MontyCarloHall

Right. If this were truly a pure scaling issue, I’d expect the interface would offer an archive.is-esque “Claude is at capacity; your prompt is #XXX/YYY in the queue; estimated time remaining: ZZZ seconds”

Instead, the whole system just shits the bed, catastrophically.

Another data point: GitHub is extremely insistent its employees maximally use AI for internal development [0], and we’ve concomitantly seen its reliability fall off a cliff in the last year or so.

[0] https://github.com/resources/insights/ai-powered-workforce-p...

   I don’t prompt Claude anymore. I have loops running that prompt Claude and figuring out what to do. My job is to write loops.
   — Boris Cherny, head of Claude Code

Reliability is a direct reflection of the quality of the underlying infrastructural code. If even Anthropic, the company with the world's best agentic vibecoders, has horribly unreliable infrastructure, it really says something about the quality of the world's best agentically produced code.

Including additional guides is negligible with respect to the overall LNP capacity, which can contain a few thousand nucleotides. The CRISPR/Cas mRNA is about 5 kilobases total, and each guide is on the order of 100 bases.

Sure, but my point is that convergent selective pressure across different individuals' tumors is strong enough that they independently arrive on the same resistance mechanisms to a given therapy, e.g. [0]. Just like if you performed the same bacterial antibiotic resistance experiment [1] across multiple independent trials, you'd repeatedly get the same result.

[0] https://pmc.ncbi.nlm.nih.gov/articles/PMC2538882/

[1] https://news.ycombinator.com/item?id=48505231#48507925

In practice, you’d use multiple guides targeting multiple mutations, so that the probability of having/acquiring multiple resistance mutations abrogating every guide from binding would be infinitesimal.

Even with 100% delivery efficacy, editing efficacy is nowhere near 100%. CRISPR/Cas editors will reliably detect the target sequence but will not reliably edit it in order to repair the mutant allele, whereas CRISPR/Cas12a2 will activate and destroy chromatin ~100% of the time when it detects the target.

As is often the case, it's a lot easier to indiscriminately destroy than precisely (re)build.

Agreed. Assuming it's ultimately proven to work in vivo, I think the endgame of this therapy is multiple guides targeting multiple mutations along with multiple delivery mechanisms (a formulation-diverse cocktail of LNPs + eVLPs [0]?). Sure, tech like [0] is futuristic and fanciful, but so is the tech of the OP, and both will probably reach in vivo maturity around the same time.

[0] https://pmc.ncbi.nlm.nih.gov/articles/PMC8809250/

>Depending on how the LNPs are designed, would resistance also potentially cripple the cancer cells?

Yes, if the LNP could be engineered to target an essential surface receptor, which is still a very tough problem. It would also not solve the issue of the payload successfully entering the cell but being subsequently degraded.

>I've heard about drug resistance in bacteria leading to slower growth / reduced virulence. Maybe the same would occur with cancers. A drug that could effectively switch an aggressive cancer into a slow-growing one wouldn't be the worst thing.

This is essentially how treatment for chronic lymphocytic leukemia happens (hence why it's called "chronic"). People with CLL can stay on BTK inhibitors for decades, often until they die of other natural causes.

>there's no "be a better/stronger cancer and spread more effectively to more hosts"

No, but there is "be a better/stronger cancer cell and don't succumb to whatever therapy is killing its neighboring cells." It's exactly akin to how dosing isolated populations of bacteria with antibiotics selects for individual cells that are resistant, which then multiply and dominate [0], just like a tumor.

[0] https://www.youtube.com/watch?v=plVk4NVIUh8

This would shorten the timeframe for cells to mutate and acquire resistance mechanisms, but would not address the issue of cells with preexisting (epi)genetic resistance mechanisms that would then be promptly selected for.

Nonetheless, we see the exact same resistance mechanisms to the same therapies recur across individuals, e.g. [0]. Convergent evolution is a harsh mistress.

[0] https://pmc.ncbi.nlm.nih.gov/articles/PMC2538882/

The idea of using CRISPR/Cas to detect tumor-specific mutations that aren't necessarily oncogenic and then kill the cell is not a new one [0, 1, 2]. However, previous studies used Cas9, which just damages the DNA at the target site; this uses Cas12a2, which is far more destructive because it shreds the chromatin in the cell once activated by detecting the target sequence.

