<n>Below are links to articles, highlighted passages<fn>, and occasional annotations<fn> for the week ending on the date in the title, <a href="{app}/view_article.php?id=4085">enriching the raw data</a> from <a href="http://www.instapaper.com/starred/rss/1890855/5c1g08eoy9skhOr3tCGqTQbZes">Instapaper Likes</a> and <a href="https://twitter.com/mvonballmo">Twitter</a>. They are intentionally succinct, else they'd be <i>articles</i> and probably end up in the gigantic backlog of unpublished drafts. YMMV.</n> <ft><b>Emphases</b> are added, unless otherwise noted.</ft> <ft>Annotations are only lightly edited and are largely <i>contemporaneous</i>.</ft> <h>Table of Contents</h> <ul> <a href="#politics">Public Policy & Politics</a> <a href="#journalism">Journalism & Media</a> <a href="#labor">Labor</a> <a href="#economy">Economy & Finance</a> <a href="#science">Science & Nature</a> <a href="#climate">Environment & Climate Change</a> <a href="#medicine">Medicine & Disease</a> <a href="#art">Art & Literature</a> <a href="#philosophy">Philosophy, Sociology, & Culture</a> <a href="#technology">Technology</a> <a href="#llms">LLMs & AI</a> <a href="#programming">Programming</a> <a href="#fun">Fun</a> </ul> <h id="politics">Public Policy & Politics</h> <img src="{att_link}prjhrxa.jpg" href="{att_link}prjhrxa.jpg" align="none" caption="Barbarie vs. Civilisation" scale="100%"> <hr> <media href="https://www.youtube.com/watch?v=5EDKRGkgLsI" src="https://www.youtube.com/v/5EDKRGkgLsI" source="YouTube" width="560px" author="Al Jazeera English / Marc Lamont Hill" caption="Democracy doesn’t exist in the United States: Chris Hedges | UpFront"> I don't even know what to transcribe because, whenever Chris Hedges speaks, it's worth citing, and he speaks for nearly the entire 30 minutes, as Hill allows him to speak at length. This is an excellent distillation of the situation in the American Empire as it is, rooted in the historical context of both its own past, as well as similar contexts in Rome, Italy, and Germany. They discuss the failures of so-called liberalism at reasonable length. Hedges doesn't waste any time pretending that Trump isn't a threat but also doesn't waste time pretending that it starts with Trump---or that it would end with him. He talks about how Carter began the immiseration of the working class, with Reagan picking up the baton and taking credit for having begun it---and with Clinton having taking the machinery of Reagan and done even more downward-spiraling horrors with it. Maybe a short quote from <b>15:30</b>, <bq><b>Chris:</b> What was wokeness? Wokeness was---the corporations love it; they love it you know---is wokeness a woman CEO? No. It's about empowering working-class women. It's a complete inversion. <b>Marc:</b> Do you see wokeness as a kind of superficial approach to dealing with identity politics or do you see identity politics itself...? <b>Chris:</b> <b>I see identity politics as furthering the goals and the rapaciousness of the corporate state</b> [...] wokeness in the hands of the ruling class has been used as a cudgel to essentially <b>punish and scold the working class. And it is also about elevating their own status</b> [...]</bq> When Marc asked Chris about the quotation from scripture at Trump's inauguration at <b>19:45</b>, he responded that it was, <bq><b>Chris:</b> Idolatry. Moloch, Worshiping at the feet of Moloch. It's idolatry. It's heresy, It's the sacralization of human and political power, which is probably the greatest sin any religious institution can make. Look, the mega churches work like this---and I learned this from Hannah Arendt [...]---they are essentially equivalent of the so-called German Christian Church, established under the fascists in Germany where, on one side, you had the Christian cross and, on the other, the Nazi flag. And let's be clear, Marc, <b>this church is bankrolled by the very billionaire class that we talked about. Why? Because with <i>Magic Jesus</i>. you don't need labor unions; with <i>Magic Jesus</i>. you don't need health care; because <i>Magic Jesus</i> is going to give you a Cadillac and make all your dreams come true.</b> And that is a shift from a reality-based world into the world of magical thinking. And <b>once people shift into that world of magical thinking. you can't reach them through rational argument.</b></bq> <hr> <img src="{att_link}mr._fish_-_explore_gaza.webp" href="{att_link}mr._fish_-_explore_gaza.webp" align="none" caption="Mr. Fish - Explore Gaza" scale="75%"> <hr> <a href="https://exileinhappyvalley.blogspot.com/2025/01/its-mad-max-world-for-us-post.html" source="Exile in Happy Valley" author="Nicky Reid">It's a Mad Max World for Us: Post Apocalyptic Daydreams in an Age of Dystopian Crisis</a> <bq>Waking up whenever the daylight calls my name through the drapes of a bulletproof yurt, I’ll hit the vacant desert plains of Central Pennsylvania in a heavily modified, rust-rod, El Camino technical with my leather clad horde of genderfuck lesbian barbarians to <b>scavenge the radiated ruins of suburbia for coffee and gunpowder</b>…. With a Buck knife in my teeth and a sawed-off M1 Carbine strapped to my thigh, <b>I’ll scale the facade of an abandoned football stadium strangled by vines and hunt white tailed buck with a crossbow between the charred rush hour carcasses on Interstate 80 from a defiled billboard</b>…. I’ll cook raw flesh on a bayonet over a flaming television set while the blistering riffs of an all-Stooges mixtape crackle over a ghetto blaster and my <b>dreadlock-laden coven of heathen sisters howl for Loki at a moon populated by the corpses of billionaires who failed to escape their demons on luxury rocket ships</b>….</bq> <bq><b>I will own up to the fact that my lust for dieselpunk daydreams isn’t exactly the most constructive response to an era of unprecedented societal collapse, but I won’t apologize either.</b></bq> <bq><b>Your average American lives under a state of constant stimulation and constant surveillance.</b> When they aren’t struggling to pay off the debts of bourgeois degrees with borderline third world wages in the boiling kitchen dungeons of your neighborhood casual dining franchise, they are <b>burning through their meager wages on clickbait smartphone crack like Candy Crush and being hounded to buy more shit that no one needs by fifteen adds at once.</b></bq> <bq>This is your precious civilization. This is what 500 years of western enlightenment has brought us too [sic]. Morbidly obese voluntary enslavement at the barrel of a drone. And all it cost us was our tribes, our villages, our gods, our dignity, and our fucking ecosystem. <b>But I’m the sicko because I’d rather shoot cannibals in the face at the end of the world than vote for backstabbing social democrats and unionizing my cell block at the nearest cubicle colony?</b> Kiss my Unabomber reading faggot ass.</bq> <bq><b>We should work like hell to do the only thing that can possibly curtail the damage of global capitalism and that’s downsize</b>; decentralize, secede, drop out, rebuild locally autonomous village communities divorced from the restraints of big business and big government.</bq> <bq><b>I can’t tell you how the next movie ends. I can only tell you that the odds of it ending harmoniously are not in our favor</b>, but that doesn’t mean that there is no hope. It simply means that <b>our best hope rests in the survival of the small amidst the wreckage of the big.</b> If Furiosa can do that with one arm chained to God’s jawbone, then <b>the very least I can do is die dreaming while civilization’s useful idiots roll their eyes.</b></bq> <hr> <a href="https://arstechnica.com/tech-policy/2025/01/copyright-office-suggests-ai-copyright-debate-was-settled-in-1965/" source="Ars Technica" author="Ashley Belanger">Copyright Office suggests AI copyright debate was settled in 1965</a> <bq>"The Office concludes that, given current generally available technology, <b>prompts alone do not provide sufficient human control to make users of an AI system the authors of the output.</b> Prompts essentially function as instructions that convey unprotectible [sic] ideas," the guidance said. "While highly detailed prompts could contain the user’s desired expressive elements, <b>at present they do not control how the AI system processes them in generating the output.</b>"</bq> <bq>"In most cases," the Copyright Office said, "humans will be involved in the creation process, and <b>the work will be copyrightable to the extent that their contributions qualify as authorship.</b>"</bq> <hr> <a href="https://www.nachdenkseiten.de/?p=127923" source="NachDenkSeiten" author="Jens Berger">Zurück zum Atom? Energiepolitische Tagträumereien im Wahlkampf</a> <bq><b>Der chinesische Projektpartner ist übrigens das Staatsunternehmen CGN und ist fein raus, da man seine Beteiligung in Höhe von 7,1 Mrd. Euro in den Verträgen festgeschrieben hat und nur der französische Partner EDF und der britische Staat nachschusspflichtig sind</b>, wenn die Kosten steigen. EDF musste – auch wegen Hinkley Point, aber auch wegen gigantischer Verluste bei den französischen AKWs – 2022 verstaatlicht werden. <b>Allein im letzten Jahr schrieb der Konzern atemberaubende 12,9 Mrd. Euro für das Projekt Hinkley Point als Verlust ab.</b> Diese Verluste trägt am Ende der französische Steuerzahler. Geht Hinkley Point irgendwann im nächsten Jahrzehnt ans Netz, ist auch der britische Steuerzahler gefragt, da er dann ja die Einspeisevergütung finanzieren muss. <b>Der britische Rechnungshof geht dabei von Gesamtkosten für den Steuerzahler in Höhe von 199,7 Mrd. britischen Pfund, also rund 240 Mrd. Euro , aus – eine Summe, bei der einem ganz schwindlig wird.</b></bq> <bq>Schwindelerregend waren und sind auch die Kosten des einzigen Atomkraftwerks, das Frankreich in den letzten beiden Jahrzehnten gebaut hat – dem Block 3 des AKW Flamanville. 2007 begann man mit dem Bau und <b>bereits 2012 sollte Flamanville-3 ans Netz gehen und das bei „überschaubaren“ Baukosten in Höhe von 3,3 Mrd. Euro. Im letzten Jahr wurde der Block dann mit 12 Jahren Verspätung endlich in Betrieb genommen</b> und ab diesem Sommer soll Flamanville-3 auch kommerziell Strom einspeisen. Aus den ursprünglich geplanten 3,3 Mrd. Euro Baukosten wurde aber ebenfalls nichts. Ein Bericht des französischen Rechnungshofs weist die endgültigen Baukosten mit 23,7 Mrd. Euro – also dem Achtfachen – aus.</bq> <bq>Privatwirtschaftliche Unternehmen aus den USA und Europa sind entweder in die Insolvenz gegangen oder wurden verstaatlicht. Auch die Hersteller aus Japan und Südkorea werden von ihren Heimatländern quersubventioniert. Die russische und chinesische Konkurrenz ist vollständig staatseigen und hat für die beiden Staaten auch die strategische Aufgabe, Material für das Kernwaffenarsenal zu produzieren. <b>Die „wahren Kosten“ tragen auch hier die jeweiligen Staaten. Für Deutschland wäre es – Stand heute – gar nicht möglich, ein neues AKW zu bauen, da es kein einziges Unternehmen gibt, das diesen Auftrag ohne milliardenschwere Staatsgarantien und Subventionen übernehmen würde.</b></bq> <bq>Vor allem für die <b>USA</b> mit ihren 94 aktiven Reaktoren und <b>Frankreich</b> mit seinen 57 aktiven Reaktoren könnte dies ein großes Problem werden. In beiden Ländern befindet sich <b>kein einziges neues AKW in Bau</b>, während zahlreiche alte Reaktoren das Ende ihrer Laufzeit schon erreicht haben. Es ist davon auszugehen, dass die Ausfallzeiten dieser alten Reaktoren in den nächsten Jahren zunehmen werden und <b>einige Reaktoren sogar aus Sicherheitsgründen ganz vom Netz genommen werden müssen.</b></bq> <bq>Bei Thema Kernenergie setzt das BSW auf Forschung, lehnt einen Neubau nach jetzigem Stand der Technik und der zu erwartenden Kosten aber ab. SPD, <b>Grüne und Linke halten weder etwas von Kernenergie noch von einer Wiederaufnahme der Gaslieferungen aus Russland, würden mit ihrem energiepolitischen Programm die Strompreise also mittelfristig nicht senken.</b> Dafür plädieren diese drei Parteien für einen forcierten Ausbau der Regenerativen, was zumindest einen langfristigen Preiseffekt hätte, da dann die teuren Reservekapazitäten seltener zum Einsatz kommen.</bq> <hr> <a href="https://mattbivens.substack.com/p/trump-has-a-path-to-mount-rushmore" source="Substack" author=" Matt Bivens, M.D.">Trump Has a Path to Mount Rushmore</a> <bq>Work began in 1927 — which means that <b>in two years, Trump will oversee a Rushmore centennial. His brain is going to be on fire about this for the next two years.</b> Already during his first term, Trump told the governor of South Dakota that seeing his own face up there was his “dream.” His White House even queried the process for adding a face to Rushmore.</bq> <bq>[...] <b>former House Speaker Nancy Pelosi just a few months ago made a pitch for Joe Biden.</b> “<i>Such</i> a consequential president of the United States,” Pelosi solemnly told CBS journalist Lesley Stahl, “a Mount Rushmore kind of president.” Stahl laughed in her face. “Are you really saying that he belongs up there on Mount Rushmore?” While Pelosi sputtered, Stahl continued, grinning at the absurdity of it: “Lincoln — and <i>Joe Biden</i>?” <b>Trump, however, isn’t laughing. He’s sure he can cut a deal to get his likeness next to Lincoln’s.</b> He really wants this. His media allies are already publicly calling for it, and this weekend a Florida Congresswoman announced she would introduce legislation demanding it. So, this is coming, friends.</bq> <bq>Why else do you think Trump is talking about Greenland? This is his wheelhouse: a big real estate deal! <b>I guarantee that Trump thinks buying Greenland earns him Rushmore.</b> I say it doesn’t. But We the People need to get organized, fast. We may not have realized it yet, but the nation has entered into a negotiation with our new president over the price of putting his face on Mount Rushmore. Everything has a price — the price of a historic alteration (desecration?) of a renowned national monument has to be a world history-making achievement. <b>If Trump follows through on his recent call for “denuclearization” and succeeds in saving the world from an eventual-inevitable nuclear war — then maybe his face on Rushmore would be a fair trade.</b></bq> <bq>[...] <b>nuclear war was constantly on young Trump’s mind. In a long 1990 interview with Playboy Magazine, he sounded like the spokesman for an anti-nuclear peace group</b>:<bq>I’ve always thought about the issue of nuclear war. It’s a very important element in my thought process. It’s the ultimate, the ultimate catastrophe, the biggest problem this world has, and nobody’s focusing on the nuts and bolts of it. It’s a little like sickness. <b>People don’t believe they’re going to get sick until they do. Nobody wants to talk about it. I believe the greatest of all stupidities is people’s believing it will never happen, because ‘everybody knows how destructive it will be, so nobody uses [the] weapons’. What bullshit.</b></bq><b>Trump was right then, and he’s right today. The looming, ever-present danger of nuclear war is indeed the biggest problem this world has.</b> It is indeed a little like a sickness that people are in denial about. It is indeed the greatest of all stupidities to assume weapons poised on hair-trigger alert for launch within minutes magically won’t ever be used. It is indeed such bullshit.</bq> <bq>Many snarky profiles of Trump from the 1980s are unintentionally and revealingly poignant today. The interviewers want to talk about his wealth, his glitz, his antics — because that’s the only thing they’re interested in. Trump wants to talk about himself, too, because certainly Donald Trump loves himself some Donald Trump. But <b>even callow and shallow Trump is a deeper soul than his journalistic tormentors. Trump repeatedly brings up the same existential question of nuclear weapons that bothers so many of us today (and he gets relentlessly ridiculed for doing so).