<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 and probably end up in the gigantic backlog of unpublished drafts.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="#economy">Economy & Finance</a> <a href="#politics">Public Policy & Politics</a> <a href="#journalism">Journalism & Media</a> <a href="#science">Science & Nature</a> <a href="#philosophy">Philosophy & Sociology</a> <a href="#technology">Technology</a> <a href="#programming">Programming</a> <a href="#fun">Fun</a> </ul> <h><span id="economy">Economy & Finance</span></h> <a href="https://www.counterpunch.org/2023/02/20/why-would-anyone-think-republicans-are-interested-in-lower-budget-deficits/" author="Dean Baker" source="CounterPunch">Why Would Anyone Think Republicans are Interested in Lower Budget Deficits</a> <bq>It is also worth noting that much of the projected shortfall in the Social Security program is due to the upward redistribution of income over the last four decades. <b>In 1982, when the last major changes to Social Security were put into place, 90 percent of wage income fell below the cap on taxable wages (currently $160,200).</b> In the last two decades, just over 82 percent of wage income was subject to the Social Security tax (see page 148). There was also a redistribution from wage income to profit income, which further reduced Social Security tax revenue. Together, this upward redistribution accounts for more than 40 percent of the program’s projected shortfall over its 75-year planning horizon. <b>If we had not shifted so much income to high-end earners and to profits, closing the projected shortfall in Social Security would be a far more manageable task.</b></bq> <hr> <a href="https://www.economist.com/finance-and-economics/2023/02/13/war-and-subsidies-have-turbocharged-the-green-transition" source="Economist" author="">War and subsidies have turbocharged the green transition</a> <bq>All over the world officials are raising renewables targets and setting aside huge sums to bankroll a buildout.</bq> You are so full of shit. Nobody is doing anything but flapping their lips and scooping cash. <bq>From December 2021 to October 2022 contract prices for the continent’s wind and solar photovoltaic projects were on average 77% below wholesale power prices. <b>At €257 per megawatt-hour ( mw h), the average price in Germany in December, a typical solar plant takes less than three years to become profitable, against 11 years at €50 per mw h, the average spot price between 2000 and 2022.</b></bq> Why does solar costing more than five times as much bode well? Oh...because the increase in prices generally mean that solar is more quickly becoming profitable. Solar didn't become cheaper---fossil fuels became more expensive. People are paying way more for energy, so the energy companies are making money hand over fist, which is why the Economist finally has a boner about it. You see, you can't just have what you want---it has to be considered economically viable by those who already control the world. The Economist's philosophy is to do only that which makes money. Imagine their relief when they realized that they could support green power while still maintaining strong class divisions. <bq>Namit Sharma of McKinsey reckons that by 2030 the EU will have to quadruple the number of people developing, building and running green plants in order to meet its targets.</bq> And they'll come from where? We drain all the brains to finance, as you well know. <bq>But in practice miserly new national rules and auction designs make doing so difficult. This winter the EU adopted a windfall tax on renewable-energy generators, and a cap on wholesale power prices, in effect placing a ceiling on returns. Germany’s new offshore-wind-tender system makes bidders compete over how much they are willing to pay to run projects, a system known as “negative bidding”. Never-ending permitting wrangles dilute returns even further. <b>In an alternative, less protectionist universe America’s and Europe’s vast spending plans would have an even bigger impact.</b></bq> There it is: the Economist always sees the world as too protectionist. This is why it's impossible to take the Economist seriously: everything bad is always the fault of the government and its purported failure to allow capital and visionaries to solve all of our problems for us. After sixty years, they're still just shouting John Gault at every problem. <bq>As green power is boosted and fossil-fuel use sags, the global economy is now expected to belch out much less carbon dioxide than had been predicted just 12 months ago.</bq> Show me their relative percentages of usage, not relative growths. If we hit our target in fifty instead of ten years, we will still have failed. The Economist is handing out participation trophies. <h><span id="politics">Public Policy & Politics</span></h> <a href="https://www.nachdenkseiten.de/?p=93877" source="NachDenkSeiten" author="Oskar Lafontaine">Hat der Dritte Weltkrieg bereits begonnen?</a> <bq>Die Vereinigten Staaten könnten sich ebenso wenig wie Russland aus diesem Krieg zurückziehen. <b>Die USA kämpften um ihre Stellung als alleinige Hegemonialmacht und seien dennoch in Gefahr, die Währungs- und Finanzkontrolle über die Welt zu verlieren</b> und damit auch die Möglichkeit, ihr riesiges Handelsdefizit umsonst zu finanzieren.</bq> <bq>Zwischen der offensiven Strategie der Amerikaner und der defensiven Strategie der Russen befänden sich die Europäer in einem atemberaubenden Zustand der geistigen Verwirrung. Das gelte ganz besonders für Deutschland. <b>Die NATO sei heute ein »Washington-London-Warschau-Kiew-Block«.</b></bq> <bq>Emmanuel Todd vertritt die Auffassung, Briten und Polen hätten sich an dem von den USA zu verantwortenden Sabotageakt, der zu der Zerstörung der Nordstream-Gasleitungen führte, beteiligt.</bq> <hr> <a href="https://jacobin.com/2023/02/seymour-hersh-interview-nord-stream-pipeline/" source="Jacobin" author="Fabian Scheidler">Seymour Hersh: The US Destroyed the Nord Stream Pipeline</a> <bq>And I’ll tell you something else. <b>The people in America and Europe who build pipelines know what happened.</b> I’m telling you something important. The people who own companies that build pipelines know the story. I didn’t get the story from them but I learned quickly they know.</bq> <bq><b>I don’t think they thought it through. I know this sounds strange. I don’t think that Blinken and some others in the administration are deep thinkers.</b> There certainly are people in the American economy who like the idea of us being more competitive. We’re selling LNG, liquefied gas, at extremely big profits; we’re making a lot of money on it. I’m sure there were some people thinking, boy, this is going to be a long-time boost for the American economy. But <b>in that White House, I think the obsession was always reelection, and they wanted to win the war, they wanted to get a victory, they want Ukraine to somehow magically win.</b></bq> <bq>It doesn’t matter what I think. What I know is there’s no way this war is going to turn out the way we want, and I don’t know what we’re going to do as we go further down the line. <b>It scares me if the president was willing to do this.</b></bq> <bq>And in the long run, this is going to be very detrimental not only to his reputation as the president but politically too. <b>It’s going to be a stigma for America.</b></bq> <bq>The one virtue of the CIA is that a president, who can’t get his agenda through Congress and nobody listens to him, can take a walk in the backyard of the Rose Garden of the White House with the CIA director and somebody can get hurt eight thousand miles away. That’s always been the selling point of the CIA, which I have problems with. But <b>even that community is appalled that he chose to keep Europe cold in support of a war that he’s not going to win.</b> And that, to me, is heinous.</bq> <bq>What’s courageous about telling the truth? Our job isn’t to be afraid. And sometimes it gets ugly. There have been times in my life, when — you know, I don’t talk about it. <b>Threats aren’t made to people like me; they’re made to children of people like me. There’s been awful stuff. But you don’t worry about it — you can’t. You have to just do what you do.</b></bq> <hr> <a href="https://www.nachdenkseiten.de/?p=93803" source="NachDenkSeiten" author="Jens Berger">Neue Bunker braucht das Land – ja ist die Politik des Wahnsinns fette Beute?</a> <bq>Gigantische Untergrundkomplexe, die sich in „Friedenszeiten“ zivil nutzen lassen – als Sportstätte, Schwimmbad oder Theater. Na klar, <b>da man derartige oberirdische Einrichtungen durch Einsparungen der öffentlichen Gelder vor die Hunde gehen lässt, kann man sie nun dem Volk unterirdisch als Zückerli für ein absurdes Projekt schmackhaft machen.</b></bq> <bq>Wenn man <b>unterirdisch Spiel, Spaß und Spannung mit dem „Schutz vor Putins Bomben“ vereinen kann, ist das offenbar ganz im Sinne der herrschenden Politik</b> und seitens der Medien gibt es nicht etwa Kritik, sondern kindischen Beifall</bq> <hr> <a href="https://scheerpost.com/2023/02/15/diana-johnstone-demonstrate-together/" source="Scheer Post" author="Diane Johnstone">Demonstrate Together</a> <bq>Leftist militants who believe a man can be transformed into a woman should have no trouble believing that a libertarian might be transformed into a socialist. Such miracles do occur.