ebolyen

Was it this exp-minus-log arxiv paper?: https://arxiv.org/html/2603.21852v2

I think people are reacting to the press-release more than the work.

I don't see why this is definitely doomed just because they discuss beta-amyloid plaques. Those exist and are real. They probably don't cause it any more than tombstones cause graveyards; very related, but not in the directly mechanistic way we wish.

Alzheimer’s disease (AD) is a prevalent neurodegenerative disorder characterized by the accumulation of amyloid-beta (Aβ) peptides in the brain.

This can be true and still not be the specific mechanism.

You can treat a specific waste product or you can repair the waste stream. The issue may be waste, but not a specific product, or the issue may not be the waste stream at all.

This work appears to demonstrate evidence of waste stream repair via a well-known waste-product. That doesn't mean that any specific waste product is or is not the problem or that this particular stream is definitely going to remove enough of the waste (if that was the problem).

Maybe there have been a lot of drugs which have similarly attempted waste-stream repair so there's good reason to doubt it on that alone. But I don't think that mentioning beta-amyloid plaque is enough to discard this out-of-hand.

Very relevant comic:

https://thejenkinscomic.wordpress.com/2024/12/01/brady-bunch...

I'm not aware of any, but it wouldn't be insane to build a seekable deflate implementation by defining offsets in a zip comment. This would leave the zip file backwards compatible to usual decompression while allowing internal seeking within an individual file if the decompressor was aware of this index.

Seems we need a .zipkeep file then.

Just kidding, I don't see how the overhead of the directory entry is even remotely enough to warrant removal. Most of the magic can be left to efficient DEFLATE compatible blocks and removing entries not in the central directory in the first place (ZIP files can support concatenation of new data so long as you re-write the central directory at the end of the file).

There's really a lot more you can look at here. Lot's a prior art on super-cookies and fingerprinting:

https://coveryourtracks.eff.org/

https://amiunique.org/

The moxy of this is inspiring.

I'm curious to know what you would rate as the most important features to make this work? It seems like calc+if do a lot of the heavy lifting, but the new function syntax is what makes instruction lookup tractable.

It does seem like the state of the art differs from popular understanding. Not only is mitochondrial DNA straight forward (although not especially useful for forensics as it is maternal), but with specialized extraction it is still possible to recover nuclear DNA, just exceedingly painful to do so.

https://www.sciencedirect.com/science/article/pii/B978032399...

It is the initial purpose of a microbiome to be at least commensal, in that it is usually prohibitively expensive to maintain a sterile environment so the odds of a true pathogen colonizing a system is greatly reduced if you simply have a crowded space of neutral participants.

Once that's true it does seem there's a lot of host and microbiome interactions we've only begun to explore, but it shouldn't be surprising that co-evolution of the microbiome and host begins to take over as soon as you have one. One great example is short-chain-fatty-acid (SCFA) producing bacteria in the human gut. [1] These seem to be essential, and if there was a general takeaway to improve health, it would be to eat your roughage so they can do their job.

This is also why high alpha-diversity (community richness in particular) is such a dead-ringer for healthy vs diseased states. And frustratingly, is often exactly where the story ends for a lot of observational studies.

Also, in case you are curious, artificially sterile mice (gnotobiotic mice) tend to act differently than other mice, which is pretty odd to be honest, and why the gut-brain axis is a plausible mechanism to research further. [2]

[1] https://pmc.ncbi.nlm.nih.gov/articles/PMC10180739/ [2]: https://www.sciencedirect.com/science/article/pii/S088915912...

No, because alignment, in the general case, is O(n^2). It is ironically one of the more tractable and well solved problems in bioinformatics.

If anyone is interested in a more formal descriptions of these control-loops, with more testable mechanisms, check out the concept of reward-taxis. Here are two neat papers that I think are more closely related than might initially appear:

"Is Human Behavior Just Running and Tumbling?": https://osf.io/preprints/psyarxiv/wzvn9_v1 (This used to be a blog post, but its down, so here's a essentially identical preprint.) A scale-invariant control-loop such as chemotaxis may still be the root algorithm we use, just adjusted for a dopamine gradient mediated by the prefrontal cortex.

"Give-up-itis: Neuropathology of extremis": https://www.sciencedirect.com/science/article/abs/pii/S03069... What happens when that dopamine gradient shuts down?