As with any cancer treatment, it's likely the tumor will evolve resistance. My guess is that cells will find ways to reject the lipid nanoparticles used to deliver the CRISPR/Cas mRNA and associated guide sequence(s), either via modifications to the cell surface (preventing LNP uptake) or via changes to endosomal/lysosomal pathways (causing the mRNA payload to get degraded before it has a chance to be translated into protein).

[0] https://pubmed.ncbi.nlm.nih.gov/28575452/

[1] https://www.nature.com/articles/s41598-018-30205-2

[2] https://www.nature.com/articles/s41467-020-18875-x

>[B]y far the most common reason given is that the cost of living has risen to such an extent that they feel that rearing children has become unaffordable.

That's certainly a factor, though very aggressive financial incentives for parents don't seem to work very well [0, 1, 2]. Not to mention that in rich countries, educational attainment and income are negatively correlated with fertility [3]. My theory there is that people's high-powered careers provide them more self-satisfaction than having kids.

>it's the fact that they can't even afford to buy somewhere to live.

It's funny you mention this. Some friends said they weren't having a second kid because they couldn't afford a three bedroom house, not realizing that kids sharing bedrooms was the norm for middle class families until very recently. Having one bedroom per kid was a luxury just 30-40 years ago.

>I won't comment on your assertion that the freedom to watch "adult-appropriate movies on a big TV in the living room" is a more fulfilling state of being than parenthood

It's not my assertion, it's something a couple deciding to not have another kid literally told me. They missed being able to have substantial amounts of adult time, and were actively counting down the days until their only child was old enough to amuse himself for long periods of time. Having another kid would reset that clock.

[0] https://www.economist.com/europe/2026/03/19/viktor-orbans-pr...

[1] https://www.theguardian.com/world/2024/feb/28/south-korea-fe...

[2] https://worldcrunch.com/culture-society/boosting-birth-rates...

[3] https://www.oecd.org/en/publications/society-at-a-glance-202...

As I wrote in another reply, I meant “hedonism” in its non-colloquial, neutral sense, i.e. the pursuit of individual pleasure and happiness above all else, which was a mistake on my part. My general point is that all the activities I listed (which only become abundantly available in rich, industrialized societies) yield more individual pleasure than having children.

Agreed with everything you wrote. I meant “hedonism” in its non-colloquial, neutral sense, i.e. the pursuit of individual pleasure and happiness above all else, which judging by the other replies was a mistake on my part.

>But having children is not intrinsically non-hedonistic. It's just one of many self-fulfilling activities we choose from

Exactly, and my point is that all the activities I listed (which only become abundantly available in rich, industrialized societies) are more self-fulfilling than having children.

I never made any value judgment on whether it’s bad or good. “Hedonism” is simply the focus of individual pleasure and happiness above all else, and everything I listed is an example of things that lead to individual happiness that are antithetical to having many children.

I'd call it hedonism if a couple wants to be able to go out on a date on a whim, easily take a vacation, watch adult-appropriate movies on a big TV in the living room, maintain good sleep/health habits, keep a flexible schedule unconstrained by school pickups/staying home with a sick kid, etc.

These are all real examples of why people I know delayed having kids, curtailed the number they had, or never had them altogether.

This has happened to every single society [*] as it industrializes [0], and offering extensive support and incentives to parents (e.g. as has been tried in Scandinavian countries) does very little to reverse this trend [1, 2].

My hypothesis is that as societies industrialize, they afford their population more and more activities that are simply more fun and rewarding than having children. So many people I know put off having children (or curtailed the number they had) because they were reluctant to give up the activities only available in a childfree/one-and-done life. Ultimately, we are hedonistic creatures, and having kids is antithetical to the myriad hedonic pursuits available in wealthy, industrialized societies.

[*] Israel is the lone exception, due to its Orthodox Jewish population.

[0] https://ourworldindata.org/global-decline-fertility-rate

[1] https://pub.nordregio.org/r-2024-13-state-of-the-nordic-regi...