</bq> This is flip, but has a kernel of truth: we now believe that ideology is more fixed than biological gender. How crazy is that? <bq>This “legitimize” threat is merely the other side of the “guilt by association” coin. Both are used to <b>evade discussion of serious matters by treating political convictions as if they were incurable contagious diseases.</b></bq> <bq>When the subject is WAR, if you can join in opposition only with people who agree with you about everything else, you have lost the sense of common humanity.</bq> <bq>Wherever you see popular resistance to war begin to come to life, go to it and make it belong to everybody.</bq> <hr> <a href="https://jacobin.com/2023/02/mussolini-in-beijing/" source="Jacobin" author="Ho-Fung Hung">Mussolini in Beijing</a> <bq>In December 2021, Xi made a speech at the Central Economic Work Conference attacking welfarism and pledging China would not opt for a model that would “uplift a group of lazy people who gain without working,” with explicit derogatory references to Latin American “populism.” <b>This hostility toward welfare could be found in any speech from any free-market fundamentalist in any capitalist country — lip service to Karl Marx and Mao Zedong aside.</b></bq> <hr> <a href="https://jacobin.com/2023/02/democrats-keep-handing-working-class-voters-to-republicans/" source="Jacobin" author="David Sirota">Democrats Keep Handing Working-Class Voters to Republicans</a> <bq>Yet laughing at the GOP’s fake populists as if they are politically irrelevant ignores a significant and dangerous trend: <b>Democrats’ genuflections to their corporate donors</b> — whether breaking a strike, authorizing corporate giveaways, or stalling a $15 minimum wage — <b>have been handing conservatives myriad opportunities to court working-class voters.</b></bq> <bq>[...] the threat of a realignment will persist if Democrats remain complacent. What does that complacency look like in practice? In the current moment, it would look like Democrats using a lame-duck session of Congress to pass strikebreaking legislation against workers trying to get paid sick days, then refusing to extend the child tax credit while preserving tax breaks for private equity billionaires — all things that have happened in the final weeks of the party’s control of Congress. That this is even a possibility <b>illustrates the “let them eat cake” nonchalance among Democratic leaders who believe they can coast on the assumption that GOP extremism makes Republicans unelectable.</b></bq> <hr> <a href="https://www.wsws.org/en/articles/2023/02/15/gben-f15.html" source="WSWS" author="Andre Damon">NATO pledges to “ramp up” defense production for “grinding war of attrition”</a> <bq>Russian forces are reportedly on the verge of a significant victory in the city of Bakhmut, amid warnings by the Financial Times that <b>Russia is preparing to increase its use of helicopters, fighters and bombers, which it has up to this point held back.</b></bq> <bq>Speaking at a meeting of NATO defense ministers in Brussels Tuesday, Stoltenberg declared, “The war in Ukraine is consuming an enormous amount of munitions and depleting Allied stockpiles. <b>The current rate of Ukraine’s ammunition expenditure is many times higher than our current rate of production. This puts our defense industries under strain. So we need to ramp up production.</b> And invest in our production capacity.</bq> Fuck you Stoltenberg; you and your boner for war are going to get so many people killed. <bq>Last month, <b>the Pentagon announced that it plans to increase ammunition production by 500 percent, to levels last seen during the Korean War.</b> The National Defense Authorization Act passed last year gave the military wartime procurement powers, allowing the Pentagon to carry out no-bid contracts, nominally in the name of increasing production.</bq> <hr> <a href="https://www.wsws.org/en/articles/2023/02/15/ruoc-f15.html" source="WSWS" author="Ela Maartens">European Union decides on massive intensification of assault on refugees</a> <bq>In contrast, the devastating earthquake disaster that had devastated the Turkey-Syria border region just two days earlier was not a topic of discussion. <b>The EU, ostensibly founded on freedom, democracy, equality, the rule of law, human rights and human dignity, has no response to the devastation occurring on its periphery, which has affected some 23 million people.</b> Hundreds of thousands have lost their loved ones, their homes and everything in the Turkish-Syrian border region. But European governments stubbornly stick to their murderous deportation routine.</bq> <bq>Based on the EU’s so-called mass influx directive, 4.8 million Ukrainians received temporary protection status. <b>Ukrainian war refugees are largely not included in official EU statistics under this measure. Ukrainian nationals are also automatically granted humanitarian residence permits in EU member states</b>, giving them access to education, employment, social benefits and medical care.</bq> <hr> <a href="https://thegrayzone.com/2023/02/14/us-crushed-struggle-somali-nation/" source="The Gray Zone" author="Ann Garrison">How the US crushed the struggle for a Somali nation</a> <bq>The US, UK, EU and NATO countries want to keep Somalia weak and fragmented, so that they can continue the toxic dumping and fish looting and the oil exploitation that is now in its early stages. And because, as you said, Somalia is so geostrategically located. <b>The United Arab Emirates would like to annex Somalia’s Puntland state, where they already have economic and political power.</b></bq> <bq>The Gulf States and other neighboring countries are scrambling to control Somali ports. <b>The UAE is already operating the Berbera port in Somaliland, the one where the US wants a naval base, without the agreement of the federal government.</b> It is also in the process of buying into the Bosaso port in Puntland.</bq> <bq>I visited Somalia’s neighbor <b>Eritrea</b>, which seemed to be the opposite of Somalia. I saw <b>a poor country on a slow but steady development path, with a calm, relaxed atmosphere. It was peaceful, no one was begging or sleeping on the streets</b>, and I never once thought to clutch my pocketbook, but I can’t remember seeing any armed military or police. It may have just been so calm that I failed to notice a cop or two. ABDIWAHAB SHEIKH ABDISAMAD: I have been to Eritrea too and I confirm everything you say. Eritrea is not the way the Western media are reporting. They are wholly negative about Eritrea. But the opposite of everything they say is true.</bq> <bq>Eritrea is working to become fully food sufficient in 2030 and so far, they have cultivated 600,000 hectares out of 2.1 million hectares of arable land. <b>In Eritrea housing, education, and health are free or relatively inexpensive compared to neighboring states.</b> Eritrea is also a debt free country. It has escaped the debt trap crippling most African nations, but none of its countless Western critics ever mention that. <b>They hate Eritrea being debt free because that means it can’t be strangled into submission by the IMF, the World Bank</b>, and the other global banking operations.</bq> <bq>In Somalia there are US troops, AFRICOM troops, UN troops, and clan militias — all of which are supposed to be fighting Al Shabaab. <b>How could all those troops, with all that firepower, fight Al Shabaab for 14 years without defeating them? You won’t find an eight-year-old child in Somalia who believes that all those troops are really fighting Al Shabaab!</b></bq> <bq>Al Shabaab is the US excuse for military presence to control resources and dominate militarily. <b>But Al Shabaab would not exist if the US had not organized the Ethiopian proxy invasion and occupation of Somalia from 2006 to 2009</b>, when the Tigray People’s Liberation Front (TPLF) was still in power in Ethiopia.</bq> <bq>The neocolonialists fix the democratic game by financing candidates that best meet their requirements: access to raw materials for their multinationals, support in foreign affairs, etc. <b>With the multi-party system in Africa, the imperialists tell you every 4 or 5 years, ‘Go and vote for the candidates we have chosen for you. They will make you poor and kill you. Vote for them!’”</b></bq> Same-same in the U.S., to be honest. <bq>He refused to sign an agreement to the continuation of Project Atalanta, the European Union navy patrol off Somalia’s coast. He thought Somalia should have its own navy and its own coast guard to stop all the fish looting and toxic dumping that this EU navy has no interest in stopping. <b>Many Somalis believe that Project Atalanta actually facilitates all this mostly European coastal aggression.