</b></bq> <bq><b>The national security state is associated with the Russiagate hoax against Trump; the absurd attempt to impeach Trump over a phone call to Ukrainian President Volodymyr Zelensky asking him to look into the Biden family’s openly corrupt dealings there</b> (dealings that President Biden just pardoned his family for on his last days in office); the 2020 influence operation to suppress and even outright censor discussion of the compelling evidence of that Biden family corruption in Ukraine;</bq> <bq><b>Success at nuclear disarmament will mean declaring war on the American national security state itself. Trump is already engaged in that war, if only to revenge himself for being outplayed during his first term.</b> He has also surrounded himself with some better people. His Middle East team is so much better that there is talk of a new Iran nuclear deal. His choice of Tulsi Gabbard to oversee the intelligence community has not yet been confirmed and is being fought against by panicked neocons. <b>Gabbard, like Trump, has been the victim of McCarthyist smears; she is a left-leaning politician, a former Bernie Sanders supporter, and a prominent and eloquent voice warning of the dangers of nuclear weapons.</b></bq> People give both Trump and Gabbard too much credit, as usual. I'm not virulently anti-Tulsi like so many others but her public statements are just as much of a mixed bag as Trump. I think it's nice to entertain the hope and notion that someone might, finally, start to reduce nuclear weapons again, and Bivens makes a strong case that a younger Trump talked about it a lot, and Trump did just mention it again on his first or second day in office---but he's mentioned a lot of stuff in his first couple of weeks in office. It's hard to separate noise from signal, and it's wildly naive to assume that the one or two decent things we've heard are the "real" ones. This is the kind of wishful thinking that allowed Obama to run roughshod over all opponents during his two administrations. <bq>Peace groups, and really all sensible people, need to tell Trump what we want. <b>We want an end to the endless wars, a draw-down of the overbearing and metastasizing national security police state — and abolition of nuclear weapons.</b> Get us that, Mr. President, and we can talk about a 60-foot tall granite representation of your face smiling down upon us for generations to come.</bq> <hr> <a href="https://scheerpost.com/2025/01/27/patrick-lawrence-trumps-failures-americas-failures/" source="Scheer Post" author="Patrick Lawrence">Trump’s Failures, America’s Failures</a> <bq>Russia’s economy is not failing. It is Europe’s economies that are failing in consequence of the sanctions regime the United States has imposed on Russia. Washington has no favors to offer Moscow. Given the progress of the war, it is the United States that is in need of a favor from Russia. <b>U.S. imports from Russia in 2022, the most recent year for which statistics are compiled, were $16 billion — taxi fare in the global trade context.</b></bq> <bq>Emmanuel Lévinas, a Lithuanian Jew who lived in France (1906–1995) and wrote in French, elevated these matters to an enduring discourse concerning the Self and the Other. <b>Indifference to others, he argued — and how radically must I simplify — lay at the root of the 20th century’s ills and evils. The cult of the individual, he posited (among a lot of other things) must be transcended in favor of relationships with all the Others among us.</b> We realize who we are only by way of these relationships; they are primary. “The Self is possible only through the recognition of the Other,” he wrote, a noted line. So, to continue my simplification: <b>We are social beings first; our individuality derives from our sociality.</b> Lévinas published <i>Totality and Infinity</i>, the book wherein he stated his case most fully and famously, in 1961.</bq> <bq><b>We long ago turned our insistence in our individuality into the “ism” of individualism</b>, an ideology that, however far it has taken America in the past, now proves a ball and chain at our ankles. Equally, America has had such power since the 1945 victories that <b>its policy cliques long ago lost interest in the perspectives of others</b>—how the world looks to them, their aspirations, their histories, all the rest. This is why, with admirable but few exceptions, America produces such poor diplomats. It has had no need of them. And <b>the policy cliques in Washington have not yet registered that we have in consequence already begun to fail.</b> And this is why, to finish off, Donald Trump thinks it is perfectly OK to declare his plans for <b>Canada, Greenland, and the [Panama] Canal</b> without so much as a preliminary consultation with a Canadian, a Dane or a Panamanian. These ideas are nonsensical to the point they embarrass. But, their loopy aspect aside, <b>are they any more nonsensical than — make your own list — Vietnam, Reagan’s invasion of Grenada, the Iraq War, Afghanistan, Syria, Ukraine</b>, indeed? Are they any more out of touch with the perspectives of others?</bq> <bq>I trace these, in spirit if not in declared fact, to <b>the Five Principles Zhou En-lai formulated in the early 1950s</b>, soon after adopted by the brand new Non–Aligned Movement.<ol>Respect for territorial integrity and sovereignty, non-aggression, non-interference in the internal affairs of others, equality and conduct for mutual benefit, peaceful co-existence</ol>I note that <b>the Chinese Foreign Ministry has now taken to stating these as the new world order’s rules of the road.</b></bq> <hr> <a href="https://www.the-hinternet.com/p/american-vacuum" source="Hinternet" author="Hinternet Editorial Board / Justin Smith-Ruiu">American Vacuum</a> <bq><b>When an adversary of the United States says “This territory matters to us for deep historical reasons”</b>, and the United States looks at that territory and sees only unexploited potential for developing resorts and casinos and airports with premium lounges for “Star Alliance” members and so on, then there really may be some <b>considerable risk of underestimating your adversary’s tenacity.</b></bq> <bq>We ourselves are at least partially sympathetic to at least some proposals for extreme alternatives to violent conflict — <b>it is after all both cheaper and far less violent to resettle every inhabitant of a conflict zone, and to do so with enviable condos with kitchens with modern amenities and all that, than to go on waging war.</b></bq> <bq>The one hitch, of course, is that <b>Gazans, like Russians, like Ukrainians, like Israelis, care deeply about history, which they conceive, as human societies always did</b> before the vacuum-packed Americans came along, as something like the temporal dimension of what is manifested spatially as land. We ourselves are torn about whether this ancient human representation merits respect, or at least so much respect as would permit it to continue indefinitely into the future.</bq> <bq>The philistine cannot accept that there should be such people, entirely outside of the goal-directed logic that necessarily governs most branches of public life — little pockets of pure freedom, where anything might happen.</bq> Not exactly, no. Some of us acknowledge that there can be special cases. We don't really like they to be self-selected. And who elected them king anyway?<fn> They survive not as parasites but because their society finds their contributions interesting. Otherwise no-one would feed or fuck them. <bq>In the future we may be able to replace war with side-taking bots that battle it out through automated contests of takesmanship, but if we fail to take advantage of the freedom such outsourcing will afford us <b>better to appreciate uses of language to speak of such things as how “the ghost of 'lectricity howls in the bones of her face”<fn>, then, well, sorry, but all our progress will have been for nothing.</b></bq> <bq><b>Just think about the last time you went to a gathering of your extended family.</b> Think of the least enlightened uncle or cousin in attendance, the one who believes the absolute dumbest things about the social good and how to attain it. <b>Does that person also perhaps excel at spinning yarns, at giving warm hugs, at mixing cocktails? Wouldn’t it be strange to insist on adding a b-moll to the warm thought you carry of them in your heart, simply because they are, well, kind of dumb?</b> How is it any less out of place to insist on doing that when it comes to, say, David Bowie, who is likewise kind of dumb, though certainly not as dumb as Bob Dylan — which is pretty close to another way of saying that he is not quite of the same rare caliber of artistic genius?</bq> <bq>We recently lost David Lynch, by <i>far</i> the dumbest American genius of the past several decades, and miraculously, so far, no one has wasted our time with any ham-fisted first-degree takes on his “politics”, with any tallying of his supposed deviations. <b>As far as we can tell, Lynch was universally adored.</b></bq> <bq>[...] [Lynch's] <b>work can only be understood to contain all things, and to show itself at its very best when these things are opposed to one another, mutually contradictory, impossible to hold together in the philistine mind — peace and violence, love and hate, suavity and monstrosity.</b> We suspect that he was an artist so great that even the philistines intuit, though they don’t have the language to articulate it, that they really just need to stay the fuck away with their usual retrospective assessments of the highs and lows of a public life. Who cares!</bq> <bq>Was Lynch, like Dylan, a product of “the capitalist PR machine”?</bq> In Chomsky's defense, that's not what he was saying. He was saying that Dylan wasn't part of the problem but neither was he part of any more just solution. Not everyone has to be, but some people have less time and use for those who aren't. Their lens is too tight. It's also a very fine line between being outside of the system and unknowingly and perhaps uncaringly serving it. Serving a distracting function is still serving a function. <bq>[...] as far as we can see deconstructionist theology is different — it is not a bunch of frivolous post-war French-theory goofballs getting a bit silly with serious questions like the existence of God. It is, rather, an insistence on continuity with what theological tradition had always held: that <b>God is beyond all mundane questions of existence or non-existence, and certainly beyond any of the deductive or inductive means by which you might prove the existence of anything else.</b></bq> <bq>As for Plantinga and the others, <b>we fear that for them the matter of belief will never move much beyond the epistemological frame in which one might also troubleshoot the repair of a malfunctioning A/C unit.</b> For that, too, is how you “do philosophy” in America. You “roll up your sleeves” and you solve some problems. You tell us what exists and what does not. You approach each day anew, full of pluck and determination, as if it were your first, and <b>you shrug with incomprehension when the “trolls” with strange accents begin their “rants” about all those best-forgotten things that happened a long time ago, before we’d figured out what matters.</b></bq> <hr> <ft>A play on <a href="https://www.elyrics.net/read/m/monty-python-lyrics/constitutional-peasant-lyrics.html" author="Monty Python" source="...and the Holy Grail">Constitutional Peasant</a>, <bq><b>Arthur:</b> Be quiet! I order you to be quiet! <b>Dennis' Mother:</b> Order, eh? Who does he think he is? <b>Arthur:</b> I am your king! <b>Dennis' Mother:</b> Well I didn't vote for you.</bq></ft> <ft>From <a href="https://songmeanings.com/songs/view/72949/" author="Bob Dylan" source="SongMeanings">Visions of Johanna</a>.</ft> <h id="journalism">Journalism & Media</h> <a href="https://www.counterpunch.org/2025/02/01/pick-up-the-pieces/" author="Jeffrey St. Clair" source="CounterPunch">Pick Up the Pieces</a> <bq><b>A new Lancet study says the life expectancy of Palestinians in Gaza has nearly been cut in half by Israel’s genocidal assault, falling from a pre-genocide average of 75.5 years to 40.5 years by Sept. 2024.</b> These are conservative estimates because the Lancet study didn’t account for indirect deaths during the war from malnutrition, hypothermia, and lack of access to essential medical care. Gazans are returning to <b>an unfathomable amount of rubble (50.7 million tons, according to the latest UN estimate), much of it containing hidden dangers, such as human remains, toxic waste, asbestos and unexploded ordinance.</b> The debris generated by the war in the Gaza Strip is <b>17 times more than the combined sum of all debris generated by other Israeli military operations in Gaza since 2008.</b> On average, there are over 365 kilograms of debris for each square kilometer of land.</bq> <bq><b>The Gaza ceasefire took effect on January 19.</b> But the killing didn’t stop. As I reported last week, 24 Palestinians were killed over the next two days in Rafah alone. <b>Over the following seven days, another 193 Palestinians were killed and 397 injured. No Israelis have been killed during the truce.</b></bq> As predicted by Chris Hedges. <bq>On Sunday morning, <b>they took me from my cell for questioning by Swiss defense ministry intelligence agents without the presence of my lawyer</b>, and they again refused to allow me to contact her or my family. <b>I refused to talk to them without my lawyer and told them to take me back to my cell.</b> During my imprisonment, I refused every meal and every cup of coffee or tea they offered me except the last meal after I knew I would be going home. I accepted only water, which is the right of every human being. All of this was after <b>I was abducted off the street around 1:30 pm on Saturday while on my way to the Palestine teach-in by undercover agents, handcuffed, forced into an unmarked car, and sped straight to the prison.</b> My “crime”? Being a journalist who speaks up for Palestine and against Israel’s genocide and settler-colonial savagery and those who aid and abet it. <b>I came to Switzerland at the invitation of Swiss citizens to talk about justice for Palestine, to talk about accountability for a genocide in which Switzerland too is complicit.</b> But while I was hauled off to prison like a dangerous criminal before I even had a chance to say a word, the <b>Israeli president Isaac Herzog, who declared at the start of the genocide that there are no civilians in Gaza, no innocents, received a red carpet welcome in Davos</b>, a carpet soaked in the blood of the more than 47,000 known victims of the genocide and the thousands more still under the rubble, or who died of deliberately inflicted starvation and denial of medical care. And on this very day, <b>Netanyahu freely travels to Poland to make a mockery of the Auschwitz commemoration despite an outstanding ICC arrest warrant.