[2] https://pmc.ncbi.nlm.nih.gov/articles/PMC10049131/

A modern take on Matthias Wandel's classic [0], which has you guess a variety of geometric attributes (e.g. angle bisection, centroid locating, shape regularization), not just simple partitioning of a line.

[0] https://woodgears.ca/eyeball/index.html

>[T]he same individuals who are able to make leasing decisions for office space are co-invested in commercial real estate

But those individuals are much bigger shareholders in the company they work for. Downsizing the office footprint of a single company has a tiny marginal effect on the overall real estate market, but can incur enormous savings for the company, and enormous personal rewards to the people making the decision to downsize and save money.

Individuals are notoriously bad at considering collective externalities. If a CFO realizes that they could save tens of millions of dollars per year on real estate by switching to remote-only, they’re not going to then think “but if everyone did this, it would hurt the commercial real estate holdings I have in my portfolio.” No, the CFO is thinking “announcing that we’re saving 8 figures per annum on real estate is going to pop the share price of my company (and thus my equity holdings), and likely net me a really fat cash bonus.” By similar logic, CTOs, who likely have considerable technology investments, would want to needlessly maximize tech spend. Instead, they get lauded for cutting IT costs.

I never understood this argument. Most companies do not own their office buildings, but rather lease space from corporate landlords. It is in the best interest of these companies to dramatically reduce their lease burden via WFH. Why would a company totally unrelated to real estate investment act against its own self-interest just to prop up real estate investors?

One of my favorite details of this movie is that the semi-antagonistic ENCOM executive Dillinger uses emacs [0], while Flynn uses vi. Clearly, the VFX artist who made the film's UNIX shells had a preference!

(Dillinger is also shown running "ENCOM Linux" -- is the VFX artist a BSD user? As he cycles through his buffers, we see a split second of `hanoi-unix`; definitely not the type to pay attention during boring board meetings!)

[0] https://www.youtube.com/watch?v=y-86iKkn6k0#t=3m55

Or that the notion of "Imaginary Property" was a seductive mirage to copyleft folks, many of whom have since realized that they do, in fact, value intellectual property rights in the wake of AI companies' free (as in freedom) use of other people's IP.

I stand somewhere in the middle: while our IP laws are far too restrictive, the folly of abolishing IP altogether has been effectively laid bare by AI companies.

Did You Say “Intellectual Property”? It's a Seductive Mirage. [0]

[0] https://www.gnu.org/philosophy/not-ipr.html

>There’s always The One Factory In North Carolina That Produces The Essential Ingredient and it turns out that it’s just the price optimal one and there is enough capacity around the world to substitute.

If you're referring to Spruce Pine in the aftermath of Hurricane Helene [0, 1], the predictions that chipmaking would be severely disrupted turned out to not come true because the Spruce Pine mine sustained a lot less damage than initially feared and was made operational within a week or two [2], not because high-purity quartz is commoditized.

[0] https://www.npr.org/2024/09/30/nx-s1-5133462/hurricane-helen...

[1] https://www.aveva.com/en/our-industrial-life/type/article/hu...

[2] https://www.cbs17.com/news/north-carolina-news/spruce-pine-q...

>Yes, because it's not an argument the letter is making.

It literally is though. The full quote from the Buzzfeed piece is:

  Reich’s claim that we need to prepare for genetic evidence of racial differences in behavior or health ignores the trajectory of modern genetics. For several decades billions of dollars have been spent trying to find such differences. The result has been a preponderance of negative findings despite intrepid efforts to collect DNA data on millions of individuals in the hope of finding even the tiniest signals of difference.

>The argument is in the second paragraph:

  But his skillfulness with ancient and contemporary DNA should not be confused with a mastery of the cultural, political, and biological meanings of human groups.

Reich never purports to make cultural or political arguments, just biological ones.

>When he says things like: But as a geneticist I also know that it is simply no longer possible to ignore average genetic differences among “races.”

Note that he put "races" in quotes. The point he was making here is that sometimes genetic ancestries can intersect quite well with traditional notions of "race" [0]. But often times they do not, especially in the case of admixed populations [1].