</b></bq> <hr> <a href="https://chinadigitaltimes.net/2023/02/word-of-the-week-huminerals-人矿-ren-kuang/" source="China Digital Times" author="人矿 RÉN KUÀNG">Word of the week: Huminerals</a> <bq>Huminerals power the motors that turn the wheels of history. Huminerals have few other choices: either fuel history’s engine, or be ground beneath its wheels. Of course the inverse is true. If huminerals were to stop propelling history, then those other huminerals who abstained would not be crushed. <b>Yet there are always huminerals who see more value in a lifetime of being fuel than to risk being flattened.</b></bq> <hr> <a href="https://scheerpost.com/2023/02/14/60-minutes-weight-loss-tip-dont-bite-the-hand-that-feeds-you/" source="Scheer Post" author="Julie Hollar">60 Minutes' Weight-Loss Tip: Don't Bite the Hand That Feeds You</a> <bq>Novo Nordisk recently predicted record earnings as a result of demand for Wegovy, with operating profits expected to increase by up to 19% (Bloomberg, Jan. 2, 2023)—from a company that made $8 billion in profit last year. And this is in an industry that already regularly expects profit margins of 15–20%—<b>Novo Nordisk’s 2022 profit margin was 31%—as compared to 4–9% for non-drug companies.</b></bq> <bq><b>Canada, like Norway and Denmark, has negotiated prices with drug companies, rather than letting them set whatever wildly inflated prices they desire</b>, which leads to those eye-popping profits. (The Inflation Reduction Act passed last year does include provisions giving Medicare the power to negotiate prices for some drugs, with the first negotiated prices to go into effect in 2026.)</bq> <bq>Because of US government policies that favor drug companies over people, <b>prices for brand-name drugs are 3.5 times higher in the US than in other high-income countries</b> (Commonwealth Fund, 11/17/21 ).</bq> <bq><b>Advertisers footing a corporate news outlet’s bills generally don’t have to tell them how to report, because those outlets understand the perils of biting the hand that feeds them.</b> If that segment had been submitted by Novo Nordisk as a paid advertisement, it would have come under more oversight than it did by 60 Minutes</bq> <hr> <a href="https://scheerpost.com/2023/02/14/patrick-lawrence-objectivity-and-its-discontents/" source="Scheer Post" author="Patrick Lawrence">Objectivity and Its Discontents</a> <bq>Bear with me a sec, as I have to turn these concussing numbers upside down to absorb them: <b>84 of every 100 Americans do not think newspapers report events truthfully; 89 of every 100 Americans consider what they see and hear in nightly newscasts unreliable.</b></bq> <bq>[...] the name of the uncorrupted ideal. This is the ideal of objective reason, which dates to the ancient Greeks. It requires that thought be conducted without reference to the desirability of its conclusions. To make any such reference is to succumb to subjective reason. <b>Socrates taught us that reason should determine belief: To allow belief to determine reason is the corruption of subjective reason.</b> The late Robert Parry, a journalist of impeccable integrity, put the case for objective reason this way: <b>“I don’t care what the truth is. I just care what the truth is.”</b></bq> <bq>Ideals are never fully realized: This is so by definition, and certainly it holds for journalists. But ideals are to be striven for nonetheless. <b>From the moment an editor or reporter decides which story to cover and which to leave alone, personal judgments and all that inform them are at work. There is nothing to be done about this</b> and only one sound way for journalists to think about it. This requires an understanding of one’s responsibilities, quite special responsibilities, and the discipline to honor them.</bq> <bq>[...] a reading of any major daily’s front page on any given day makes this point quite clearly: <b>We find the same sonorous, authoritative diction and the same faux disinterest used to naturalize contempt for whomever or whatever the press wants to attack and to approve of whatever it wishes to favor.</b> It is by way of this professional sleight of hand that advocates of subjectivity propose to advance their ideological proclivities as none other than objective truth. We are back in the 1920s.</bq> <hr> <a href="https://www.cjr.org/special_report/trumped-up-press-versus-president-part-2.php/" source="Columbia Journalism Review" author="Jeff Gerth">The press versus the president, part two</a> <bq>Spayd, in an email to me, complained that the Times had “two standards.” Before the election, she wrote, the October 31 piece was “downplayed” because the paper “didn’t know whether the allegations held up,” but after the election, “the Times produced a steady stream of stories about whether Trump conspired with Russians to win the election without knowing whether the allegation was actually true.” <b>Trump told me he noticed the difference in coverage once he took office. Not only did he have to run the country, he had to fight off “unbelievably fake” stories.</b> Spayd, a former editor of CJR, left the Times a few months after the column was published, and the position of public editor was ultimately abolished.</bq> <a href="https://www.cjr.org/special_report/trumped-up-press-versus-president-part-4.php" source="Columbia Journalism Review" author="Jeff Gerth">The press versus the president, part four</a> <bq>Reporters who ferreted out the details of the FBI inquiry into Trump’s campaign couldn’t, or wouldn’t, confirm the Justice Department investigation into the future president’s son. <b>Whereas the specter of purported Russian ties to Trump spurred an explosion of social media and journalistic interest, this time Twitter and Facebook temporarily curbed the reach of the Post story.</b></bq> <hr> <a href="https://www.counterpunch.org/2023/02/13/remembering-vladimir-putins-speech-of-10-february-2007-at-the-munich-security-conference/" source="CounterPunch" author="Alfred De Zayas">Remembering Vladimir Putin’s speech of 10 February 2007 at the Munich Security Conference</a> <bq>Allow me to <b>quote Kennedy</b>: “while defending our own vital interests, <b>nuclear powers must avert those confrontations which bring an adversary to a choice of either a humiliating retreat or a nuclear war.</b> To adopt that kind of course in the nuclear age would be evidence only of the bankruptcy of our policy—or of a collective death-wish for the world.”[</bq> <bq author="George Kennan"><b>Why, with all the hopeful possibilities engendered by the end of the Cold War, should East-West relations become centered on the question of who would be allied with whom and, by implication, against whom in some fanciful, totally unforeseeable and most improbable future military conflict?</b>…[B]luntly stated…expanding NATO would be the most fateful error of American policy in the entire post-Cold War era. Such a decision may be expected to inflame the nationalistic, anti-Western and militaristic tendencies in Russian opinion; to have an adverse effect on the development of Russian democracy; to restore the atmosphere of the cold war to East-West relations, and to impel Russian foreign policy in directions decidedly not to our liking …</bq> <bq><b>Most people in the West were and remain unaware of Putin’s speech or for that matter of the texts of the two proposals that he put on the table in December 2021</b>, two draft treaties solidly anchored in the UN Charter concretising the necessity of agreeing on a modus vivendi and building a security architecture for Europe and the world.</bq> <bq><b>Imagine if all the financing that went and still goes into the military</b>, military bases, procurement of tanks, missiles and nuclear weapons <b>became available for financing education, health, housing, infrastructure</b>, research and development!</bq> <bq>Western politicians gloated over the fact that Russia would not be able to do anything about our breach of trust. <b>We cheated, as we so often cheat in international relations.</b> I would even say that we have developed a “culture of cheating”[4], of taking advantage of the other guy whenever possible. <b>It is perceived almost as cleverness, a secular virtue.</b></bq> <bq>I remember the negative caricatures in the press and the incessant defamation of the Russians as totalitarians. <b>It is the artificial creation of such negative feelings toward other peoples and cultures that facilitates war propaganda and serves to justify sanctions and war crimes</b>, all of this in violation of article 20 of the International Covenant on Civil and Political Rights and in violation of the UNESCO Constitution.</bq> <hr> <a href="http://exileinhappyvalley.blogspot.com/2023/02/confessions-of-keyboard-forest-defender.html" source="Exile in Happy Valley" author="Nicky Reid">Confessions of a Keyboard Forest Defender</a> <bq>Tortuguita was shot dead by a joint police task force that included federal agents during the latest in a long series of raids on these encampments. Police claim that Tortuguita fired first, injuring one of their co-conspirators in uniform in the process but their story continues to change, and no body-camera footage actually exists of the alleged shootout. <b>I honestly don't know if Tortuguita fired first, but quite frankly, I don't care. He was an American citizen on public property that was being stolen at gunpoint by a runaway police state.