</b> That is the perverse, unjust world we live in. </bq> <hr> <a href="https://old.reddit.com/r/YesAmericaBad/comments/1ijogx1/why_oppressed_americans_dont_care_about/" author="" source="Reddit">Why oppressed Americans don't care about nationalistic narratives that call on them to have solidarity with their oppressors against their oppressors' enemies</a> <img src="{att_link}useless_eater.jpg" href="{att_link}useless_eater.jpg" align="none" caption="Useless eater" scale="75%"> They may be bad in other ways, but you've personally treated me and everyone around me like shit, so there's no way that I'm going to side with you. I also think it's cute how she pops up the Chinese flag. <hr> <a href="https://podcasts.apple.com/us/podcast/behind-the-news-1-30-25/id73801817?i=1000687303052" author="Doug Henwood" source="Apple Podcasts">Behind the News 1/30/25</a> It was pretty shocking to be exposed to Doug's superficial and ignorant broadside against Glenn Greenwald and Matt Taibbi, who he caricatured after having almost certainly not watched, listened, or read anything that they've produced in a decade. This doesn't stop him from judging them as right-wing fanatics cozying up to power, and to platforms like Substack and Rumble as right-wing havens. This is the world of the upper-west-side New York liberal, completely siloed and masturbatory. I was shocked to hear Henwood so deep in the tank for this completely unsubstantiated myth. This is just stupid in-fighting based on ignorance. There is a lot of leftism in Taibbi and Greenwald's work---but none of the unwavering support for the Democrats that apparently still forms the base of Henwood's ideology. A pity. I expected better of him. Henwood's interviewee Eoin Higgins was more fair than Doug, but the premise of his book is that everyone on the left has been bought up by the right. They spend more time defending Rachel Maddow's creep to the right just because she's ostensibly "on their side." These fucking people really can't get out of their own way. Eoin even says that Matt's premise the Maddow is as bad as Tucker Carlson is mostly correct. Henwood is fucking relentless, though. It's interesting that these two fail to be able to see that part of the problem that Greenwald and Taibbi are noting is ... them: Henwood and Higgins literally can't tell that their inability to notice as much wrong on the MSNBC side as on the FOX side is definitely a problem. Don't bother listening to this podcast episode. Henwood's a raging, stupid, petty asshole in this one. It just gets more and more disappointing. He keeps calling Matt Taibbi right-wing. I don't think he's actually read a word of what Taibbi's written in years. This is what it sounds like when high priests tell their flock about who the apostates and heretics are. Henwood expresses a belief in a very clear world of black and white, red and blue, if-you're-not-with-us-you're-against-us mindset that surprised me quite a bit. Matt Taibbi is not an anti-vaxxer. For fuck's sake. Jesus Christ, they're talking about something called "Substack brain," which is a completely invented phenomenon. They just literally call anything that's not controlled and censored by the elites to which they kowtow "right wing". It's disgusting. Eoin manages to disparage the reporting in the Twitter Files a few times. Even though he and Henwood don't take the accusations in the Twitter Files seriously---how could they? The Twitter Files are mostly about how a Democratic-run administration absolutely gutted the first amendment and how could they believe something like that? Or even care about it when their intellectual leaders have ordered them not to?---the government is taking it seriously. The wheels of justice turn slowly. <h id="labor">Labor</h> <a href="https://freddiedeboer.substack.com/p/we-have-to-take-some-kind-of-an-l" author="Freddie deBoer" source="Substack">We Have to Take Some Kind of an L on Immigration, For Now</a> <bq>I am an internationalist, which is to say that I don’t respect the concept of country. As shorthand I sometimes refer to myself as an open borders guy, but this isn’t quite right, as I am in fact a no-borders guy, in common with people from my political tradition. <b>The nation-state is a fiction, and a very recent one, invented for the benefit of capital and imperialism. As such, in my ideal world we’d take in whoever wants to live here; indeed, there would be no formal legal difference between “here” and “there.”</b></bq> <bq>[...] you have to recognize that <b>a permanent system of endless mass migration looks to most like a huge burden for the receiving countries, a potentially catastrophic situation for the countries of origin, and chaos for everyone.</b> This is just to say that you can be an open borders type like me and still understand that <b>“All the poor people can move to the rich countries” is not a sustainable solution for anyone. The rich countries would cease to be rich and the poor countries would be totally devastated.</b></bq> <bq>The fact that we have created a system in which undocumented people work as racially-and-linguistically marginalized Morlocks for the good of the American rich is ugly and awful and not something any progressive person should be defending. [...] I hear this claim from supposedly-progressive people, <b>the idea that undocumented immigrants are good because our employers can exploit them, and it drives me insane. Lefty people support labor protections and don’t celebrate when people break them.</b></bq> <h id="economy">Economy & Finance</h> <media href="https://www.youtube.com/watch?v=qj-gAUaUF-4" src="https://www.youtube.com/v/qj-gAUaUF-4" source="YouTube" width="560px" author="Jonathan Pie" caption="The Corporate Con."> <bq>most of our public services are now owned by private companies whose main purpose---and, in most cases, <i>only</i> purpose---is to make profit. They don't work for you or the government or the council. <b>They work for shareholders and nobody else. And it's a pretty good system, if <i>you own shares in that company</i>.</b> Privatization! It was sold to the nation as giving us more choice as consumers. And there is a choice: <b>get fucked hard over a barrel with zero lube or go without heating and water. There's your choice.</b></bq> <bq>All they talk about is growth and growth and business and growth, with zero acknowledgement that <b><i>we live in a corrupt corporate state</i>, which is not an <i>inevitability</i> but a <i>political choice</i>. And not doing anything about it <i>is also a political choice</i>.</b></bq> <bq>You can't do much but you better do something about this unbridled corporate greed and do it fast, <b>cuz there's a dead-eyed, populist, GB-news, lopsided minge just one election away.</b></bq> <hr> <a href="https://www.bloomberg.com/opinion/articles/2025-01-28/deepseek-disruption-has-its-upside" source="Bloomberg" author="Matt Levine">DeepSeek Disruption Has Its Upside</a> <bq>There is, however, a much funnier approach. The approach is:<ol>Build a good AI model that can compete with the leading large language models built by tech giants, but cheaply, with fewer and less sophisticated chips and less electricity. <b><i>Sell short</i> the stocks of the tech giants with expensive AI models</b>, and the big chipmakers, and electric utilities and everyone else who is exposed to the “AI is a gusher of capital spending” trade. <i>Then</i> announce your cheap good open-source model. <b>Wipe out almost $1 trillion of equity market value, and take some of that for yourself.</b></ol></bq> <bq>“Disruptors can profit not by selling their product more cheaply than incumbents, but by giving it away after shorting the stock of incumbents,” something like that. You don’t see it a lot in practice, in part because people seem to find it icky and in part because they just don’t think of it. <b>There is some psychological incongruity between trying to build world-changing products and trying to find short trades.</b> Most founders of disruptive consumer technology startups are not hedge fund managers! But Liang is. <b>It just seems like a missed opportunity if he doesn’t own a pile of Nvidia puts.</b></bq> That is an incongruity. The assumption is that the people working in AI are trying to make world-changing products. If we were to suppose that they were more interested in making a pile of money for themselves in the short run, then things make a lot more sense. <bq>“It is difficult to know exactly how to make money in AI” does seem like an essential aspect of the AI trade; we have talked about OpenAI’s claim that “it may be difficult to know what role money will play in a post-[artificial general intelligence] world,” and also about <b>a venture capital bet that the way to make money on AI is by buying up homeowners’ association management companies.</b> But the actual answer turns out to be “build a cheap AI model and short Nvidia.”</bq> <bq>How do you buy groceries? Well, classically, <b>a bank will lend you a bunch of money secured by your stock, at a fairly low interest rate</b> (because the loan is pretty safe), and you use that money to buy groceries and houses and yachts and things. <b>The loan is not a taxable event. You never really have to pay it back: You just borrow more to fund your lifestyle</b> (and to pay the interest on the loan), and if your company keeps succeeding, the loan compounds at a lower rate than your stock. You keep this up for the rest of your life, and then when you die <b>your heirs can sell some of your stock to pay back the loan. And they don’t pay any tax on the stock sale</b>, because they get a “basis step-up.” This strategy is sometimes called <b>“buy, borrow, die.”</b></bq> <bq>And so you want to find some situation where tax accounting (how much income you have to report to the Internal Revenue Service) differs from GAAP accounting (how much income you have to report to your shareholders). <b>Specifically a situation where your GAAP net income is high and your tax net income is low. I will not describe these situations — ask your accountant! — but I assert that many of them exist and that finding them is a lucrative business.</b></bq> <hr> <a href="https://www.bloomberg.com/opinion/articles/2025-01-27/hedge-fund-ai-is-cheap-ai" source="Bloomberg" author="Matt Levine">Hedge Fund AI Is Cheap AI</a> <bq>[...] nobody learns the essential skills to build fusion reactors or perform brain surgery at a hedge fund. It is perhaps a fortunate coincidence that maybe the highest-profile generally useful technology these days is the sort of thing that you can learn at a hedge fund. But maybe not. <b>Maybe that’s just path dependency. Perhaps someone would have built, like, a Star Trek replicator by now</b> — an even better genie! — <b>if the smartest technically inclined people had spent the last 30 years going into, uh, particle physics or whatever that is.</b> But instead the draw was hedge funds and the output was large language models.</bq> Exactly what I was going to say. Brain drain is real, and it's one of the biggest problems that the west and its acolyte nations has. The people most capable of providing value to society are drained away into building their own personal fortunes instead. <bq><b>Financial alchemy has turned a pest control business into a high-finance operation</b> that can attract HBS graduates and get financing from sophisticated investors.</bq> Finance is so fucking enamored with itself that it can't even be seriously self-critical enough not to metastasize. It's mostly useless but it thinks it's indispensable. <bq>[...] the trade is that investors give the AI some money, it uses the money to buy the pest control company, [...]</bq> What nutsy fantasy world is this? All of these schemes sound like Golgafrinchan leaves. <bq><b>The thesis of an index fund is that a lot of smart hard-working people compete to pick the best stocks.</b> As a product of their competition, all the stocks are pretty fairly priced; if they weren’t, those smart hard-working competitors would buy the underpriced ones and sell the overpriced ones. So <b>you can just opt out of that competition: You buy all the stocks at their market prices, trusting that everyone else's hard work will make the prices right.</b> And so you will be able to buy long-term economic growth at the right price. <b>This is nice for you, but if everyone did it, it would be bad: There’d be no one to make the prices right.</b></bq> <bq>Bloomberg’s Katherine Doherty reported Friday : For the first time on record, <b>the majority of all trading in US stocks is now consistently occurring outside the country’s exchanges,</b>, according to data compiled by Bloomberg. This off-exchange activity — which <b>happens internally at major firms or in alternative platforms known as dark pools</b> — is on course to account for a record 51.8% of traded volume in January.</bq> <hr> <a href="https://www.hamiltonnolan.com/p/on-having-a-maximum-wealth" source="How Things Work" author="Hamilton Nolan">On Having a Maximum Wealth</a> <bq>The single most ridiculous aspect of human history is how much of it has been <b>driven by the goal of allowing a tiny portion of a large population to live in luxury.</b></bq> <bq>I think that every once in a while it is well worth taking a moment to gape at the basic <i>ludicrousness</i> of this fact. As societal goals go, an honest reading tells us that we are often not aiming for “better technology” or “philosophical progress.” No, the reality is that, thousands of years and around the globe, <b>the primary purpose of all the work that everyone is doing is “allowing a few jerks and their unbearable kids to live lavishly.”</b></bq> <bq>If nothing is done to prevent it, the gains of AI will accrue to a small pool of already wealthy investors and tech company executives, at the cost of countless normal people losing their jobs and becoming worse off. This is why <b>the relevant discussion about AI is not really one of technology, but rather a political one: How will the gains of this technology be distributed?</b></bq> <bq>This is the deal that we already have for the gains of the past 30 years of tech advancement, by the way. Jeff Bezos gets a hundred billion dollars and you get an easy way to order toothpaste. Elon Musk gets four hundred billion dollars and you get a neat car you can buy. This is the standard offer of capitalism. <b>The population at large is supposed to be satisfied with the incidental benefits of the technology itself, as those who control the technology ensure that it is deployed in service of maximizing their own wealth.</b></bq> <bq>It is easy to imagine Facebook as a useful public utility; instead, it is a trashy den of extremism and slop. That is because <b>the company makes its decisions in service of making money rather than in service of being useful to the public.</b> Despite the fact that Mark Zuckerberg has a net worth of more than $200 billion, far more than his descendants could ever spend in several generations, <b>he continues to make his primary product worse because it will make him more money.</b></bq> <bq><b>Debating the appropriate number is what happens after you accept the underlying premise</b> that there should in a fact be a wealth limit.</bq> You have to accept that, just like the minimum is set at what we can afford, so, too, the maximum wage would be. Once you agree that a limit is needed---examples abound---you're discussing price. <bq>Such a limit on net worth would <b>eliminate the incentive of every single tech CEO, already rich, to get richer.</b></bq> <bq>The drudge work that it automates could be <b>paid back to the public in the form of shorter hours for the same pay, rather than having those gains taken by CEOs and investors</b>, while workers were stuck with fewer jobs.</bq> <bq>Simply capping how rich people are allowed to get could radically reorient the goals of society by removing <b>the (insane!) thing that our mighty corporations now work towards like insatiable robots of doom.</b></bq> "Simply." <hr> <a href="https://wallstreetonparade.com/2025/02/trumps-hedge-fund-guy-is-now-overseeing-the-u-s-treasury-irs-occ-u-s-mint-fincen-f-soc-and-the-consumer-financial-protection-bureau/" author="Pam & Russ Martens" source="Wall Street on Parade">Trump’s Hedge Fund Guy Is Now Overseeing the U.S. Treasury, IRS, OCC, U.S. Mint, FinCEN, F-SOC, and the Consumer Financial Protection Bureau</a> <bq>What did Bessent do previously to qualify for this powerful position? He ran a hedge fund, Key Square Capital Management LLC, with 25 employees. But, more important to the transactional world of Donald Trump, Bessent gave $1.25 million to PACs supporting Trump and tens of thousands of dollars to state and national Republican parties and candidates.