[0] https://www.nature.com/articles/s41598-023-32325-w/figures/1

[1] https://link.springer.com/article/10.1186/s12859-019-2680-1/...

The crux of that letter is the "need to recognize that meaningful patterns of genetic and biological variation exist in our species that are not racial." This is true. However, this does not mean that there aren't also meaningful patterns of genetic and biological variation that do stratify according to ancestry (not race!). The letter tries to handwave this away, claiming that "[f]or several decades billions of dollars have been spent trying to find such differences. The result has been a preponderance of negative findings despite intrepid efforts to collect DNA data on millions of individuals in the hope of finding even the tiniest signals of difference." This is simply not true, as studies like the subject of this discussion demonstrate.

The letter also states that "[t]he public should not cede the power to define race to scientists who themselves are not trained to understand the social contexts that shape the formation of this fraught category." Also true! This is exactly why Reich explicitly avoids discussing "races" but rather populations and ancestries, which are rigorously defined strictly in terms of genetics. With respect to population structures and ancestry, Reich is indeed an expert.

I'll add that very few of the signatories of that letter have any experience, let alone expertise in genetics. Here are the first few:

  Jonathan Kahn, James E. Kelley Professor of Law, Mitchell Hamline School of Law

  Alondra Nelson, Professor of Sociology and Gender Studies, Columbia University; President, Social Science Research Council

  Joseph L. Graves Jr., Associate Dean for Research & Professor of Biological Sciences, Fellow of the American Association for the Advancement of Science, Section G: Biological Sciences, Joint School of Nanoscience & Nanoengineering, North Carolina A&T State University, UNC Greensboro

  Sarah Abel, Postdoc, Department of Anthropology, University of Iceland

  Ruha Benjamin, Associate Professor, Department of African American Studies, Princeton University

  Sarah Blacker, Postdoctoral Research Fellow, Max Planck Institute for the History of Science, Berlin

  Catherine Bliss, Associate Professor, Social and Behavioral Sciences, UC San Francisco

Out of the 67 signatories, I counted approximately 5 who might have sufficient genetics expertise to offer a meaningful scientific counterpoint to Reich's work (this is being charitable, as I included titles like "Professor of Biological Sciences," which is no guarantee.) The rest were in fields like anthropology, sociology, law, and history.

>Reich in particular is borderline deliberate about attracting those sorts with his lab's research, because of how badly he chooses to handle the topic and terminology of race.

Sorry, do you have any examples? His views that I've read [0, 1] are scientifically rigorous and terminologically precise, deftly navigating the politics that some consider extremely controversial. To wit, one of my favorite passages from [1], which deals specifically with terminology:

   But “ancestry” is not a euphemism, nor is it synonymous with “race.” Instead, the term is born of an urgent need to come up with a precise language to discuss genetic differences among people at a time when scientific developments have finally provided the tools to detect them. It is now undeniable that there are nontrivial average genetic differences across populations in multiple traits, and the race vocabulary is too ill-defined and too loaded with historical baggage to be helpful. If we continue to use it we will not be able to escape the current debate, which is mired in an argument between two indefensible positions. On the one side there are beliefs about the nature of the differences that are grounded in bigotry and have little basis in reality. On the other side there is the idea that any biological differences among populations are so modest that as a matter of social policy they can be ignored and papered over. It is time to move on from this paralyzing false dichotomy and to figure out what the genome is actually telling us.

This particular passage is on p. 253 of [1], but everything in Chapter 11 ("The Genomics of Race and Identity," pp. 247-273) is well worth the read.

[0] https://www.nytimes.com/2018/03/23/opinion/sunday/genetics-r...

[1] https://sackett.net/reich_who_we_are_and_how_we_got_here.pdf

>It will be violence from the rich towards the poor, because democratic AI is not acceptable for them.

Unless the rich somehow manage to completely stifle the progress of consumer-level computing advancement (all chip manufacturers would just collude to quit selling to consumers?) and exert an iron-fisted control over the dissemination of software (when has this ever worked?), I'm not sure how they could control the democratization of AI.