</b></bq> <bq><b>$30 million of the taxpayer's hard-earned money has been pilfered by the city of Atlanta and awarded to a conglomeration of private corporations to hijack the commons of a community that was never offered a vote on whether or not they wanted the fucking Death Star built in their backyard.</b> This land belongs to that community, to the children who find shelter from a hateful world beneath its branches and as far as I'm concerned, volunteers like Tortuguita have every right to defend that land by any means that community finds necessary.</bq> <bq>Turn on the news and you can find some braying asshole willing to blame the rise of violent nihilism in first world society on everything from handguns to puberty blockers, but <b>nobody seems willing to consider the fact that human beings are animals and animals tend to become violent when they live in cages.</b> That is precisely what the modern city has become, a sprawling kennel for domesticated beasts and I believe that this is killing us</bq> <bq>The people shooting up Walmart's and pushing people in front of subway trains aren't evil, they're rabid. <b>It is the society that fosters this desperation for profit that is evil, and it must be smashed before it smashes us all.</b></bq> <hr> <a href="https://scheerpost.com/2023/02/12/pentagon-wants-to-return-special-ops-propagandists-to-ukraine/" source="Scheer Post" author="Caitlin Johnstone">Pentagon Wants To Return Special Ops Propagandists To Ukraine</a> <bq>The US empire has been frantically ramping up propaganda and censorship because the “great power competition” it has been preparing against Russia and China is going to require economic warfare, massive military spending, and nuclear brinkmanship that no one would consent to without lots of manipulation. <b>Nobody’s going to consent to being made poorer, colder, and less safe over some global power struggle that doesn’t benefit them unless that consent is actively manufactured.</b></bq> <hr> <a href="https://www.nachdenkseiten.de/?p=93672" source="NachDenkSeiten" author="Florian Schwinn">Was Willi will</a> <bq>„Sie wollen keine Massentierhaltung, keine Gentechnik, keine Monokulturen, keine ‚Pestizide‘ (die wir Pflanzenschutzmittel nennen), und auch sonst stellen Sie viele Ansprüche. <b>Als Verbraucher kaufen Sie (trotz allen anderen Geredes immer noch) wenig regional, wenig saisonal, wenig Bio. Sie kaufen vor allen Dingen billig.“</b></bq> <bq>„Die Überschrift über allem lautet: <b>Wir Bauern können alles.</b> Wir können Naturschutz, wir können Artenschutz, wir können Klimaschutz, wir können Tierwohl.“ <b>Das allerdings müsste dann bitte auch bezahlt werden.</b></bq> <bq><b>Noch nie in der Geschichte waren die Lebensmittelkonsumenten weiter von den -produzenten entfernt als heute.</b> Der Verbraucher weiß schon lange nicht mehr, wie das Essen entsteht, das täglich auf seinem Teller landet, wie es angebaut, gezüchtet, geerntet, geschlachtet, verarbeitet oder kurz: hergestellt wird.“</bq> <bq>Es geht darum, die Kluft zwischen den Esserinnen und Essern und den Essenmacherinnen und Essensmachern zu überwinden. <b>Nur wenn wir wieder lernen, wie und unter welchen Bedingungen unsere Lebensmittel entstehen, können wir den Bäuerinnen und Bauern, und damit auch uns, beim Überleben helfen.</b></bq> <hr> <a href="https://chrishedges.substack.com/p/there-are-no-permanent-allies-only" source="Substack" author="Chris Hedges">There Are No Permanent Allies, Only Permanent Power</a> <bq>To those who suffer directly from U.S. aggression, these demands are not academic and theoretical issues. The victims of this militarism do not have the luxury of virtue-signaling. They want the rule of law to be reinstated and the slaughter stopped. So do I. They welcome any ally who opposes endless war. For them, it is a matter of life or death. If some of those on the right are anti-war, if they also want to free Julian Assange, it makes no sense to ignore them. <b>These are urgent existential issues that, if we do not mobilize soon, could see us slip into a direct confrontation with Russia, and perhaps China, which could lead to nuclear war.</b></bq> <bq>“A left-right alliance on issue after issue, whether it’s on a living wage, ending endless wars of aggression by the United States; whether it’s striking down hard on corporate crime, fraud and abuse; whether it’s universal health insurance is an unbeatable movement,” Nader told me when I reached him by phone. “<b>Just think of a senator receiving ten constituents from back home and five are liberals and five are conservatives. How is a senator going to game them? How is a senator going to sugarcoat them? It’s very difficult.</b></bq> <hr> <a href="https://zeitgeschehen-im-fokus.ch/de/newspaper-ausgabe/nr-1-vom-18-januar-2023.html" source="Zeitgeschen im Fokus" author="General a. D. Harald Kujat">Ukrainekonflikt: «Jetzt wäre der richtige Zeitpunkt, die abgebrochenen Verhandlungen wieder aufzunehmen»</a> <bq>Die Ukraine kämpft um ihre Freiheit, um ihre Souveränität und um die territoriale Integrität des Landes. Aber die beiden Hauptakteure in diesem Krieg sind Russland und die USA. Die Ukraine kämpft auch für die geopolitischen Interessen der USA. Denn <b>deren erklärtes Ziel ist es, Russland politisch, wirtschaftlich und militärisch so weit zu schwächen, dass sie sich dem geopolitischen Rivalen zuwenden können, der als einziger in der Lage ist, ihre Vormachtstellung als Weltmacht zu gefährden: China.</b></bq> <bq>Man muss sich die heutige Situation einmal vorstellen. Die Leute, die von Anfang an Krieg führen wollten und immer noch wollen, haben den Standpunkt vertreten, <b>mit Putin kann man nicht verhandeln. Der hält die Vereinbarungen so oder so nicht ein. Jetzt stellt sich heraus, wir sind diejenigen, die internationale Vereinbarungen nicht einhalten.</b></bq> <bq>«Niemand bezweifelt den grossen Wert der Beziehungen zwischen Europa und den Vereinigten Staaten. Aber ich bin der Meinung, dass Europa seinen Ruf als mächtiger und selbständiger Mittelpunkt der Weltpolitik langfristig nur festigen wird, wenn es seine Möglichkeiten mit den russischen menschlichen, territorialen und Naturressourcen sowie mit den Wirtschafts-, Kultur- und Verteidigungspotenzialen Russlands vereinigen wird.» <b>Mit dieser Rede hat Putin zu Beginn seiner Präsidentschaft seine aussenpolitischen Ziele formuliert. Was ist danach geschehen? Nichts, was ernsthaft den weitsichtigen Überlegungen Putins Rechnung getragen hätte.</b></bq> <bq>Doch die EU im Verbund mit der Nato führte ihre Politik weiter. Höhepunkt der Entwicklung war der Putsch gegen den Staatspräsidenten der Ukraine, bei dem die USA unbestritten die Finger im Spiel hatten. Das abgehörte Telefonat, in dem die Aussenbeauftrage der USA, Victoria Nuland, mit dem amerikanischen Botschafter in Kiew, Goeffrey Pyatt, die neue Regierung in der Ukraine besprachen, als der gewählte Präsident noch in Amt und Würden war, legt ein beredtes Zeugnis von US-amerikanischer Einmischung in die inneren Angelegenheiten eines Staates ab. <b>Dies stellt einen Verstoss gegen die Uno-Charta, also einen Völkerrechtsbruch, dar. Danach nahm die Geschichte ihren Lauf.</b></bq> And Victoria Nuland rides on, Cheney-esque in her career arc. <bq>Der Konflikt zwischen dem Westen und Russland, der in der Ukraine ausgetragen wird, hat also eine lange Vorgeschichte, die den wenigsten bekannt sein wird und die auf unseren Informationskanälen, sprich Medien, nicht thematisiert wird. Für neutrale Staaten wie die Schweiz würde das äusserste Zurückhaltung in einseitigen Schuldzuweisungen bedeuten. Leider ist das Gegenteil passiert. <b>Die Schweiz, insbesondere in der Person von Ignazio Cassis, hat sich, unbesehen aller Ereignisse im Vorfeld des Konflikts, in moralischer Überhöhung auf die Seite der Ukraine gestellt und damit die Neutralität schwer geschädigt.</b></bq> <hr> <a href="https://www.wsws.org/en/articles/2023/02/20/fnlj-f20.html" author="Peter Symonds" source="WSWS">US menaces China over balloon, arms for Russia</a> <bq>Blinken bluntly told Yi that the Chinese balloon which strayed into US airspace late last month was an “unacceptable violation of US sovereignty and international law” and “must never occur again,” according to State Department spokesperson Ned Price.</bq> <bq>Wang confirmed that the meeting with Blinken had taken place and called on the US to repair the “damage” to the countries’ relations. He told the media: “To have dispatched an advanced fighter jet to shoot down a balloon with a missile, such behaviour is unbelievable, almost hysterical.”</bq> <bq>The extraordinary way in which a vagrant Chinese research balloon has been absurdly inflated into a major threat to US security is a measure of just how reckless and far advanced US war preparations are.</bq> <hr> <a href="https://www.counterpunch.org/2023/02/20/spying-vs-spying/" author="John Feffer" source="CounterPunch">Spying vs. Spying</a> <bq>Meanwhile, this definitely happened: in a rare show of unanimous bipartisanship, the House of Representative voted 491-0 to condemn China over its balloon belligerence.