Not an expert, but I have a bit of formal training on Bayesian stuff which handles similar problems.

Usually Gibbs is used when there's no directly straight-forward gradient (or when you are interested in reproducing the distribution itself, rather than a point estimate), but you do have some marginal/conditional likelihoods which are simple to sample from.

Since each visible node depends on each hidden node and each hidden node effects all visible nodes, the gradient ends up being very messy, so its much simpler to use Gibbs sampling to adjust based on marginal likelihoods.

It's been a long time since I took a class like this, but I definitely had a similar experience to the author.

Ideas like fold and map where _never_ mentioned in lisp (to exaggerate, every function had to have the recursive implementation with 5 state variables and then a simpler form for the initial call), at no point did higher-order functions or closures make an appearance while rotating a list by 1 and then 2 positions.

The treatment of Prolog was somehow worse. Often the code only made any sense once you reversed what the lecturer was saying, realizing the arrow meant "X given Y" not "X implies Y", at which point, if you could imagine the variables ran "backwards" (unification was not explained) the outcome might start to seem _possible_. I expect the lecturer was as baffled by their presentation as we were.

In general, it left the rest of the class believing quite strongly that languages other than Java were impossible to use and generally a bad move. I may have been relatively bitter in the course evaluation by the end.

I wish you luck with tracking down versions of software used when you're writing papers... especially if you're using multiple conda environments.

How would you do this otherwise? I find `conda list` to be terribly helpful.

As a tool developer for bioinformaticians, I can't imagine trying to work with OS package managers, so that would leave vendoring multiple languages and libraries in a home-grown scheme slightly worse and more brittle than conda.

I also don't think it's realistic to imagine that any single language (and thus language-specific build tools or pkg manager) is sufficient. Since we're still using fortran deep in the guts of many higher level libraries (recent tensor stuff is disrupting this a bit, but it's not like openBLAS isn't still there as a default backend).

This is true of corals, and they are often considered "colonial" organisms instead of an individual.

That said, I don't think anyone who studies biology is particularly concerned with hard-line definitions, as nature tends to eschew them every chance it has.

I think Pando and corals being considered "modular bodyplans/habits" is perhaps a more useful concept than individual or clone.

I see, so it's a procedural language that is well understood by those who fly (not just some semi-structured data or ontology). This is a great example of the advantage of domain experience. Thanks for sharing!

I like that approach.

I don't know, that sounds like a complex kind of ingest which could be arbitrarily subtle and diverge over time for legal and bureaucratic reasons.

I would kind of appreciate having two formats, since what are the odds they would change together? While there may never be a 3rd format, a DRY importer would imply that the source generating the data is also DRY.

Branding works on primitive types as well, which is I think the most interesting use case.

I would also agree that it's harder to confuse complex types as any single instance of a type is unlikely to overlap once you have a few fields.

Slightly tangential, but is there any hope of seeing Polar return as a (maintained) open source system?

I absolutely love the concept of using a logic language for authorization, and I think Polar's aesthetic qualities make it significantly more approachable for most people (over Prolog/Datalog).

But even without the authorization problem, Polar is just... really nice looking. It would be awesome to be able to use it as its own language outright.

You don't want negative numbers because that would require the description of a category to change based on whether it is positive or negative

That's actually already how it works, the kind of inverse you use depends on if the account is credit-normal or debit-normal. You end up in the same place.

Imagine showing negative revenue. That would be mostly useless.

Revenue is already credit instead of debit, which if you were consistent in using negative numbers, a negative revenue be completely correct (so you would want to sweat if revenue was a positive number). It's weird, but it lets your cash on hand be a positive number, which is just a necessary consequence of the system.

Imagine how misleading it could be to show $0 in expenses last month, when in reality, you had $100 in expenses, but you subtracted $100 out because you returned some big purchase from last month and got a refund.

Wouldn't that be accurate? This would be a debit of 100 and a credit of 100 which nets a balance of 0. But individually you would see the transactions and could sum the credits and debits if you were so inclined.

Alternatively, contra-accounts exist if reporting these individually is material for some reason.

I see a lot of consternation about credits and debits and the nomenclature.

Something that makes this simpler to think about from a modern perspective is that accounting is older than the popular use of negative numbers. By a lot. If we were to invent accounting today, we'd probably use positive and negative accounts instead of debit and credit accounts.

