Kind of Bloop, the chiptune tribute to Miles Davis’ Kind of Blue that I produced? I went out of my way to make sure the entire project was above board, licensing all the cover songs from Miles Davis’s publisher and giving the total profits from the Kickstarter fundraiser to the five musicians that participated.

But there was one thing I never thought would be an issue: the cover art.

Before the project launched, I knew exactly what I wanted for the cover — a pixel art recreation of the original album cover, the only thing that made sense for an 8-bit tribute to Kind of Blue. I tried to draw it myself, but if you’ve ever attempted pixel art, you know how demanding it is. After several failed attempts, I asked a talented friend to do it.

Well, the government has spent billions paying Boeing et al to develop a (non-reusable) public option, Orion, but it has had multiple problems (thanks, Boeing!). So if they are publicly endorsing SpaceX, that’s a serious rebuke to the public-private industrial complex behind Orion.

Someone should price this out on transcript aka https://strateos.com/ cloud lab...

article: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7507405/

some quotes:

"Consider for example, the arrangement of iron filings sprinkled over a magnet. A conventional computer could calculate their configuration by inputting the initial random configurations of the filings into an algorithm that implements either Maxwell’s equations or the equations of quantum electrodynamics to output their final equilibrium configuration. Yet the EM field at each point in space, generated by the electron spin of iron atoms within the magnet, instantly computes this solution. In this sense, the field represents an algorithm in space, rather than the algorithms in time that are implemented by Turing machines. And, most importantly, the information involved in the computation is simultaneously available in the space of the magnet and its surroundings. It is spatially, rather than temporarily, integrated information. The EM field’s information is complex information that is physically bound."

...

"I previously proposed that ‘Nearly all qualia—the sound of C minor, the meaning of the number seven, the image of a triangle, the concept of a motor car, the feeling of anger, etc.—are similarly complex … conscious states [that] integrate parallel information streams to form a model that is both complex and physically unified within the cemi field.' That is, the qualia associated with hearing the musical note middle C is what an a EM field perturbation in the brain that correlates with the sensory input of middle C feels like, from the inside."

The infographic ranks the pandemics by total deaths, but "deadlyness" should be compared by normalizing against global population I think.

After some searching I found https://en.wikipedia.org/wiki/Estimates_of_historical_world_...

Here are their first 7 "deadliest" normalized against estimated global population and resorted.

figures in parenthesis are (total deaths / world population at end of pandemic) in millions

end date - plague - global mortality rate

---------------------------------------------

565 Plague of justinian: 2.27% (50 / 220)

1351 Bubonic Plague: 0.56% (25 / 443)

1977 Smallpox*: 0.4% (500 / 12,000?) [0.8% if assuming total pop was only 6000M)

1547 Cocoliztli: 0.33% (15 / 460)

1918 Spanish Flu 0.27% (50 / 1,840)

1520 Mexico Smallpox Epidemic 0.11% (5 / 460)

2020 HIV/AIDS 0.05% (32 / 7,300)

2020 Covid19: 0.0016% (1.13 / 7,300) (global deaths as of 22 Oct 2020)

I wasn't sure how to handle smallpox, their total deaths estimate spans a 100 year period from 1877 to 1977. Global population increased from 1.33 billion to 5.9 billion people, and over the 100 year period many more than 5.9 billion people were alive and then died (i.e. the 1.33 B alive in 1877 are not the same as those alive in 1977).