</bq> Look, this isn't great. But ... honestly? I'm happy if some stumblebum who'd only run a hedge fund with 25 employees is in charge of Treasury. Maybe he won't be able to get out of his own way. Will he really be worse than Janet Yellen, who never saw a war she didn't want to finance the shit right out of? Or Larry Summers, that Jabba-the-fucking-Hut sonofabitch who's so elitist that he makes most other elitists look like sons of the soil? Let's see, who else we got? Hank Paulson? Tim Geitner? Robert Rubin? Steve Mnuchin? All craven creatures of Wall Street. Bessent is not going to be any worse than any of those buffoons, who worked as hard as they could to funnel money upward, toward themselves and their ilk. Bessent will do the same but our best hope is that <i>he will be more incompetent at it</i>. <hr> <media href="https://www.youtube.com/watch?v=j1zMA_vlHVw" src="https://www.youtube.com/v/j1zMA_vlHVw" source="YouTube" width="560px" author="Evan Edinger" caption="The Dark Truth About Tipping in America"> I learned quite a few things about tipping in the U.S. <ol>The federal minimum wage for tipped workers was legally set at 1/2 of the federal minimum wage for everyone else in the 1970s. The law was originally set to keep the federal minimum wage for tipped workers at 1/2 of the federal minimum wage in perpetuity. In the early 1990s, the law was changed to lock in the federal minimum wage for tipped workers at $2.13. It has not moved in over 30 years. Almost every state enforces tip-sharing, which includes sharing tips out to the people working federal-minimum-wage jobs. The sharing extends to all employees, including managers. More than 1/2 of your tip in those states does not go to your server; most of it channels through the management of the restaurant and you can only hope that it trickles back down to the person for whom it was intended. The law is that, if a server does not earn the federal minimum wage based on their wages and tips, then the employer is <i>legally obligated</i> to make up the difference. Over 90% of employers fail to do so. Studies have shown time and again that nearly all restauranteurs are breaking this law in a grand wage-theft. Generally, nothing happens outside of some minor fines. That means that, when you tip someone, you can only psychologically feel that you're "helping" the person who served you if you <i>have no idea how the system actually works.</i> What you're actually doing is paying the restaurant more money. You are not even filling the wage-gap <i>for that worker</i>. This applies even more if you're using an electronic payment system, in which the money doesn't even spend a single second with the server. Instead, it goes straight to the restaurant's account and it would have to voluntarily give that money back to its servers. Given the massive fraud and wage-theft in that industry, this is incredibly unlikely to happen. I would imagine that this applies almost equally to Europe and Switzerland, where there are <i>no laws</i> about tips being applied back to workers. When you "tip" with an electronic-payment mechanism in Switzerland, you're almost certainly just paying more for your meal, a gift that the restaurant is more than happy to just scoop up. I don't even know that there is a legal framework in Switzerland to determine what a business is obligated to do with the extra money that you give it for "tips". Imagine if you just paid CHF5.- more for a T-shirt than it costs. Do you think that that money actually ends up with the person running the cash register that day? How would it? Why would it? The business would probably just pocket it. You might as well just put it in the mailbox. I wonder the same now about tips in restaurants. It's a whole fictitious mechanism whose functioning the way you think it does is contingent on the theory that the same business that is happy to pay its workers $2.13 per hour and happy to otherwise engage in wage-theft would do everything required to ensure that the extra $5.73 that you paid with your credit card will actually end up with the person who served you. Even thinking that they would bother to do that seems increasingly unlikely, especially with oversight being so shoddy. In the states, there are laws. In Europe and Switzerland, there are labor laws, but I wouldn't even be so sure that there is oversight for a system like tipping that people just kind of start doing. Why would there be? Tip in cash if you're going to tip. At least this gives the server the opportunity to collect it for themselves---and to not declare or under-declare it on their taxes, so that Uncle Sam doesn't get his filthy hands on it. </ol> <hr> <a href="https://coreyrobin.com/2025/02/05/it-was-never-going-to-be-the-proud-boys/" author="Corey Robin" source="">It was never going to be the Proud Boys</a> <bq>For the end has turned out to be at it was at the beginning: Fear, American Style, <b>where millions of people fear that they will be punished by losing their jobs if they dare speak up or or speak out, is the reigning principle of the moment.</b></bq> I am so frustrated by otherwise salient commentators who seem to legitimately believe that all of this evil started 2.5 weeks ago, when Trump was inaugurated as president of the U.S. Mr. Robin, what do you think people where doing in the government when they sat on their hands about contradicting the prevailing COVID orthodoxy? What do you think they were doing as the previous administration ran a genocide for 15 months? Do you think that millions of people were all cheerily speaking their minds, secure in the knowledge that they would never lose their jobs in retribution? Did you not hear about any of the whistleblowers? Did you not read about the exodus of Biden administration employees who ended up quitting because their voices weren't being heard? This is not new. It damages your argument nearly beyond repair when you pretend that it is, or when you couch your argument in terms that allow your readers to assume that it is. <bq>I wonder if the analogy we might be looking for can be drawn from a combination of two movies from the 1980s: Wall Street and Die Hard. I keep thinking, as I read the press, that what Trump and Musk are doing looks like the hostile takeovers that used to be part of the cultural landscape in film, where corporate raiders would seize control of a company, strip it of labor and for parts, and leave just a battered shell, the ghost towns of deindustrialization in the 1970s and 1980s. <b>This is one big massive project of deindustrialization of the national government.</b> The other film, as I said, is Die Hard, where very <b>savvy criminals use the optics of political terrorism as a cover for their real racket of stealing a lot of money for themselves.</b></bq> I'm not sure if Corey intends for this to also be seen as a new development, or whether he was simply unable to see it---or write about it---before Trump showed up on the scene, but <i>this is how it has been for at least 40, if not 50, years.</i> They do this <i>everywhere</i>. They do it in America. It's just that, places like the former Soviet Union or countries in the developing world aren't the <i>center of the empire</i> and are thus not continually replenished by plunder. The plunder in American has perhaps never been as evident to many people because the level in Dagobert's money tank only went down slowly, as more and more gold coins were constantly being added. It was also the case that prior administrations didn't make it so obvious what they were doing. Corey shouldn't be lauded for finally understanding how the system works. He should be disparaged for having allowed himself to so easily be deluded into thinking that it wasn't working like this when oligarchs more politically adjacent to him were running the show. <bq quote-style="none">Marx’s daughters loved to play the game “Confessions,” which we now call the Proust Questionnaire. They once got their father to play it with them and answer its questions. Almost two hundred years later, his responses resonate. <b>What is your idea of happiness? “To fight,” said Marx. What is your idea of misery? “Submission,” said Marx. What is the vice you detest most? “Servility,” said Marx.</b></bq> F'in' A, Marx, F'in' A. <h id="art">Art & Literature</h> <a href="https://www.theparisreview.org/blog/2025/01/29/the-last-day-of-his-life/" source="The Paris Review" author="J.D. Daniels">The Last Day of His Life</a> <bq>And Aeschylus, in Seven against Thebes, has Eteocles tell the chorus: “<b>The gods, I am sure, have already ceased to think of us. The offering they desire from us is that we die.</b>”</bq> <hr> <media href="https://www.youtube.com/watch?v=RdcSFsQRsnc" src="https://www.youtube.com/v/RdcSFsQRsnc" source="YouTube" width="560px" author="CinemaStix" caption="when you watch the movie with all the exposition removed"> This is a great 12-minute video that shows what <i>The Bourne Identity</i> would be like if it had been included only scenes from Bourne's point of view, where the audience learns about Bourne at the same speed that he does. It argues that the movie would work very well without the interstitial scenes explaining everything to us about the Treadstone Project <i>before</i> Bourne finds out about it. This is pretty great. <hr> <media href="https://www.youtube.com/watch?v=CTi08kOgPU0" src="https://www.youtube.com/v/CTi08kOgPU0" source="YouTube" author="Evan Edinger" caption="9 words I’m surprised Brits pronounce differently" width="560px"> Evan discussed the difference between how the British and Americans pronounce "patent". He mentioned that "patent" means to "lay open" in Latin and then thought it strange, since patents "hide" something. That is a common misconception about how patents work. When a legal entity (person or company) is granted a patent, it exchanges the opening of the design for a government-granted monopoly on more-or-less exactly that design. The government is the final arbiter on whether a design _based on_ that design is "patent-infringing" and subject to retraction and fines, or whether it is considered sufficiently different to be its own thing. Why would a society have patents? They're ostensibly to encourage inventors to share their ideas publicly so that others can learn and improve on them, while also encouraging inventors to keep inventing because they get a window during which they can recoup their up-front investment of capital and time. That was the original idea, anyway. It's been somewhat perverted by now. Still, patenting something does "lay open" its design. A bit later in his video, I was barely able to hold myself back from going down the rabbit hole of <a href="https://en.wikipedia.org/wiki/Linking_and_intrusive_R" source="Wikipedia">non-rhoticity, non-linking and linking Rs</a>. <h id="philosophy">Philosophy, Sociology, & Culture</h> <a href="https://www.the-hinternet.com/p/a-note-on-literacy-from-brother-martin" source="Hinternet" author="Sam Kahn">A Note on Literacy from Brother Martin XoPo</a> <bq>[...] literacy always was a rarer skill than commonly believed. Your noted literary scholar-zoologist, Maryanne Wolf, in contends that <b>reading is an extraordinarily unnatural endeavor requiring virtually the whole of childhood to pursue.</b> “Every child in every generation must do a lot of work,” she writes.</bq> <bq>[...] the masses found themselves with far more engaging forms of entertainment — these were pictures without sound, and then treacly films made for cynical profit motives, and then treacly TV shows with advertising breaks sometimes longer than the programs themselves, and then shoot-‘em-up video games of no edifying value whatsoever, and then private screens allowing you to follow the highly curated, highly mendacious feeds of casual acquaintances and minor celebrities. <b>Every one of these was vastly more interesting and entertaining to you than the wisdom of the ages or the outpouring of another’s soul.</b></bq> <bq>[...] information was so readily available between the video and the podcast that <b>it started to seem like a waste —especially to the terrestrial children themselves— to dedicate all of childhood to reading uselessly when that time could more profitably be spent learning Python or developing an Instagram following.</b></bq> <bq>[...] by mid-Musk, even <b>professors —knowing that their students could no longer read— simply recorded lectures on relevant material, which their students listened to at 1.5x speed.</b></bq> <bq>The encouragement I have is that you are doing more or less the right thing given the inevitable collapse of literacy. You are not mourning it too much. You understand that, <b>for many people, emojis are easier than letters and voice memos more comfortable than text messages</b>, that it can in fact be a smart adaptation to dedicate one’s childhood to videos and online chat as opposed to reading dense texts — but that <b>there are those who find reading and writing valuable for their own reasons and that they should band together.</b></bq> <bq>The warning is that if continuity is lost, it really is irreparable, that the time coming up for you will be dark and difficult, that <b>the entire treasure house of the Analog Age —you might call it the Age of Literacy— is at risk of being forgotten and deemed irrelevant, and that it will be an immeasurable catastrophe if that does happen.</b> Whether it does or not will be a close-run thing. There is no posterity to help you —<b>we in the future are just as vapid as you are, if not more so—, and we are perfectly willing to forget you if you do not leave anything memorable for us.</b> Whether you are forgotten or not is, in the end, on you.</bq> <hr> <a href="https://old.reddit.com/r/LateStageCapitalism/comments/1ifkemg/the_absolute_state_of_ethics_under_capitalism/" author="Corey Mohler / Existential Comics" source="Reddit">The absolute state of ethics under capitalism</a> <img src="{att_link}is_it_ethical_to_hoard_beard_when_families_are_starving_.webp" href="{att_link}is_it_ethical_to_hoard_beard_when_families_are_starving_.webp" align="none" caption="Is it ethical to hoard beard when families are starving?" scale="50%"> <bq>Why are ethics questions always like: "is it ethical to steal bread to feed your starving family?" And not: "is it ethical to hoard bread when families are starving?"</bq> <hr> <a href="https://www.caitlinjohnst.one/p/the-tiger-show" author="Caitlin Johnstone" source="Substack">The Tiger Show</a> <bq>Nowhere is it inscribed upon the fabric of reality that <b>control of the world must be ceded to the dumbest, crudest and cruelest among us.</b> Nowhere is it written in adamantine that <b>our happiness and the fate of our biosphere must depend on what the market will bear.</b></bq> <bq>We have galaxies and sorcery roiling within us. We have barbed wire wings beneath our sundresses waiting to be unfurled. <b>The tiger show is over when we decide it is over.</b> <b>They can lock up Luigi, but they can’t lock up everyone.</b></bq> I just finished watching <i>Gandhi</i> last night. His philosophy was similar but without the violence. They can't lock up everyone. <hr> <a href="https://www.thenationalliteracyinstitute.com/post/literacy-statistics-2024-2025-where-we-are-now" author="" source="The National Literacy Institute (U.S.)">Literacy Statistics 2024- 2025 (Where we are now)</a> <bq><ul>On average, 79% of U.S. adults nationwide are literate in 2024. <b>21% of adults in the US are illiterate in 2024.</b> <b>54% of adults have a literacy below a 6th-grade level</b> (20% are below 5th-grade level). Low levels of literacy costs the US up to 2.2 trillion per year. 34% of adults lacking literacy proficiency were born outside the US.