Algebra over addition is second nature to us at this point, but it would not have been obvious to the average merchant in 1604, and even then, negative numbers were viewed quite poorly.

What is important is that there are always two sides to a transaction and that they are inverses of each other. This is all a credit and debit are. Inverse operations on a number.

Therefore, we can make a rule that a transaction balances when credit = debit. (aka we didn't invent money as debit + credit = 0, but remember that we didn't like negative numbers when this system was invented, so this last fact is more of a happy coincidence that happens to ALWAYS WORK, for some reason, rather than the goal).

What is then reasonable is to consider literal cash on hand to be the most positive (debit) kind of account there can be, and work backwards from there. If I want to handle an expense, then I need to invert the cash account somehow (credit it) and therefore have the opposite kind of entry for where the money went (debit it). So an expense account must have a generally debit (or positive) balance.

But where did the cash come from? Well that comes from income and I want the cash to be debit-ey so the place it came from must be credit-ey otherwise I risk writing a transaction that doesn't balance (equal zero). So then income accounts must generally be credit accounts rather than debit account (aka they hold a typically negative balance or are "credit normal").

What is really killer about this system, is it just so happens that every mundane transaction you could ever write will end up as balanced transactions and each of these possible transaction accounts end up having the same usual balance of credit or debit (negative or positive). I think that is just so elegant.

Relatedly, what about integers like 9007199254740995. Is that a legal integer since it rounds to 9007199254740996?

It does seem unclear what it means to exceed precision (given rounding is such an expected part of the way we use these numbers). Magnitude feels easier as at least you definitely run out of bits in the exponent.

The interpolation of animations at the bottom is really neat, but I can't help but wish the models were a little less _active_ on the rest of the page. Math is plenty hard without a small elephant cheerleader.

Background:

The tax cuts and jobs act of 2017 in its plain language forces all costs and expenses related to software development to be amortized over 5 years (starting mid-year so 6 in practice)[0].

This act would revert that requirement to the original language permitting a complete deduction of software wages in the year accrued, rather than treating those wages as investment in a capital(ized) asset.

0: https://www.law.cornell.edu/uscode/text/26/174

Instead of imagining that it works like a number line and then observing that you can use induction, you could measure outcomes: suicide, suicidal ideation, quality of life, etc.

That puts a pretty reasonable lower bound on what is or isn't depressed. It is absolutely an altered state of mind, and an information processing disorder.

For anyone who might feel like "it's ok to be depressed", this is not a normal thought, and there is no normal amount of suicidal ideation. Seek professional help or strong social support; just ask for help anywhere it might exist (go to an emergency room and just have a seat if you must). If you aren't depressed, these thoughts literally don't happen, as strange as that may sound if you are in the thick of it.

Disclaimer, I am not an expert in these fields, and while I have some academic-ey looking references, the references are also on the periphery of their respective fields, so like.. big grain of salt, maybe several.

---

You can't, but you don't have to "want more" in the same "direction".

I think this article by Nathaniel Travis gives a lot of insight into the human condition with a remarkably simple premise:

[1] https://nathanieltravis.com/2022/01/17/is-human-behavior-jus...

The reason we can't ever be entirely satisfied is our fundamental control loop cannot consider scale (and for good reason if you think about it for a few minutes). So no matter how good things are, you will always sensibly consider how things could be better, no matter how little it would actually improve your day to day (if at all).

Fortunately, it does seem we have some control over what it is we expect to gain, so I would suggest you start "tumbling" vigorously in ways that are not detrimental to your current life and you may come upon something that really sparks an interest (at least enough of one).

I suspect there's also a darker parallel with these systems and give-up-itis, in which a lack of dopamine from the prefrontal cortex seems to cause stasis and eventually death or suicide as the basal ganglia loses the ability to perform self-directed activity (the control loop has shut down) [2]. This isn't entirely different from a bacteria failing to find a gradient and simply waiting out the situation. But from our highly derived perspective, waiting out a situation when we are the only thing that can create situations is not a useful feedback loop.

[2] https://www.sciencedirect.com/science/article/pii/S030698771...

Great summary! Although I would argue that 16S is still a perfectly good (and cost-effective) method, especially with DADA2. There are also neat sequencing techniques that like CCS which give you really high resolution of a target region (amplicon) without sequencing a lot of redundant/uninformative DNA.