</bq> <h><span id="journalism">Journalism & Media</span></h> <a href="https://freddiedeboer.substack.com/p/the-bitter-end-of-content" author="Freddie De Boer" source="SubStack">The Bitter End of "Content"</a> <bq><b>The video is nonsensical, not in some avant garde way but to fulfill its economic purpose.</b> Leaving the viewer confused as to what exactly is being conveyed is a feature, not a bug - the more people are baffled by the video, the more they’ll comment on it to register their confusion, the more times they’ll send it to friends to try and figure out that which cannot be figured out. It is “content,” to use that wretched term, that is devoid of content, <b>a human centipede of virality, monetizing fleeting interest. It’s the inevitable outcome of every bad incentive we’ve created online.</b></bq> <bq>All of this was eminently predictable based on the values that we baked into online life years ago. This road was always going to lead to nowhere. The internet is like a person you know who you think can’t possibly stoop any lower, and then manages to pull it off, over and over again. <b>Years ago, the idea that online life would be dominantly funded by advertising coalesced into conventional wisdom, and we’ve been living with the consequences ever since.</b> The utopian assumption that views and clicks would accrue to the highest-quality content failed to understand a basic lurking reality - that the monetization of attention leads inevitably to the weaponization of attention. You can get eyeballs on your work by having talent and working diligently. Or <b>you can get eyeballs by exploiting the system. And the worst part is that the big players have no particular financial incentive to challenge that exploitation.</b></bq> <bq>The marketplace of attention was supposed to solve this problem; we were told that the good channels would be elevated by the platforms and that people would stop watching the bad channels. But <i>the marketplace of attention cares only about attention.</i> The assumption that low-quality or dishonest or dangerous content wouldn’t get clicked on was always entirely and obviously wrong. The videos are slickly made; the results look plausible. People try them, fail, and assume it’s their own fault. And <b>since the videos get views and views make money, the platforms have no reason to do anything about them.</b></bq> <bq><b>So long as advertising is the dominant funding source of the online world, any and every creative platform will be a race to the bottom.</b> People will find ways to abuse the system to receive attention and money based on nothing more than manipulation.</bq> <bq>[...] advertising has been ingrained into the internet as the basic model for so long and to such an extent that it’s hard to envision online life without the systematic manipulation of attention and all its evils. So we’re bound to wind up here, at the bitter end of “content.” Which is <b>a good excuse to withdraw deeper into books, movies, albums, and art, stuff that was created for a deeper purpose than mining fleeting bits of attention for fractions of a penny.</b> The question is whether generations who have grown up immersed in these platforms can imagine life without them, and <b>whether we're cursed to live with them ourselves.</b></bq> <h><span id="science">Science & Nature</span></h> <a href="https://arstechnica.com/science/2023/02/the-bird-flu-outbreak-has-taken-an-ominous-turn/" source="Ars Technica" author="Maryn McKenna">The bird flu outbreak has taken an ominous turn</a> <bq>That includes the US, where <b>43 million laying hens were either killed by avian flu last year or slaughtered to prevent the disease from spreading.</b> Those losses took out almost a third of the national flock of laying hens;</bq> <bq>“When there’s public discussion of addressing zoonotic disease, it almost immediately turns to vaccination, preparedness, biosecurity—but no one discusses addressing the root cause,” says Jan Dutkiewicz, a political economist and visiting fellow at Harvard Law School’s Brooks McCormick Jr. Animal Law and Policy Clinic. <b>“We would never have a debate about preventing cancer from tobacco products without talking about stopping smoking. Yet when it comes to zoonotic disease risk, there is a huge reticence to discuss curbing animal production.”</b></bq> <hr> <a href="https://arstechnica.com/science/2023/02/mini-robot-shifts-from-solid-to-liquid-to-escape-its-cage-just-like-the-t-1000/" source="Ars Technica" author="Jennifer Ouellete">Mini-robot shifts from solid to liquid to escape its cage—just like the T-1000</a> <bq>The new mini-robot is made of magneto-active phase transitional matter (MPTM), capable of switching back and forth between solid and liquid states. <b>When the MPTM is heated with an alternating magnetic field, it melts into a liquid, while ambient cooling lets it resolidify when the magnetic field is removed.</b></bq> <bq>[...] <b>the team demonstrated a minimally invasive miniature machine that removes a foreign object from an artificial model stomach filled with water.</b> Here, one would need to tailor the melting point to be a bit higher than human body temperature (around 38° C, or 100° F) by embedding the microparticles in a gallium-based alloy instead of pure gallium.</bq> <bq>[...] two remotely controlled MPTMs passed through a narrow space and settled to the top of threaded holes. <b>Upon heating, the MPTMs turned into liquid and filled the threaded holes to form screws</b>, which solidified when cooled, thereby joining two plastic plates together.</bq> <h><span id="philosophy">Philosophy & Sociology</span></h> <a href="https://astralcodexten.substack.com/p/trying-again-on-fideism" source="Astral Codex Ten" author="Scott Alexander">Trying Again On Fideism</a> <bq>I get emails every single day from P=NP crackpots and quantum mechanics crackpots and now AI crackpots too. <b>Some of them probably *would* be better off never trying to think for themselves again, and just Trusting Science and Trusting the Experts.</b> Sure, the experts are sometimes confidently wrong, but not as consistently so as they are! And for my part, I can’t possibly write 25,000 words to explain why each and every crackpot is wrong. As a matter of survival, I *have* to adopt a Kavanagh-like heuristic: “this person seems like an idiot.”</bq> <bq>When you take conspiracy theorists arguments seriously, it implies a higher prior on conspiracy theories than when you dismiss them out of hand. <b>This can lead to your readers (consciously or not) increasing their priors on conspiracy theories and being more likely to believe future conspiracy theories they come across.</b></bq> <bq><b>When people shrug off conspiracy theories easily, it’s either because the conspiracy theory isn’t aimed at them - the equivalent of an English speaker feeling smug for rejecting a sales pitch given entirely in Chinese</b> - or because they’re biased against the conspiracy theory with a level of bias which would also be sufficient to reject true theories.</bq> <bq>It’s easy for a scissor statement like “is the chess set black or white?” to become the basis for a social/political movement, which then evolves the anti-epistemology necessary to protect its own existence (I’m still in awe of the way <b>ivermectin advocates have made “small studies are more trustworthy than big studies” sound like a completely reasonable and naturally-arrived-at position</b>).</bq> <bq>What I do think is that “trust the experts” is an extremely exploitable heuristic, which <b>leads everyone to put up a veneer of “being the experts” and demand that you trust them.</b></bq> <bq>If you are too quick to seek epistemic closure because “you have to trust the experts”, <b>you will be easy prey to people misrepresenting what they are saying.</b></bq> If you believe liars…um, ok. But kind of obvious? <bq><b>When you run into conspiracy theories you don’t believe, feel free to ignore them. If you decide to engage, don’t mock them or feel superior. Think “there, but for the grace of God, go I.”</b> Get a sense of what the arguments for the conspiracy theory look like - not from skeptics trying to mock them, but from the horse’s mouth - so you have a sense of what false arguments look like. <b>Ask yourself what habits of mind it would have taken the people affected by the theory to successfully resist it.</b> Ask yourself if you have those habits of mind. Yes? ARE YOU SURE?</bq> <bq>If you feel tempted to believe something that has red flags for being a conspiracy theory, at least keep track of the Inside vs. Outside View. <b>Say “on the Inside View, this feels like the evidence is overwhelming; on the Outside View, it sounds like a classic conspiracy theory”. You don’t necessarily have to resolve this discomfort right away.</b> You can walk around with an annoying knot in your beliefs, even if it’s not fun. Look for the strongest evidence against the idea.</bq> <bq>All advice along the lines of “don’t do X unless you’re smart and sophisticated” is useless, because <b>everyone believes themselves smart and sophisticated.