</ul></bq> Below a sixth-grade level means that people can't read their taxes, they can't read end-user license agreements, they can't read insurance forms, they can't read their mortgages, or their credit-card contracts. They can barely read more complicated menus. This is a sleeping giant of a problem. This is why people are so easy to scam: they can't read what's happening, but they're ashamed to admit it, so they sign things that they don't understand and get fleeced. When you're a very good reader and capable of understanding complex concepts relatively easily, it is hard to remember the relatively superficial degree to which most people understand what's going on around them. My reading comprehension in Italian and French is probably better than many Americans' in their native language. And I'm at about high B1 in those languages. <hr> <media href="https://www.youtube.com/watch?v=WteQGkACmLw" src="https://www.youtube.com/v/WteQGkACmLw" source="YouTube" width="560px" author="Nobel Prize" caption="Facts, fictions and critical thinking | The Future of Decision Making | Nobel Prize Dialogue Sydney"> At about <b>30:00</b>, <bq><b>Tim:</b> I also think it's about what we get cred for. And this is a lot further down the track, but people get cred at the moment for being sure, and [for] being declarative. And that's good and certainly in activism that can be very, very important. And it can create good change but it's mostly not at the moment. Mostly it's causing tribalization. And so, <b>the idea is that what a cool thing to be sure about is your <i>unsureness</i>. What a cool thing to be knowledgeable about is how hard it is to have absolute knowledge.</b> I don't know how to do that but it feels accessible to me. <b>Saul:</b> It does feel like we've often gone the other way, where we're teaching essay-writing, that can be 'show how you give your arguments to prove your point.' <b>Or a debate course where it can feel too much like the goal is to win, rather than to try to figure out how you're wrong and what you could learn.</b></bq> At about <b>35:00</b>, <bq>I don't get too much into this to sound like an old ranting guy on a porch, but [...] I'm a year clean, I'm off social media. And I don't let the news tell me when to read it. It's really hard. So my self-esteem was attached to the likes. But it's about agency. My kids seem to get it because we've instilled it. I think parents are now very anxious, so that this next generation of kids are growing up, understanding that our generation have discovered it to be wanting at best and dangerous at worst. And so I talk a lot to my kids about agency. <b>You choose when you're going to read the world news, don't have the news read you. You choose when you want to go look at a cat video video. Don't get fed a cat video in the middle of your work</b> And so I feel like it's as simple as that to get through this bit where we're being bombarded by digital information that we have no agency in consuming [...]</bq> At about <b>42:45</b> <bq>[...] when you're looking at your phone. you're not in your community. I noticed I wasn't being as good a dad. I mean, that was the thing that just made me go: mate, there's one thing you can do for the world is put good kids in it.</bq> Because this class of liberal doesn't actually believe in leftist or progressive principles more than they enjoy their creature comforts and elite status. <hr> <a href="https://www.taoistic.com/taoteching-laotzu/taoteching-17.htm" author="Lao Tzu" source="Tao Te Ching">Chapter 17</a> On the subject of management and hierarchy. <bq>The supreme rulers are hardly known by their subjects. The lesser are loved and praised. The even lesser are feared. The least are despised. Those who show no trust will not be trusted. <b>Those who are quiet value the words. When their task is completed, people will say: We did it ourselves.</b></bq> <h id="technology">Technology</h> Can we talk about what a shitshow the Netflix---and pretty much any streaming UI---is? It's all just carousels of content, with no real ability to filter, and everything you click maddeningly just starts playing stuff that you don't even know what it is yet. That's why you clicked it: you were trying to find out. There are places where there is literally no way to find out if Mark Wahlberg actually is in a show that is presented as a postage stamp, without just starting to watch the show. It just starts playing. You can't add to list. You can't find out more. It just plays. I guess I'll just binge an entire season of whatever the hell happened to be under my mouse because I'm a slack-jawed, dead-eyed mental infant who's got nothing better to do. What is the point? How does this drive engagement? It's just random clicks. All of the statistics that Netflix presents to its customers (read: advertisers and content-providers) are completely fake. The numbers are real, but what they purport to mean is fake. And don't even get me started on Apple TV, which doesn't even let you configure it <i>not</i> to immediately play the next episode. You also can't tell it to stop playing its own <i>fucking</i> trailers of other shows before you get to watch the show you actually selected. I will watch <i>Severance</i> when I'm good and ready. I don't need to be presented with the trailer 400 times. And, once you're done with a show, Apple TV will give you five seconds to leap for the remote to prevent it from <i>just starting a completely new show for you.</i> Do they think that their content is like oxygen? Do they think that I'll die if a show isn't playing for <i>more than five seconds</i>? Are they afraid that I'll not be able to figure out how to push a button to start a new show? How is this good, even for them? <hr> <a href="https://www.schneier.com/blog/archives/2025/02/ais-and-robots-should-sound-robotic.html" author="Bruce Schneier" source="">AIs and Robots Should Sound Robotic</a> <bq>And even those robot voices are being made obsolete by new AI-generated voices that can mimic every vocal nuance and tic of human speech, down to specific regional accents. And <b>with just a few seconds of audio, AI can now clone someone’s specific voice.</b> This technology will replace humans in many areas. Automated customer support will save money by cutting staffing at call centers. <b>AI agents will make calls on our behalf, conversing with others in natural language. All of that is happening, and will be commonplace soon.</b> But there is something fundamentally different about talking with a bot as opposed to a person. A person can be a friend. <b>An AI cannot be a friend, despite how people might treat it or react to it. AI is at best a tool, and at worst a means of manipulation.</b> Humans need to know whether we’re talking with a living, breathing person or a robot with an agenda set by the person who controls it. That’s why <b>robots should sound like robots.</b></bq> I don't to waste my time being friendly or diplomatic with a robot that can't be wheedled or guilted into following up on something when it says it will. <hr> <media href="https://www.youtube.com/watch?v=CwZW0CO7F-g" src="https://www.youtube.com/v/CwZW0CO7F-g" source="YouTube" width="560px" author="Half as Interesting" caption="The Obscure System That Syncs All The World’s Clocks"> It's not really an <i>obscure</i> system, in that it is incredibly well-documented, but this is still a reasonably informative and entertaining explainer. The system is called NTP---the Network Time Protocol---and comprises four tiers. Tier 0 is Atomic clocks, which measures the resonant frequency of Cesium atoms to obtain a regular "ticking" from nature itself. These are attached to servers in Tier 1, usually a machine that is on-site. These are, in turn, attached to Tier 2 servers, things like <c>time.windows.com</c> or <c>pool.ntp.org</c>. Any machines that we use are almost certainly in Tier 3, which are connected to Tier 2 machines. The machines coordinate between layers by relying primarily on their local clocks (usually kept running run by a CMOS battery on what passes for a motherboard) and re-synchronizing occasionally by "pinging" the layer above. They account for lag by including time sent and time received in messages, so that the sending system has four times with which to calculate the current time. <h id="llms">LLMs & AI</h> <a href="https://freddiedeboer.substack.com/p/the-shitting-in-the-yard-test" source="Substack" author="Freddie deBoer">The Indoor Plumbing Test</a> <bq><b>The AI conversation remains absurd, hype-ridden, and utterly out of touch with actual material reality. I could have written that sentence in 2024, 2023, or 2022, and it would have also been true.</b> But somehow it just gets more and more true.</bq> <bq>You’ll have to take my word for it that, in a list that was released with great fanfare, they rated the iPhone as the most important invention of all time. <b>Not antibiotics, the plow, or alternating current, not anesthesia or the lightbulb, but the iPhone</b>, which took a bunch of things that already existed (cellular telephone service, email on the go, a touchscreen) and put them in one remarkably profitable package.</bq> <bq>it wasn’t until the 1920s or so that indoor plumbing became a true mass phenomenon, again only in wealthy countries, and it was perfectly common for a soldier coming home from World War II in 1945 to be coming home to a house with a well and an outhouse. <b>It wasn’t until the 1960s that a majority of American homes had indoor plumbing, which means that the beginning of the Space Age overlapped with a period where most Americans couldn’t wash their hands whenever they wanted.</b></bq> <bq>[...] you also have to justify saying that AI is more important than, like, the bowl. By which I mean, bowls. To put food in. To eat out of. <b>Try and spend the rest of your life without ever using another food container and get back to me about whether ChatGPT is more important.</b> Food containers are inventions!</bq> <bq>[...] you’ll find immense frustration with them even among their most devoted users, and of course for every amazing image someone shares that came from an AI, there’s dozens that came out borked and were discarded. (<b>Hard to think of a more obvious example of selection bias than when someone generates ninety-nine shitty AI images and one good, then shares it online saying “Look at the power of AI!”</b>)</bq> <bq>I mean that’s exactly what that commercial is conveying, right? <b>They create a protagonist who is intended to appear as helpless and intellectually vacant as possible.</b> They then demonstrate the great value of the product they’re selling, Apple Intelligence, by having it take an email he spends 30 seconds writing and converting it into a more professional email that any human being who doesn’t have some sort of serious cognitive disability could also write in 30 seconds. And <b>Apple is not the only company that’s selling AI by demonstrating its ability to shepherd the tragically stupid through life.</b></bq> <bq>What artificial intelligence can actually do, in 2025 - two and a half years after people declared the world forever changed by the release of ChatGPT - is remarkably limited. ChatGPT can access and synthesize information, but not better than an educated adult, and <b>for important tasks almost anyone will choose to do that work themselves, especially given ongoing issues with LLM outputs, and anyway doing that work yourself is how you get and stay smart.</b></bq> <bq>The absolutely constant hype inflation never stops. Here ’s someone named Ross Lazer <b>claiming that the ability for an AI “agent” to order a pizza for you is as transformative as the automobile.</b> To be clear! What’s not being referred to here is the ability to order a pizza online, which is an affordance so old that pizza is believed to be the first thing ever sold via the internet. No, what’s as transformative as the automobile - which not only utterly changed human commerce and socialization forever but also resulted in the largest intentional transformation of our lived environment, ever - is <b>simply the ability to get a bot to do that simple, decades-old task for you.</b></bq> <bq>[...] <b>this belief in the miraculous potential of automating mundane human tasks only underlines how embarrassing ongoing AI struggles are.</b> As John Herrman of New York magazine writes, discussing ongoing difficulties faced by these agents, <b>“If buying groceries through a streamlined interface is deceptively complicated, what isn’t?”</b></bq> It treats even programmers like morons. <bq>I find it profoundly easy to order a pizza online. <b>Can it really be a socially optimal use of resources (immense amounts of money, manpower, and electricity) to create incredibly complex systems that can, with tons of training and eye-watering power costs, take that simple task off my hands?</b></bq> <bq>“They’re very sincere,” he says. To which I would say, <b>you mean everyone whose stock price and thus net worth is directly related to AI hype is in agreement about AI hype?</b> You don’t say! I’m sure many of them really are sincere, the same way that the guy who spends half his take-home income on sports gambling sincerely believes that his ship will eventually come in.</bq> <bq>Sam Altman says their is no slowdown in improvements to LLM-based AI systems; <b>his wealth is directly tied to public perception of whether there is a slowdown or not. These are not unrelated phenomena.</b></bq> <bq>The whole industry floats on the idea of limitless growth and massive market caps, the assumption that tech operates outside of ordinary financial constraints. And <b>with so many mature product categories and saturated fields, right now the only vehicle that can realistically power Silicon Valley is AI… whether it’s actually useful or not.</b></bq> <bq>why do people need this so bad? I get it, when it comes to corporations. <b>I don’t get it when it comes to people who are not directly financially remunerated by AI hype.</b></bq> <bq>[...] <b>that’s how human life works. You’ll still have to stand in impotent resentment while you wait for a subway that will arrive already stuffed with too many riders.</b> And if they invent the teleporter, then you’ll find other reasons to feel bored and annoyed.</bq> <bq><b>After 9/11</b>, people were afraid that we were going to live in an America where the local grocery stores were constantly getting bombed and we had to fight the terrorists in the streets, where nothing would ever be the same. <b>They were afraid, but there was such yearning in that fear. I think AI hype comes from the same place.</b></bq> <hr> <a href="https://youtubetranscriptoptimizer.com/blog/05_the_short_case_for_nvda" source="YouTube Transcript Optimizer" author="Jeffrey Emanuel">The Short Case for Nvidia Stock</a> <bq>Because I am such a dyed-in-the-wool believer in the long term transformative impact of this technology— I <b>truly believe it's going to radically change nearly every aspect of our economy and society in the next 5-10 years, with basically no historical precedent</b>— it has been hard for me to make the argument that Nvidia's momentum is going to slow down or stop anytime soon.</bq> Quit your bullshit. What about running water? I agree about the change though. There is a tiny minority in power. They see this as the next big play to make them more rich and powerful. Too much is never enough. It won't change things for the better though. Most people's lives will be worse in twenty years if the West has anything to say about it. It's just so tiring that these people think that this technology is more important than running water, sanitation, vaccines, etc. They're made so stupid by their greed. <bq>Deep learning and AI are the most transformative technologies since the internet, and <b>poised to change basically everything in our society.</b></bq> This is the non-hyped take? There's still no real product, Three years and a trillion dollars later. It's a get-even-richer scam for the rich. Fight me. <bq>Besides things like <b>the rise of humanoid robots</b>, which I suspect is going to take most people by surprise when they are rapidly able to perform a huge number of tasks that currently require an unskilled (or even skilled) human worker (e.g., <b>doing laundry, cleaning, organizing, and cooking; doing construction work like renovating a bathroom or building a house in a team of workers; running a warehouse and driving forklifts</b>, etc.), there are other factors which most people haven't even considered.</bq> OMG he thinks those things are a <i>lock</i>. OMG <i>he believes in the Jetsons</i>. Like, ...<i>soon</i>. None of that is going to happen. It's not even being worked on. The margins are too low. The up-front investment doesn't exist. These people are in charge of the economy. They're in charge of deciding where society invests its effort. They get to decide which value to produce. What a shitshow. <bq>[...] are <b>not properly licensed</b> for use as training data.</bq> AHAHAHAHA like you or anyone even cares about the outright unfairness and criminality of the whole operation. <bq>[...] a large chunk of that is already included in the training corpora used by the big labs, <b>whether it's strictly legal</b> or not.</bq> "Not strictly legal" is how you refer to something that is most definitely illegal but from which you benefit. <bq>The market is so excited about AI that it has thankfully ignored this, allowing companies like OpenAI to post breathtaking from-inception, cumulative operating losses while garnering increasingly eye-popping valuations in follow-up investment rounds (although, to their credit, they have also been able to demonstrate very fast growing revenues). But <b>eventually, for this situation to be sustainable over a full market cycle, these data center costs do need to eventually be recouped, hopefully with a profit, which over time is competitive with other investment opportunities on a risk-adjusted basis.</b></bq> Sure, sure, on paper and in classical economics where you're not just trying to avoid holding the bag when the jig is up. You only have to care if you're still invested when the bill comes due. He's writing as if anyone involved in this whole bubble cares about a sustainable business model. That's hopelessly naive. The rug-pull/pump-and-dump is just as reliable as a Ponzi scheme. <bq>[...] people really just started focusing on in the past year: <b>inference time compute scaling.</b></bq> Which means massively increased per-request compute and latency. Even more brute-forcing with different names. Isn't that one of the signs of a scam or a cult? That you just use different names for commonly known things? "inference-time compute-scaling" is "pouring more horsepower into each request to improve the quality of the result." <bq>[...] although <b>researchers have made breathtaking algorithmic improvements</b> on this front relative to the initial quadratic scaling people originally expected in scaling this up [...]</bq> This is not true. Sparse usage of attention to save memory is not panning out. <a href="https://en.wikipedia.org/wiki/No_such_thing_as_a_free_lunch" source="Wikipedia">TANSTAFL</a>. <bq>[...] the main difference is that <b>O1-Pro thinks for a lot longer before responding</b>, generating vastly more COT logic tokens, and <b>consuming a far larger amount of inference compute</b> for every response.</bq> <bq>[...] even a very long and complex prompt for Claude3.5 Sonnet or GPT4o, with ~400kb+ of context given, generally takes less than 10 seconds to begin responding, and often less than 5 seconds. Whereas <b>that same prompt to O1-Pro could easily take 5+ MINUTES before you get a response</b> [...]</bq> <bq>The Wright Brothers' airplane company in all its current incarnations across many different firms today isn't worth more than $10b despite them inventing and perfecting the technology well ahead of everyone else. And <b>while Ford has a respectable market cap of $40b today, it's just 1.1% of Nvidia's current market cap.</b></bq> This isn't because of an inherently just logic to the system. It's because pirates will always defeat inventors. Our system incentivizes piracy nearly exclusively. It is only curiosity and honor that adds anything at all into a system whose leaders scoop all rewards into their insatiable maws. <bq>[...] the fact that AMD's drivers suck, that popular AI software libraries don't run as well on AMD GPUs, that you can't find really good GPU experts who specialize in AMD GPUs outside of the gaming world (why would they bother when there is more demand in the market for CUDA experts?), that you can't wire thousands of them together as effectively because of lousy interconnect technology for AMD— <b>all this means that AMD is basically not competitive in the high-end data center world, and doesn't seem to have very good prospects for getting there in the near term.</b></bq> <bq>[...] you can imagine that <b>absolutely biblical amounts of capital, brainpower, and effort are being expended in this area.</b></bq> It will be a tragedy that will perhaps remain unremarked by a history that will never get the chance to be written that this level of investment only happens for something as hair-brained as this kind of thing, but not for feeding the world or combating climate change. The only way anyone loosens this much cash in the west is if they think that they can earn more money for themselves. <bq>[...] instead of trying to battle Nvidia head-on by using a similar approach and trying to match the Mellanox interconnect technology, Cerebras has used a radically innovative approach to do an end-run around the interconnect problem: <b>inter-processor bandwidth becomes much less of an issue when everything is running on the same super-sized chip.</b> You don't even need to have the same level of interconnect because one mega chip replaces tons of H100s.</bq> <bq><b>Cerebras chips also work extremely well for AI inference tasks.</b> In fact, you can try it today for free here and use Meta's very respectable Llama-3.3-70B model. It responds basically instantaneously, <b>at ~1,500 tokens per second.</b> To put that into perspective, anything above 30 tokens per second feels relatively snappy to users based on comparisons to ChatGPT and Claude, and <b>even 10 tokens per second is fast enough that you can basically read the response while it's being generated.</b></bq> <bq>Using a comparable Llama3 model with "speculative decoding," <b>Groq is able to generate 1,320 tokens per second, on par with Cerebras and far in excess of what is possible using regular GPUs.</b> Now, you might ask what the point is of achieving 1,000+ tokens per second when users seem pretty satisfied with ChatGPT, which is operating at less than 10% of that speed. And the thing is, it does matter. <b>It makes it a lot faster to iterate and not lose focus as a human knowledge worker when you get instant feedback.</b> And if you're using the model programmatically via the API, which is increasingly where much of the demand is coming from, then it can enable whole new classes of applications that require multi-stage inference (where the output of previous stages is used as input in successive stages of prompting/inference) or which require <b>low-latency responses, such as content moderation, fraud detection, dynamic pricing, etc.</b></bq> <bq>At the very least, <b>Cerebras and Groq can chip away at the lofty expectations for Nvidia's revenue growth</b> over the next 2-3 years that are embedded in the current equity valuation.</bq> A bunch of Nvidia's value is more "greater fool" than "expectation of future gains". <bq>How should one think about the future of this business when literally <b>every single one of these VIP customers is building their own custom chips</b> specifically for AI training and inference?</bq> <bq>When thinking about all this, you should keep one incredibly important thing in mind: <b>Nvidia is largely an IP based company. They don't make their own chips.</b> The true special sauce for making these incredible devices arguably comes more from <b>TSMC, the actual fab, and ASML, which makes the special EUV lithography machines used by TSMC</b> to make these leading-edge process node chips.</bq> <bq><b>TSMC will sell their most advanced chips to anyone who comes to them with enough up-front investment and is willing to guarantee a certain amount of volume.</b> They don't care if it's for Bitcoin mining ASICs, GPUs, TPUs, mobile phone SoCs, etc.</bq> <bq>These frameworks allow developers to write their code once using high powered abstractions and then target tons of platforms automatically— doesn't that sound like a better way to do things, which would give you a lot more flexibility in terms of how you actually run the code?</bq> Sure, unless you need the low-level optimizations you get from writing to a specific ABI. Cross-compilers aren't magic. You'll leave some performance on the table with any LCD approach, but it's unclear whether it matters. It might be fine, especially if most of the other compile targets are basically just emulating CUDA's ABI surface anyway. <bq>[...] instead of having that code compiled for use on Nvidia GPUs like you would normally do, <b>it can instead be fed as source code into an LLM which can port it into whatever low-level code is understood by the new Cerebras chip, or the new Amazon Trainium2, or the new Google TPUv6, etc.</b> This isn't as far off as you might think; it's probably already well within reach using OpenAI's latest O3 model, and <b>surely will be possible generally within a year or two.</b></bq> Why the fuck would you transpile with an LLM? Are you so afraid of writing a relatively straightforward transpiler? CUDA isn't even that complex. I can't think of a more error-prone and inefficient approach and this guy seems to be typing one-handed about it. <bq>DeepSeek cracked this problem by developing a clever system that breaks numbers into small tiles for activations and blocks for weights, and strategically uses high-precision calculations at key points in the network. Unlike other labs that train in high precision and then compress later (losing some quality in the process), <b>DeepSeek's native FP8 approach means they get the massive memory savings without compromising performance.</b> When you're training across thousands of GPUs, this dramatic reduction in memory requirements per GPU translates into needing far fewer GPUs overall.</bq> <bq>DeepSeek figured out how to predict multiple tokens while maintaining the quality you'd get from single-token prediction. <b>Their approach achieves about 85-90% accuracy on these additional token predictions, which effectively doubles inference speed without sacrificing much quality.</b> The clever part is they maintain the complete causal chain of predictions, so the model isn't just guessing— it's making structured, contextual predictions. One of their most innovative developments is what they call <b>Multi-head Latent Attention (MLA). This is a breakthrough in how they handle what are called the Key-Value indices</b>, which are basically how individual tokens are represented in the attention mechanism within the Transformer architecture. Although this is getting a bit too advanced in technical terms, suffice it to say that <b>these KV indices are some of the major uses of VRAM during the training and inference process</b>, and part of the reason why you need to use thousands of GPUs at the same time to train these models— each GPU has a maximum of 96 gb of VRAM, and these indices eat that memory up for breakfast.</bq> <bq>This means that the entire mechanism is "differentiable" and able to be trained directly using the standard optimizers. All this stuff works because these models are ultimately finding much lower-dimensional representations of the underlying data than the so-called "ambient dimensions". So <b>it's wasteful to store the full KV indices, even though that is basically what everyone else does.</b></bq> <bq>They also made major advances in GPU communication efficiency through their DualPipe algorithm and custom communication kernels. This system intelligently overlaps computation and communication, carefully balancing GPU resources between these tasks. <b>They only need about 20 of their GPUs' streaming multiprocessors (SMs) for communication, leaving the rest free for computation. The result is much higher GPU utilization than typical training setups achieve.</b></bq> <bq>The "cost" of that extreme level of knowledge is that the models become very unwieldy both to train and to do inference on, because <b>you always need to store every single one of those 405B parameters (or whatever the parameter count is) in the GPU's VRAM at the same time in order to do any inference with the model.</b></bq> <bq>The beauty of the MOE model approach is that you can decompose the big model into a collection of smaller models that each know different, non-overlapping (at least fully) pieces of knowledge. DeepSeek's innovation here was developing what they call <b>an "auxiliary-loss-free" load balancing strategy that maintains efficient expert utilization without the usual performance degradation that comes from load balancing.</b></bq> <bq>[...] the technical papers mention several other key optimizations. These include their <b>extremely memory-efficient training framework that avoids tensor parallelism, recomputes certain operations during backpropagation instead of storing them, and shares parameters between the main model and auxiliary prediction modules.</b></bq> <bq>[...] <b>the cost differential of the DeepSeek API relative to the OpenAI and Anthropic API could be simply that they are nearly 50x more compute efficient</b> (it might even be significantly more than that on the inference side— the ~45x efficiency was on the training side).</bq> <bq>I think it's more likely that they are telling the truth, and that they have simply been able to achieve these incredible results by being extremely clever and creative in their approach to training and inference. <b>They explain how they are doing things, and I suspect that it's only a matter of time before their results are widely replicated and confirmed by other researchers at various other labs.</b></bq> <bq>[...] unlike OpenAI, which is incredibly secretive about how these models really work at a low level, and won't release the actual model weights to anyone besides partners like Microsoft and other who sign heavy-duty NDAs, these DeepSeek models are both completely open-source and permissively licensed. <b>They have released extremely detailed technical reports explaining how they work, as well as the code that anyone can look at and try to copy.</b></bq> <bq>[...] using pure reinforcement learning with carefully crafted reward functions, they managed to get models to develop sophisticated reasoning capabilities completely autonomously. <b>This wasn't just about solving problems— the model organically learned to generate long chains of thought, self-verify its work, and allocate more computation time to harder problems.</b></bq> Here's where you lose me. This is a wildly romanticized description trying desperately to round up to AGI. He can't just blow up one hype bubble; he has to puff up another one simultaneously. We love our myths. <bq>[...] <b>they developed a clever rule-based system that combines accuracy rewards (verifying final answers) with format rewards (encouraging structured thinking).</b> This simpler approach turned out to be more robust and scalable than the process-based reward models that others have tried.</bq> It's a program. You've described a classic program. It might be cleverly done but I feel like the author and his ilk are slowly training themselves to believe that anything that is like non-LLM programming is witchcraft whereas LLM-based queries are "normal." <bq>Apparently, the Llama project within Meta has attracted a lot of attention internally from high-ranking technical executives, and as a result they have something like <b>13 individuals working on the Llama stuff who each individually earn more per year in total compensation than the combined training cost for the DeepSeek-V3 models which outperform it.