</b></bq> <h><span id="technology">Technology</span></h> <a href="https://writings.stephenwolfram.com/2023/02/what-is-chatgpt-doing-and-why-does-it-work/" source="" author="Stephen Wolfram">What Is ChatGPT Doing … and Why Does It Work?</a> <bq>ChatGPT effectively does something like this, except that (as I’ll explain) it doesn’t look at literal text; it looks for things that in a certain sense “match in meaning”. But the <b>end result is that it produces a ranked list of words that might follow, together with “probabilities”</b></bq> <bq>The big idea is to <b>make a model that lets us estimate the probabilities with which sequences should occur—even though we’ve never explicitly seen those sequences in the corpus of text we’ve looked at.</b> And at the core of ChatGPT is precisely a so-called “large language model” (LLM) that’s been built to do a good job of estimating those probabilities.</bq> <bq>It worth understanding that there’s never a “model-less model”. <b>Any model you use has some particular underlying structure</b>—then a certain set of “knobs you can turn” (i.e. parameters you can set) to fit your data. And in the case of ChatGPT, lots of such “knobs” are used—actually, 175 billion of them.</bq> There's also no such thing as no preconceptions or bias. <bq>The most popular—and successful—current approach uses neural nets. Invented—in a form remarkably close to their use today— in the 1940s, neural nets can be thought of as simple idealizations of how brains seem to work. In human brains there are about 100 billion neurons (nerve cells), each capable of producing an electrical pulse up to perhaps a thousand times a second. The neurons are connected in a complicated net, with each neuron having tree-like branches allowing it to pass electrical signals to perhaps thousands of other neurons. And <b>in a rough approximation, whether any given neuron produces an electrical pulse at a given moment depends on what pulses it’s received from other neurons—with different connections contributing with different “weights”.</b></bq> <bq>When we “see an image” what’s happening is that when photons of light from the image fall on (“photoreceptor”) cells at the back of eyes they produce electrical signals in nerve cells. These nerve cells are connected to other nerve cells, and eventually the signals go through a whole sequence of layers of neurons. And <b>it’s in this process that we “recognize” the image, eventually “forming the thought” that we’re “seeing a 2”</b></bq> <bq>Bigger networks generally do better at approximating the function we’re aiming for. And in the “middle of each attractor basin” we typically get exactly the answer we want. But <b>at the boundaries —where the neural net “has a hard time making up its mind”—things can be messier.</b></bq> <bq>At each stage in this “training” the weights in the network are progressively adjusted—and we see that eventually we get a network that successfully reproduces the function we want. So how do we adjust the weights? <b>The basic idea is at each stage to see “how far away we are” from getting the function we want—and then to update the weights in such a way as to get closer.</b></bq> <bq>[...] now consider differentiating with respect to these weights. It turns out that the chain rule of calculus in effect lets us “unravel” the operations done by successive layers in the neural net. And <b>the result is that we can—at least in some local approximation—“invert” the operation of the neural net, and progressively find weights that minimize the loss associated with the output.</b></bq> <bq>In other words—somewhat counterintuitively—it can be easier to solve more complicated problems with neural nets than simpler ones. And the rough reason for this seems to be that when one has a lot of “weight variables” one has a high-dimensional space with “lots of different directions” that can lead one to the minimum—whereas <b>with fewer variables it’s easier to end up getting stuck in a local minimum (“mountain lake”) from which there’s no “direction to get out”.</b></bq> <bq><b>But which of these is “right”? There’s really no way to say. They’re all “consistent with the observed data”.</b> But they all correspond to different “innate” ways to “think about” what to do “outside the box”. And some may seem “more reasonable” to us humans than others.</bq> <bq>[...] it’s hard to know if there are what one might think of as tricks or shortcuts that allow one to do the task at least at a “human-like level” vastly more easily. <b>It might take enumerating a giant game tree to “mechanically” play a certain game; but there might be a much easier (“heuristic”) way to achieve “human-level play”.</b></bq> <bq>[...] what we is that if the net is too small, it just can’t reproduce the function we want. But above some size, it has no problem—at least if one trains it for long enough, with enough examples. And, by the way, <b>these pictures illustrate a piece of neural net lore: that one can often get away with a smaller network if there’s a “squeeze” in the middle that forces everything to go through a smaller intermediate number of neurons.</b> (It’s also worth mentioning that “no-intermediate-layer”—or so-called “ perceptron ”-networks can only learn essentially linear functions—but as soon as there’s even one intermediate layer it’s always in principle possible to approximate any function arbitrarily well, at least if one has enough neurons, though to make it feasibly trainable one typically has some kind of regularization or normalization.)</bq> <bq>[...] indeed it’s a standard strategy to just show a neural net all the examples one has, over and over again. <b>In each of these “training rounds” (or “epochs”) the neural net will be in at least a slightly different state, and somehow “reminding it” of a particular example is useful in getting it to “remember that example”.</b> (And, yes, perhaps this is analogous to the usefulness of repetition in human memorization.)</bq> <bq>It’s also necessary to show the neural net variations of the example. And it’s a feature of neural net lore that those “data augmentation” variations don’t have to be sophisticated to be useful. <b>Just slightly modifying images with basic image processing can make them essentially “as good as new” for neural net training.</b></bq> <bq><b>The fundamental idea of neural nets is to create a flexible “computing fabric” out of a large number of simple (essentially identical) components</b>—and to have this “fabric” be one that can be incrementally modified to learn from examples.</bq> <bq>[...] neural net training as it’s now done is fundamentally sequential, with the effects of each batch of examples being propagated back to update the weights. And indeed with current computer hardware—even taking into account GPUs—most of a neural net is “idle” most of the time during training, with just one part at a time being updated. And in a sense this is because our current computers tend to have memory that is separate from their CPUs (or GPUs). But <b>in brains it’s presumably different—with every “memory element” (i.e. neuron) also being a potentially active computational element. And if we could set up our future computer hardware this way it might become possible to do training much more efficiently.</b></bq> <bq>And in the end there’s just a fundamental tension between learnability and computational irreducibility. <b>Learning involves in effect compressing data by leveraging regularities. But computational irreducibility implies that ultimately there’s a limit to what regularities there may be.</b></bq> <bq>[...] there’s an ultimate tradeoff between capability and trainability: the more you want a system to make “true use” of its computational capabilities, the more it’s going to show computational irreducibility, and the less it’s going to be trainable. And <b>the more it’s fundamentally trainable, the less it’s going to be able to do sophisticated computation.</b></bq> <bq>[...] computers can readily compute their individual steps. And instead what we should conclude is that <b>tasks—like writing essays</b>—that we humans could do, but we didn’t think computers could do, <b>are actually in some sense computationally easier than we thought.</b> In other words, the reason a neural net can be successful in writing an essay is because writing an essay turns out to be a “computationally shallower” problem than we thought. And in a sense <b>this takes us closer to “having a theory” of how we humans manage to do things like writing essays, or in general deal with language.</b></bq> <bq>And so, for example, we can <b>think of a word embedding as trying to lay out words in a kind of “meaning space” in which words that are somehow “nearby in meaning” appear nearby in the embedding.</b> The actual embeddings that are used—say in ChatGPT—tend to involve large lists of numbers.</bq> <bq>[...] instead of just defining a fixed region in the sequence over which there can be connections, transformers instead introduce the notion of “ attention ”—and the idea of “paying attention” more to some parts of the sequence than others. Maybe one day it’ll make sense to just start a generic neural net and do all customization through training. But at least <b>as of now it seems to be critical in practice to “modularize” things—as transformers do</b>, and probably as our brains also do.