</b></bq> <bq>[...] if AI really is as transformational as I expect, if the real-world utility of this tech is measured in the trillions, if inference-time compute is the new scaling law of the land, <b>if we are going to have armies of humanoid robots running around doing massive amounts of inference constantly</b> [...]</bq> Humanoids robots? How did you round up to that? People in this tech-investor space are so weird and gullible. Did he just want to be the one to say it? You know, so he can point to it later and say "see? I predicted it." This constant grandstanding is exhausting. <bq>Most importantly, we're seeing the emergence of <b>LLM-powered code translation that could automatically port CUDA code to run on any hardware target</b>, potentially eliminating one of NVIDIA's strongest lock-in effects.</bq> These people can't see anything getting done without LLMs these days. See above. There is no reason to use an LLM to transpile code when you could just write a transpiler once. Stop fucking around and wasting time. That is the stupidest possible thing you could do with an LLM---just have it generate one ad-hoc transpiler after another, each one a bit different, each one with the potential to be just slightly wrong or less efficient. God forbid, you should collect wisdom in a tool. These people can't think beyond just having the LORD GOD LLM do it every time, for free! <bq><b>The economics here are compelling</b>: when DeepSeek can match GPT-4 level performance while charging 95% less for API calls, it suggests either NVIDIA's customers are burning cash unnecessarily or margins must come down dramatically.</bq> That's Just it: no-one ever talks about the societal benefit. The need for speed is purely driven by first-movers wanting money, but they have no idea what they could achieve with the thing that will make them money. There is no incentive to care. The assumption is that, if it makes money, there's a useful side-effect. But they care so little about that, that this hour-long article doesn't mention anything except humanoid robots and washing dishes. <hr> <a href="https://x.com/karpathy/status/1886192184808149383" author="Andrej Karpathy" source="Twitter">"Programming"</a> <bq>There's a new kind of coding I call "vibe coding", where you fully give in to the vibes, embrace exponentials, and forget that the code even exists. <b>It's possible because the LLMs (e.g. Cursor Composer w Sonnet) are getting too good. Also I just talk to Composer with SuperWhisper so I barely even touch the keyboard.</b> I ask for the dumbest things like "decrease the padding on the sidebar by half" because I'm too lazy to find it. I "Accept All" always, I don't read the diffs anymore. When I get error messages I just copy paste them in with no comment, usually that fixes it. The code grows beyond my usual comprehension, I'd have to really read through it for a while. Sometimes the LLMs can't fix a bug so I just work around it or ask for random changes until it goes away. <b>It's not too bad for throwaway weekend projects, but still quite amusing. I'm building a project or webapp, but it's not really coding - I just see stuff, say stuff, run stuff, and copy paste stuff, and it mostly works.</b></bq> I just put this here because this is what people will call the future of programming---some claiming that it's already here---but Karpathy is very explicit that he's doing this for "throwaway" projects. These are, at best, prototypes. This style only works because Karpathy <i>already kind of knows how to program</i> and <i>he's smart</i>. This style will not make someone magically able to produce anything other than a crude facsimile of what has already been produced before. They will likely get stuck in myriad cul-de-sacs where the LLM avows very confidently that something will work and it will not work, sometimes subtly, sometimes overtly. Yesterday, I was confronted three times with outright inaccurate, or simply outdated or inapplicable suggestions from Copilot. In one case, someone asked in a PR whether we really could eliminate the framework declaration in <c>app.config</c> files. Copilot confidently said that you absolutely needed the declaration for reasons that might have been somewhat applicable in 2009, or maybe even 2012. The Microsoft documentation still confidently declares that you need to include this declaration but says nothing about why it's not included in any of its starter templates. The app in question works with .NET Framework 4.8. That's been out for three years. There is no follow-up version in sight. That version of the framework has been included on all versions of Windows for years. The app doesn't run with a lower version. We don't need to artificially pin the version with an outdated mechanism. The version of the framework give in the project file is not only sufficient, but it's distracting to have an outdated alternative that says something slightly different but will be ignored anyway hanging out in the project files. It's a recipe for configuration cargo cults, which is why I threw it out. Copilot would chirpily encourage people to put it right back in, and to waste time fine-tuning it. In another case, a colleague was having trouble upgrading Entity Framework from using the <c>System.Data.SqlClient</c> to using the <c>Microsoft.Data.SqlClient</c> instead. There was an error where a registration wasn't being honored and the app couldn't find the new provider. The Copilot solution was to confidently recommend adding a whole bunch of crap that wouldn't solve the problem in any sane manner---and couldn't. The EF support took a dependency on a 5.x component but the latest version was 6.x. The suspicion was that perhaps something would work better in 6.x. When my colleague added the newer version to all projects, things started working again, so apparently a bug had been fixed. But which bug? When I was called in, I read the error message and it was pretty clear that something in the system was overriding the settings we wanted to use. It turned out that there was a reference to the old data provider in one of the core libraries that we'd upgraded. It's unclear why that would the configuration to ignore subsequent registrations of other data providers. However, removing that single line of configuration in the base library solved the error. My theory is, that the 5.x version wasn't capable of properly managing multiple, registered data providers but that the 6.x version could. When the library was forced to the newer version, everything worked again. The maintainable solution that we ended up using was: <ol>Remove the unwanted configuration entry in the base library. Add an explicit reference to the newer version of the transient library to ensure that we were using the 6.x version---with the bug fix that we technically no longer needed---in all consumers of the library. This step wasn't even strictly necessary but there's no harm in it. We marked the package-inclusion with a reason and a link to the work item that was tracking the work and included a write-up of what had happened.</ol> Finally, I had updated a build-pipeline template to accept a new parameter: <c>OutputFolder</c>, which was to default to a value based on the value of another parameter <c>SolutionFolder</c>. The following solution feels like a natural way of expressing this intent. <code>parameters: SolutionFolder: '' OutputFolder: '${{ parameters.SolutionFolder }}\Output'</code> This is what it looked like in the template: <img src="{att_link}this_is_not_allowed.png" href="{att_link}this_is_not_allowed.png" align="none" caption="This is not allowed" scale="100%"> However, if you try it, you will get an error message saying that "A template expression is not allowed in this context": <img src="{att_link}template_expression_not_allowed.png" href="{att_link}template_expression_not_allowed.png" align="none" caption="Template Expression not allowed" scale="100%"> However, if you ask one of the machines (this is DuckDuckGo's, which is based on ChatGPT by default), it will cheerily tell you that you can totally do exactly what you clearly can't do. <img src="{att_link}ai_thinks_prameters_can_reference_each_other.png" align="none" caption="AI thinks parameters can reference each other"> This is the <a href="https://thecloudtheory.github.io/posts/handling-default-parameter-value-in-azure-devops/">cloud theory</a> link shown in the screenshot. There's nothing in here that has anything to do with the question or its answer. It's about Azure pipelines but it's a pretty banal article that explains the basics and it will utterly waste your time if you were looking for a definitive answer on how to reference parameters from one another, as it claims is possible. This is my worry about this whole LLM-based tool thing: it's wasting people's time and making them inefficient. The best-case scenario is if it manages to give them a mediocre solution that works. The worst-case scenario is that they get routed down a seemingly productive rabbit-hole that doesn't lead them to a solution but which they can then claim that had been productive because they'd been doing "work" and who can blame them if it was the LLM that started them on that path? How do you do this for real? You use the solution that I came up with after quickly learning from <a href="https://github.com/microsoft/azure-pipelines-agent/issues/2737">Allow parameter defaults to be calculated based on other parameters #2737</a> on GitHub that it was never implemented. I used a variable and <a href="https://learn.microsoft.com/en-us/azure/devops/pipelines/process/expressions?view=azure-devops#coalesce">coalesce</a>, as shown below. <code>variables: defaultOutputPath: '${{parameters.SolutionFolder}}/Output' templateOutputFolder: ${{ coalesce(parameters.OutputFolder, defaultOutputPath) }}</code> Curious, though, I went to Copilot to what it would suggest (assuming that Copilot is a bit more advanced than the default ChatGPT used by DuckDuckGo). As usual, it returned a lot of boilerplate text that may or may not have actually corresponded to the code. I didn't bother reading most of it and just kept the code examples. <box title="🧐 1️⃣ prompt">azure yaml use parameter in default for another parameter</box> <box title="🤖 1️⃣ answer"><code>parameters: - name: repositoryName type: string default: '' - name: sonarQubeProjectKey type: string default: '' jobs: - job: SetDefaults steps: - script: | echo "##vso[task.setvariable variable=sonarQubeProjectKey]com.company.${{ parameters.repositoryName }}" displayName: 'Set SonarQube Project Key'</code></box> This is an impenetrable and unmaintainable disaster that writes a variable to the job's terminal, which also sets it in the pipeline environment as a side-effect.Why would you do it that way when you just need a local variable? I don't even know if this would work because I wasn't even willing to copy/paste it. I already know what the solution is, so I know that I can keep prompting instead of wasting my time trying something that looks OK but which I'm not in a position to judge. <div align="right" style="font-size: 500%">🎰</div>That's the advantage of being a senior developer: I will know a good answer when I see it, so the LLM can save me time as I pull the slot-machine arm until it comes up sevens. <box title="🤨 2️⃣ prompt">Not even close. Try harder.</box> <box title="🤖 2️⃣ answer"><code>parameters: - name: repositoryName type: string default: 'default-repo' - name: sonarQubeProjectKey type: string default: '' variables: - name: sonarQubeProjectKey value: ${{ parameters.sonarQubeProjectKey != '' && parameters.sonarQubeProjectKey || 'com.company.' + parameters.repositoryName }} jobs: - job: Build steps: - script: | echo "SonarQube Project Key: $(sonarQubeProjectKey)" displayName: 'Display SonarQube Project Key'</code></box> That's a bit better. Now there's a variable declaration, even though it still uses the unfortunate old-style name/value syntax that is not super-intuitive. It also uses a bizarre bastardization of the <c>||</c> operator to <i>coalesce</i> to a default value if it's not set. I'm not an expert on pipeline-expression syntax but the placement of the <c>&&</c> looks pretty wonky. As with the original answer, I didn't bother even trying to see if that one-liner mess even compiles, to say nothing of getting me what I want. Even if it works, it's an impenetrable and unmaintainable line that no-one will ever dare touch again. Since I already knew the answer, I told it straight-up how I wanted it to write the expression. <box title="😒 3️⃣ prompt">Use coalesce</box> <box title="🤖 2️⃣ answer"><code>parameters: - name: repositoryName type: string default: 'default-repo' - name: sonarQubeProjectKey type: string default: '' variables: sonarQubeProjectKey: $[coalesce(parameters.sonarQubeProjectKey, 'com.company.' + parameters.repositoryName)] jobs: - job: Build steps: - script: | echo "SonarQube Project Key: $(sonarQubeProjectKey)" displayName: 'Display SonarQube Project Key'</code></box> It's unclear why it's using <a href="https://learn.microsoft.com/en-us/azure/devops/pipelines/process/variables?view=azure-devops&tabs=yaml%2Cbatch#runtime-expression-syntax" author="" source="">Runtime expression syntax</a> instead of the more-standard <a href="https://learn.microsoft.com/en-us/azure/devops/pipelines/process/variables?view=azure-devops&tabs=yaml%2Cbatch#macro-syntax-variables" author="" source="">Macro syntax variables</a>. It doesn't matter, though because, according to the documentation (and as far I knew already), you have to use the <a href="https://learn.microsoft.com/en-us/azure/devops/pipelines/process/variables?view=azure-devops&tabs=yaml%2Cbatch#template-expression-syntax" author="" source="">Template expression syntax</a> when referencing template parameters (which we most certainly are---it's right there in the question). Is it complicated that there's a table as well as a section called <a href="https://learn.microsoft.com/en-us/azure/devops/pipelines/process/variables?view=azure-devops&tabs=yaml%2Cbatch#what-syntax-should-i-use" author="" source="">What syntax should I use?</a> that explains which form to use where and what to watch out for? Yes, it is. But I don't make the rules. I just play by them. Unlike Copilot, which would have cheerily helped a less-savvy developer piss away an entire afternoon if not a couple of days chasing phantom errors, trying desperately to get the code produced by the <i>genius machine</i> to work, as advertised. The thing about pipeline scripts is...you have to executed them in the cloud. The turnaround time is murder. Let's take a quick peek at my version, the one I had before we starting playing around the <i>the tool of the future that you're a fool and knave for not using because junior devs super-powered by it will blow by your ancient ass if you don't learn how to use it</i>. <code>variables: defaultOutputPath: '${{parameters.SolutionFolder}}/Output' templateOutputFolder: ${{ coalesce(parameters.OutputFolder, defaultOutputPath) }}</code> Succinct, easy to read, with useful variable names. Easy, peasy, lemon-squeezy. Also, it works! That sets it apart from the Copilot suggestions above. The only trouble I had in the PR was that my reviewer didn't know what the coalesce function did...and the <a href="https://dict.leo.org/german-english/coalesce?side=both" author="" source="">top German translation</a> was for <i>Verbinden</i>, which means to link, tie, or join. That meant he thought it was a synonym for <c>concatenate</c>, which it most certainly is not. It was a good learning opportunity, where I reminded him of the <c>coalesce</c> function in SQL, as well as the <a href="https://learn.microsoft.com/en-us/dotnet/csharp/language-reference/operators/null-coalescing-operator">null-coalescing operators</a>---<c>??</c> and <c>??=</c>---in C#. <h id="programming">Programming</h> <a href="https://tratt.net/laurie/blog/2024/can_we_retain_the_benefits_of_transitive_dependencies_without_undermining_security.html" source="Laurence Tratt" author="">Can We Retain the Benefits of Transitive Dependencies Without Undermining Security?</a> <bq>In this post I’m going to argue that <b>the growth in transitive dependencies in software is the equivalent of jamming our door open and hoping for the best — we are putting too much trust in things we don’t and can’t know in detail.</b> However, I don’t think that the best long-term solution is to avoid transitive dependencies all together — we’re increasing our use of direct and indirect dependencies because it makes us more productive and our software better. <b>Is it possible to get the advantages without the disadvantages?</b></bq> <bq>Simplifying only slightly, every machine code instruction executed has the ability to read from, and write to, anywhere within a processes’ memory. If the software building my website does something clever with passwords , <b>any one of those 181 dependencies could decide that it will scan my processes’ memory for passwords, and send any it finds over the internet to a bad person.</b></bq> <bq>In essence, capabilities in <b>CHERI</b> are double width pointers with fine-grained permissions: code can only access capabilities to which it is given permission. One can use this, for example, to lock code into a subset of a process, only able to escape via a single well-defined exit point. Exploring the possibilities is great fun, but the more sophisticated one’s compartment mechanism becomes, the more likely it is to be incomplete. <b>I have come to think that this style of compartmentalisation most useful for ensuring that cooperative (i.e. trusted) software doesn’t go wrong accidentally. Ensuring that actively malicious code doesn’t undermine the hoped-for security guarantees is much harder.</b></bq> <bq>All this has led me, slowly and reluctantly, to the conclusion that <b>our dependency-heavy approach to building software is fundamentally incompatible with security.</b> I say this with great reluctance — I find it much easier to write large, reliable software than I did 10 years ago, and the quality and quantity of dependencies that is now available is a big part of that. However, I am painfully aware that <b>this approach means that I’m taking on more risk than I should be comfortable with.</b></bq> <bq><b>History is replete with examples of people who thought that they could avoid bad behaviour by others by asserting that it couldn’t happen</b> — and to whom the bad behaviour later happened. Money spent on defence might seem wasted, but war is much more expensive.</bq> <bq><b>The more dependencies we use within a single process, the less suitable the process is as a security mechanism.</b> However, stating that there is a problem that needs solving doesn’t mean that there is an obvious solution, or that any such solution is practical, or that there is even a “solution” at all. <b>Sometimes we have to accept that all the trade-offs available to us are unpalatable.</b></bq> <bq><b>I don’t want my image decoding component to have network access, or the ability to access RAM with passwords in</b>; but I do want my network downloading component to have network access, and I do want to be able to create a component that can manage and use passwords.</bq> <bq>Another way of looking at this is that it is a more rigorous enforcement of the age-old principle of least privilege: <b>our current approach to software hands out far too many privileges to dependencies, and we need to rethink how we build software</b> for this to become a thing of the past.</bq> <bq><ol>It is reminiscent of privilege separation as found in OpenSSH which <b>splits single “static programs” into multiple dynamic processes.</b> Compromising one process does not compromise another. <b>The actor model, which defines how interacting “things” can communicate with each other.</b> This is a fairly large umbrella term, ranging from languages such as Erlang to various libraries and frameworks; few have security as an explicit aim.</ol></bq> <bq><b>The software industry is more productive than ever, but arguably less imaginative.</b> We’re mostly using operating systems that are almost exclusively recognisably 1960s/1970s in style, and programming languages and CPUs that are mostly recognisably 1970s/1980s in style. <b>I hope that this reflects a current period of consolidation, and that it is not indicative of permanent stasis.</b></bq> <bq>We sometimes underappreciate how many <b>simple security mitigations</b> included in modern operating systems have <b>made it harder for programming flaws to result in security flaws.</b></bq> <hr> <a href="https://blog.ploeh.dk/2025/01/27/dependency-inversion-without-inversion-of-control/" source="Ploeh Blog" author="Mark Seemann">Dependency inversion without inversion of control</a> <bq>Despite the name similarity, <b>the Dependency Inversion Principle isn't equivalent with Inversion of Control or Dependency Injection.</b> There's a sizeable intersection between the two, but the DIP doesn't require IoC. I often use the Functional Core, Imperative Shell architecture, or the Impureim Sandwich pattern to invert the dependencies without inverting control. <b>This keeps most of my code more functional, which also means that it fits better in my head and is intrinsically testable.</b></bq> <hr> <a href="https://blog.ploeh.dk/2025/02/03/modelling-data-relationships-with-c-types/" author="Mark Seemann" source="Ploeh Blog">Modelling data relationships with C# types</a> <bq><b>This article demonstrates how to use the <i>Ghosts of Departed Proofs</i> technique in C#.</b> In some ways, I find that it comes across as more idiomatic in C# than in F#. I think this is because rank-2 polymorphism is only possible in F# when using its object-oriented features. Since F# is a functional-first programming language, it seems a little out of place there, whereas it looks more at home in C#. [...] I think I actually like how the C# API turned out, although having to define and implement a class every time you need to supply a Visitor may feel a bit cumbersome. Even so, <a href="https://blog.ploeh.dk/2024/05/13/gratification">developer experience shouldn't be exclusively about saving a few keystrokes</a>. After all, <a href="https://blog.ploeh.dk/2018/09/17/typing-is-not-a-programming-bottleneck">typing isn't a bottleneck</a>.</bq> <hr> <media href="https://www.youtube.com/watch?v=T9UqIkuGnuo" src="https://www.youtube.com/v/T9UqIkuGnuo" source="YouTube" author="Nick Chapsas" caption="The Exciting Future of C# with Mads Torgersen" width="560px"> At about <b>45:00</b>, Mads describes a higher-level typing system that allows <a href="https://github.com/dotnet/csharplang/discussions/8711" source="GitHub">Proposal: Existential types for interfaces and abstract types #8711</a> <bq>There are various features that are related but not quite the same. There are 'associated types' in some languages and there's what one scholar calls 'abstract types' [I think he meant 'existential types'], which might be my favorite version of the feature, which are kind of an alternative to generics or a kind of generics. The really short version is they help you not have so damn many type arguments all the time, yeah? Okay; essentially think of it as a class. Instead of a class having a type parameter saying, I'm an animal with a type argument saying which kind of food it eats...that means <b>every time you talk about animals, you have to pass type arguments around. That's really annoying because what kind of food it eats is inherent to it. It shouldn't be like something on the outside; it should be a member, saying my food type is [whatever]</b> ... ...and if you do that, then you can kind of tamp down on a lot of the...sometimes you just end up in generics overload, or passing the same stuff around. And every one of these related types has a type argument for which particular implementation of the other related interfaces it is using and <b>they all carry the same five type arguments around all the time.</b> So that would be a feature, if we can get it right, and if we can work it into the runtime, and it's limited---and we are occasionally talking about it---that <b>could be a really really beautiful and quite impactful addition to C#.</b></bq> The <a href="https://github.com/dotnet/csharplang/discussions/8711">proposal</a> does a better job of explaining it, though. The interface would look like this: <code>interface ICounter<protected> { T Start { get; } void Next(T current); bool Done { get; } }</code> And then you could just use it without the generic parameter, like this: <code>void M(ICounter ic) { var x = ic.Start; while (!ic.Done) { x = ic.Next(x); } }</code> The implementation is the only one that has to pass the type parameter: <code>class Counter : ICounter<int> { int Start => 0; int Next(int current) => current + 1; bool Done => current == 42; }</code> At <b>58:00</b>, they discuss what something like "discriminated unions" or "tagged unions" would look like in a decidedly object-oriented language like C#. <bq><b>Mads:</b> We arrived at this degree of clarity around what our options are. The type unions---use of the word 'types' there reflects that one conclusion that we reached is that, in C, unions should be 'unions of types'. If you look at F# or other functional languages, discriminated unions are not unions of types. They're unions of something with a name, a tag-discriminator, whatever you want to call it, the tagged unions that can then be deconstructed to give you values of one or more types. So <b>the different options are like 'named options' but they're not things in and of their own right. They're just a means to get to what's inside.</b> In C#, one thing that we agree on is that that has to change. It has to be that the things are types. So you don't have to pattern on a union and get an <c>A</c> or you get a dog and immediately you have to decompose it into how many legs it has and and what it eats. You know what its name is. You can carry it around as a dog. It makes sense in its own right and it can be its own object. So <b>that's essentially trying to take an object-oriented view on what discriminated unions would look like in C#.</b> <b>Nick:</b> Fundamentally, <b>discriminated unions are a very functional-programming-like concept</b> and they don't fit in a language that already has inheritance right? Because the idea is <b>it's their version of inheritance.</b> <b>Mads:</b> Exactly.</bq> At <b>01:22:00</b>, Nick asks what are Mads's three least-favorite/most-hated features of C#, things that he wishes had never gone in or that, in a better world, he would remove. <ol><b>Events</b>, because they should never have been a language feature. They should have, at most, been a runtime/library feature. <b>Delegates</b>, because they were forced to be collection types because of events. They execution dynamics are indeterminate (ordering) and they actually break covariance (kind of like arrays). <b>Void</b> should have been a type. Its being a language feature means that they bifurcate all other support, like requiring a distinction between <c>Func</c> and <c>Action</c>. <b>dynamic</b> is a pretty great feature <i>academically</i> but the value of the feature became less than hoped. <iq>Performance-wise, it's a disaster. There's a whole bunch of infrastructure to maintain it. It doesn't carry its own weight.</iq> </ol> The final part of the discussion is about nullability. <iq>The point is that it's so much better than nothing.</iq> The feature makes you think about what even should be allowed to be <c>null</c>. You should avoid using <c>null</c> unless you absolutely can't avoid it. Don't ruin your API, of course, but be absolutely sure that <c>null</c> is an option. It is much easier to write code with APIs that never return <c>null</c>. Consider sentinel objects. The feature in C# has gotten a lot better and is very good now. <c>init</c> and <c>required</c> properties helped finalize the feature. <hr> <a href="https://old.reddit.com/r/apple/comments/1ihufa0/doom_running_on_apple_lightning_to_hdmi_dongle/" source="Reddit">DOOM running on Apple Lightning to HDMI dongle</a> The first comment explains why this works. <bq><b>The dongle's firmware is super stripped-down iOS, basically</b> There is SecureROM, iBoot and XNU as a kernel - just like some iPhone or iPad of that era (now is the same, but obviously they did a lot of development since then) Production firmware's userspace is ultra-minimalistic though - there's a ramdisk, but it's not even a filesystem, but a statically compiled Mach-O (it's like ELF, but for Apple *OS) Internal development bundles do have a proper ramdisk with filesystem and a bunch of executables/shared libraries on it The Mac here just loads such firmware into it, since the dongle doesn't have any persistent storage. The colorful logs going in one of the terminals are UART output from it - first iBoot and then kernel and userspace <b>Arbitrary code execution is achieved due to iOS-world bootrom exploit - checkm8, which also works here because codebase is literally the same.</b></bq> The first response to this was <iq>Damn I feel like I failed myself, I have a CS degree and I feel I'm reading gibberish, well done.</iq> That comment made me realize that I read the explanation without even blinking. It's 100% clear to me what's going on. I couldn't have done it myself, but I understand how it was done. It's good to be reminded that people see words and explanations within their own context. <hr> <a href="https://blog.plover.com/2025/02/06/#just-answer-the-question-2" author="Mark Dominus" source="The Universe of Discourse">Just give the man the fish!</a> <bq>I'm kind of an asshole, but I'm not that big an asshole. I'm callous, but I'm not sadistic. <b>Someone who says they don't have time to help you, but who does have time to explain to you in detail why they aren't helping you, is cruel.</b></bq> Why just answer the question---give them a fish---when they could have read the manual themselves---learned how to fish? <bq>Because it's easy. Because it's helpful. Because <b>I think the theory that says that people will become dependent on it is bullshit.</b> Because I think the theory that says that telling them to read the man page is more helpful is also bullshit. Because <b>in my experience people are much more likely to heed your suggestion to read the man page after you have established that you are a helpful concerned person</b> by assisting them.</bq> I agree with all of what Mark is writing here with the caveat that there are now a lot of people who don't put any thought into trying to figure out something for themselves, who simply paste their homework assignments into forums like this, and who will be back with equally insipid questions again and again and again. This has, to be fair, reduced significantly in the last couple of years, as LLMs are now fielding and incorrectly answering these kinds of questions, which none of the queriers will be able to notice are wrong. But the LLM has infinite patience for handing out fish, and isn't capable of caring about whether the prompter ever learns how to fish. <h id="fun">Fun</h> <media href="https://www.youtube.com/watch?v=x9nrBF5gl8A" src="https://www.youtube.com/v/x9nrBF5gl8A" source="YouTube" width="560px" author="DJ Cummerbund" caption="Everybody and The Sunshine Band"> <hr> <media href="https://www.youtube.com/watch?v=XmuGBNBBu3A" src="https://www.youtube.com/v/XmuGBNBBu3A" source="YouTube" width="560px" author="SNL" caption="Museum of Hip-Hop Panel">