</bq> <bq>First, it takes the sequence of tokens that corresponds to the text so far, and finds an embedding (i.e. an array of numbers) that represents these. Then it operates on this embedding—in a “standard neural net way”, with values “rippling through” successive layers in a network—to produce a new embedding (i.e. a new array of numbers). <b>It then takes the last part of this array and generates from it an array of about 50,000 values that turn into probabilities for different possible next tokens.</b></bq> <bq>[...] it’s part of the lore of neural nets that—in some sense—<b>so long as the setup one has is “roughly right” it’s usually possible to home in on details just by doing sufficient training</b>, without ever really needing to “understand at an engineering level” quite how the neural net has ended up configuring itself.</bq> <bq><b>Within each such attention block there are a collection of “attention heads” (12 for GPT-2, 96 for ChatGPT’s GPT-3)—each of which operates independently on different chunks of values in the embedding vector.</b> (And, yes, we don’t know any particular reason why it’s a good idea to split up the embedding vector, or what the different parts of it “mean”; this is just one of those things that’s been “found to work”.)</bq> <bq>What the “attention” mechanism in transformers does is to allow “attention to” even much earlier words—thus <b>potentially capturing the way, say, verbs can refer to nouns that appear many words before them in a sentence.</b></bq> <bq><b>Essentially it’s to transform the original collection of embeddings for the sequence of tokens to a final collection.</b> And the particular way ChatGPT works is then to pick up the last embedding in this collection, and “decode” it to produce a list of probabilities for what token should come next.</bq> <bq>[...] in the end what we’re dealing with is just a neural net made of “artificial neurons”, <b>each doing the simple operation of taking a collection of numerical inputs, and then combining them with certain weights.</b></bq> <bq>The original input to ChatGPT is an array of numbers (the embedding vectors for the tokens so far), and what happens when ChatGPT “runs” to produce a new token is just that these numbers “ripple through” the layers of the neural net, with each neuron “doing its thing” and passing the result to neurons on the next layer. <b>There’s no looping or “going back”. Everything just “feeds forward” through the network.</b></bq> <bq>[...] there are millions of neurons—with a total of 175 billion connections and therefore 175 billion weights. And <b>one thing to realize is that every time ChatGPT generates a new token, it has to do a calculation involving every single one of these weights.</b> Implementationally these calculations can be somewhat organized “by layer” into highly parallel array operations that can be conveniently be done on GPUs. But <b>for each token that’s produced, there still have to be 175 billion calculations done (and in the end a bit more)—so that, yes, it’s not surprising that it can take a while to generate a long piece of text with ChatGPT.</b></bq> <bq>[...] the remarkable thing is that all these operations—individually as simple as they are—can somehow together manage to do such a good “human-like” job of generating text. It has to be emphasized again that (at least so far as we know) there’s no “ultimate theoretical reason” why anything like this should work. And in fact, as we’ll discuss, I think we have to view this as a—potentially surprising—scientific discovery: that <b>somehow in a neural net like ChatGPT’s it’s possible to capture the essence of what human brains manage to do in generating language.</b></bq> <bq>[...] it’s certainly not that somehow “inside ChatGPT” all that text from the web and books and so on is “directly stored”. Because <b>what’s actually inside ChatGPT are a bunch of numbers—with a bit less than 10 digits of precision—that are some kind of distributed encoding of the aggregate structure of all that text.</b></bq> <bq>[...] while the results from this may often seem reasonable, they tend—particularly for longer pieces of text—to “wander off” in often rather non-human-like ways. <b>It’s not something one can readily detect, say, by doing traditional statistics on the text. But it’s something that actual humans reading the text easily notice.</b></bq> <bq>[...] how can the neural net use that feedback? The first step is just to have humans rate results from the neural net. But then another neural net model is built that attempts to predict those ratings. But now this prediction model can be run—essentially like a loss function—on the original network, <b>in effect allowing that network to be “tuned up” by the human feedback that’s been given. And the results in practice seem to have a big effect on the success of the system in producing “human-like” output.</b></bq> <bq>[...] try to give it rules for <b>an actual “deep” computation that involves many potentially computationally irreducible steps and it just won’t work.</b> (Remember that at each step it’s always just “feeding data forward” in its network; never looping except by virtue of generating new tokens.)</bq> <bq>[...] when ChatGPT continues a piece of text this corresponds to tracing out a trajectory in linguistic feature space. But now we can ask what makes this trajectory correspond to text we consider meaningful. And <b>might there perhaps be some kind of “semantic laws of motion” that define—or at least constrain—how points in linguistic feature space can move around while preserving “meaningfulness”?</b></bq> <bq>[...] yes, this seems like a mess—and doesn’t do anything to particularly encourage the idea that one can expect to identify “mathematical-physics-like” “semantic laws of motion” by empirically studying “what ChatGPT is doing inside”. But <b>perhaps we’re just looking at the “wrong variables” (or wrong coordinate system) and if only we looked at the right one, we’d immediately see that ChatGPT is doing something “mathematical-physics-simple” like following geodesics.</b></bq> <bq>When we start talking about “semantic grammar” we’re soon led to ask “What’s underneath it?” What “model of the world” is it assuming? A syntactic grammar is really just about the construction of language from words. But <b>a semantic grammar necessarily engages with some kind of “model of the world”—something that serves as a “skeleton” on top of which language made from actual words can be layered.</b></bq> <bq>From its training ChatGPT has effectively “pieced together” a certain (rather impressive) quantity of what amounts to semantic grammar. But its very success gives us a reason to think that it’s going to be feasible to construct something more complete in computational language form. And, <b>unlike what we’ve so far figured out about the innards of ChatGPT, we can expect to design the computational language so that it’s readily understandable to humans.</b></bq> <bq>[...] the original basic constructs of syllogistic logic could now be used to build huge “formal towers” that include, for example, the operation of modern digital circuitry. And so, we can expect, it will be with more general semantic grammar. At first, it may just be able to deal with simple patterns, expressed, say, as text. But <b>once its whole computational language framework is built, we can expect that it will be able to be used to erect tall towers of “generalized semantic logic”, that allow us to work in a precise and formal way with all sorts of things that have never been accessible to us before</b>, except just at a “ground-floor level” through human language, with all its vagueness.</bq> <bq>[...] unlike even in typical algorithmic computation, <b>ChatGPT doesn’t internally “have loops” or “recompute on data”. And that inevitably limits its computational capability—even with respect to current computers, but definitely with respect to the brain.</b> It’s not clear how to “fix that” and still maintain the ability to train the system with reasonable efficiency. But to do so will presumably allow a future ChatGPT to do even more “brain-like things”. Of course, <b>there are plenty of things that brains don’t do so well — particularly involving what amount to irreducible computations. And for these both brains and things like ChatGPT have to seek “outside tools”—like Wolfram Language.</b></bq> <hr> <a href="https://blog.plover.com/2023/02/22#four-digit-numbers" author="Mark Dominus" source="The Universe of Discourse">ChatGPT on the subject of four-digit numbers</a> <bq><bq><b>MJD</b> What is the largest number that is less than 1000? <b>ChatGPT</b> The largest number that is less than 1000 is 999. Any number that is less than 1000 will have three digits, with the first digit being 9, the second digit being 9, and the third digit being any number from 0 to 8. The largest possible three-digit number that is less than 1000 is therefore 999. This is the largest number that is less than 1000, as any number with four or more digits will be greater than 1000.</bq>I was impressed by this, the most impressed I had been by any answer it had given. It had answered my question correctly, and although it should have quit while it was ahead the stuff it followed up with wasn't <i>completely</i> wrong, only somewhat wrong. But it had made a couple of small errors which I wanted to probe.</bq> In my continuing series of meta-commenting about people's comments about ChatGPT, I continue to be horrified by what people consider to be "good" or "impressive". I think it's great that this supposed AI can write whole sentences that are reasonably grammatically correct. What it's writing about is a misleading soup that is more than just a little wrong. It is riddled with errors declared authoritatively. These wrong answers are not just coming from strangers on the Internet now, but from sources that a lazy society has historically promoted to <i>unimpeachable</i> status (i.e., "the system says you're not registered" will become "1000 is not a four-digit number"). From GPTChat's paragraph above, <ol> ✅ <iq>The largest number that is less than 1000 is 999</iq> ⛔️ <iq>Any number that is less than 1000 will have three digits</iq> ⛔️ <iq>with the first digit being 9</iq> ⛔️ <iq>the second digit being 9</iq> ⛔️ <iq>and the third digit being any number from 0 to 8</iq> ⛔️ <iq>The largest possible three-digit number that is less than 1000 is therefore 999.</iq> (I'm marking this wrong because the only purpose of the sentence is "therefore", which has no place in that paragraph because the preceding sentences didn't provide any chain of logic that proved anything.) ⛔️ <iq>any number with four or more digits will be greater than 1000.</iq> </ol> I suppose the author's original statement <iq>it should have quit while it was ahead</iq> was a fair summary. The author followed up with, <bq><bq><b>MJD</b> That is a good answer, but you said that any number with four or more digits will be greater than 1000. That is not true. Can you think of a counterexample? <b>ChatGPT</b> You are correct, my previous statement was not accurate. Any number with four or more digits can be less than 1000, depending on the specific digits that are used. For example, the number 9991 is a four-digit number that is less than 1000.</bq>This reminds me of Richard Feynman's story about reviewing science textbooks for the State of California. He would be reading the science text book, and it would say something a little bit wrong, then something else a little bit wrong, and then suddenly there would be an <b>enormous pants-torn-off blunder that made it obvious that the writers of the book had absolutely no idea what science was or how it worked.</b></bq> <h><span id="programming">Programming</span></h> <a href="https://people.kernel.org/linusw/rust-in-perspective" source="linusw" author="linusw">Rust in Perspective</a> <bq><b>C provided freedom</b>, where high-level languages were considered as straight-jackets enforcing unwanted discipline. It was <b>an invitation to use tricks which had been necessary to achieve efficiency in the early days of computers.</b></bq> <bq>Alonzo Church was a scholar of mathematical logic and computability, the supervisor of Alan Turing's doctoral thesis and active in the same field as Kurt Gödel (those men quoted each other in their respective articles). <b>The lambda calculus ties into the type set theory created by Bertrand Russell and the logical-mathematical programme</b>, another universe of history we will not discuss here.</bq> <bq>While regular expressions can express how to parse a body of text in a language with regular grammar, <b>expressions in λ-calculus can go on from the abstract syntax tree and express what an addition is, what a subtraction is, or what a bitwise OR is.</b> This exercise is seldomly done in e.g. compiler construction courses, but <b>defining semantics is an inherent part of a programming language definition.</b></bq> <bq>ML still has one imperative language feature: assignment. Around this time, some scholars thought both the J operator and assignment were unnecessary and went on to define purely functional languages such as Haskell. We will not consider them here, they are outside the scope of this article. <b>ML and everything else we discuss can be labelled as impure: a pejorative term invented by people who like purely functional languages.</b> These people dislike not only the sequencing nature of imperative languages but also the assignment (such as happens with the keyword let ) and <b>prefer to think about evaluating relationships between abstract entities.</b></bq> <bq>But make no mistake. The current underlying ambition is definitely nothing different from the ambition of the ALGOL committee between 1958 and 1968: <b>to raise the abstraction of the language through the ambition to join computer programming with formal logic.</b> This comes from the arrival of strong academic support for the language.</bq> I especially liked the notion that bringing more formal-verification mechanisms into the language brings us closer to making TDD unnecessary because, once it compiles, it runs. While this is powerful, I take this prognostication with a bit of skepticism because it strongly prefers verification of low-level semantics (e.g., no dangling references or referencing of freed memory, no race conditions, no deadlocks) while kind of ignoring the high-level semantics. I.e. the program is perfect and will never crash, but <i>does it do want you want it to do</i>? Unless I've missed something, there is nothing in Rust that prevents you from writing a program that subtracts two numbers when you really meant to add them. Trying to formally verify such high-level semantics ends up in a <i><a href="https://en.wikipedia.org/wiki/Quis_custodiet_ipsos_custodes%3F">Quis custodiet ipsos custodes</a></i>? loop, where you'd have a program that claims to add two numbers, and a meta-program that claims to verify that it adds two numbers, and a ... meta-meta-program to verify that the prover is actually doing what it says... and so on. But not to distract: great article and a great language that should be a boon to stability in Linux (even if I have my doubts that we'll be able to replace all of the C with Rust, simply because Rust does not allow the kind of "cheating" that can be incredibly efficient, performance-wise). <hr> <a href="https://goomics.net/361/" author="" source="Goomics: Comics about Life at Google">2021-01-22: Fix a typo in a comment</a> <img src="{att_link}2021-01-22_lgtm.jpg" href="{att_link}2021-01-22_lgtm.jpg" align="none" caption="2021-01-22: Fix a typo in a class comment" scale="50%"> Look, I get that the red tape that this bug-submission form suggests feels like way too much. It is, of course. It's exaggerated. But laughter at it is hollow when you start to think about how much is "just enough". <ul> Who gets to fix typos? What qualifications do you have to have? Is it really an obvious typo? Is it a translation error? Do we need to check the original, too? Has this already been changed to that version for other reasons? PC reasons, perhaps? Are there ways of misinterpreting the correction? Does the correction fit with the style? Which style of English is accepted? Who can actually see this change? Is the source published publicly? </ul> Sure, maybe all of these questions are answered easily and implicitly when you're just trying to fix "synchronise" to "synchronize", but you can't deny that these questions are relevant for the class of repair that is "fix a typo". <h><span id="fun">Fun</span></h> <a href="https://www.smbc-comics.com/comic/engaged" author="Zack Weinersmith" source="SMBC">Engagment</a> <img src="{att_link}smbc_engaged.jpg" href="{att_link}smbc_engaged.jpg" align="none" caption="SMBC: engaged" scale="75%"> <bq>I already spend 4 hours a day on this app. It feeds me an infinite quantity of short videos. I spend an average of 4 seconds on each; like or dislike, share or don't share, then move on Do i like it? I don't know! Eventually I run out of Internet. But it won't be long until AI can generate content designed to surgically pull apart every bit of my attention span until all that remains is 150 pounds of warm flesh existing only to minimally service its bodily functions while looking at the next video</bq> <hr> <a href="https://www.smbc-comics.com/comic/novels" author="Zack Weinersmith" source="SMBC">Novels</a> <img src="{att_link}smbc_novels.jpg" href="{att_link}smbc_novels.jpg" align="none" caption="SMBC: Novels" scale="75%"> <bq>Jane Austen Novels: ...and now we introduce character number 412. son of the aunt whose sister-in-law's cousin was briefly affianced to the troubled love interest of the protagonist's estranged brother's half-nephew's wife's best friend.</bq> Hover-over title: <bq>I feel like all these people trying to measure attention span degradation should just look at what percent of the population is cognitively capable of reading Sense and Sensibility.</bq>