Astera wants to fund ~10 “icebox experiments… for start-ups that want to publish their science as/after they ramp down” (@seemaychou, application form here)
RIP, Charlie Munger
Munger giving the USC Law Commencement 2007 (via @garrytan):
The last idea that I want to give you, as you go out into a profession that frequently puts a lot of procedure, and a lot of precautions, and a lot of mumbo-jumbo into what it does, this is not the highest form which civilization can reach.
The highest form that civilization can reach is a seamless web of deserved trust. Not much procedure, just totally reliable people correctly trusting one another.
That’s the way an operating room works at the Mayo Clinic. If a bunch of lawyers were to introduce a lot of process, the patients would all die.
So never forget, when you’re a lawyer, that you may be rewarded for selling this stuff, but you don’t have to buy it. In your own life, what you want is a seamless web of deserved trust. And if your proposed marriage contract has forty-seven pages, my suggestion is you not enter.
Pika 1.0, an idea-to-video platform (via @pika_labs). We are at the dawn of a golden age of creativity: 100x as many people will be able to make their vision into reality, new blockbusters will come out of nowhere, and everyone will benefit. (And we will escape the endless sequels and franchise reboots.) “When digital filmmaking became viable in the 90s, I expected a boom in independent cinema that never materialized…. AI may finally deliver on this promise, as one creative person can prompt videos that previously would take the collaboration of hundreds of people to capture. The next 10 years of cinema could be weird and wonderful” (@ryrzny)
Future House announces WikiCrow, which “generates Wikipedia style articles on scientific topics by summarizing full-text scientific articles” (via @SGRodriques). “As a demo, we are releasing a cited technical summary for every gene in the human genome that previously lacked a Wikipedia article: 15,616 in all.” They are hiring, see link above
Energy/climate announcements
On Dec 2, “representatives from 20 countries signed a declaration establishing a goal of TRIPLING global nuclear energy production by 2050” (@Atomicrod)
British Library hit by ransomware: “The level of cultural vandalism here is truly shocking…. Like the digital version of the burning of the Library of Alexandria,” says (@antonhowes)
Michael Nielsen on “wise accelerationism”. “Will a sufficiently deep understanding of nature make it almost trivial for an individual to make catastrophic, ungovernable technologies?” If so, “what institutional ideas or governance mechanisms can we use to stay out of this dangerous realm?”
“We need some serious people who are also builders and not decelerationists. These issues are too crucial to be left to vibes-based shitposters” (me). And: “Like Jason, I like e/acc vibes, but agree they are not enough. Build the future with evidence, careful thought, well-roundedness, courage, earnestness, creativity, strength, love of life, curiosity, vulnerability, nuance, hard work, and a can-do spirit” (@elidourado)
“I’m neither an accelerationist nor a decel when it comes to AI / AGI. This is technology we absolutely must build, and we’ll build it regardless. But I hope thoughtfully, not carelessly. Reducing a debate to a memetic tribal saying to tweet or add to bio is fun—but not a strategy” (@rsg). Related: “Threading the needle between being an AI panglossian and being an AI doomer is hard sometimes. Social dynamics often force people to identify with a simple ideological camp, but reality doesn’t always fit neatly into a single camp” (@MatthewJBar)
“One difference between the EA and progress studies communities: ‘Progress is inevitable and maybe even happening too fast, and maximum efforts against it can only slow it by a bit (which might not be enough)’ vs. ‘Progress can be stopped, and without active efforts to protect it, it will probably happen too slowly’” (me)
“A personal take on how Sam’s done as CEO with respect to AI safety. Far from perfect. But way better than most alternatives and a huge margin better than the next-most-likely outcomes” (thread from @jachiam0). I particularly appreciated this point: “When people ask me what safety is, my answer is always something along the lines of: ‘safety is a negotiation between stakeholders about acceptable risk.’ All parts of this are important. Safety is not a state; it’s a conversation, a negotiation”
“I take AI safety pretty seriously, but it really bothers me that some in the AI safety world don’t take the upsides of AGI seriously. Yes, we should try to avoid an AGI that wipes out humanity, but an AGI that could end death isn’t something to give up trying to develop ASAP” (@s8mb)
“There will almost certainly come a point (p=.999?) where an AI is developed that a) poses no threat to humanity civilization and b) is able to perform enough economically meaningful work to cause some humans to lose their jobs. At that point, there will be bootleggers and Baptists situation. Make-workers = bootleggers; Safetyists = Baptists. We have few social antibodies against this kind of alliance. I continue to worry we will never reap the economic benefits of AI. We need: decentralized and open source AI; freedom to compute; preemptive attacks on make-workism; a norm against restrictive forms of safetyism unless there is some kind of empirical evidence that danger is imminent” (@elidourado)
More thoughts from me
If no one is criticizing you, it’s not because your work is unimpeachable: it’s because you’re not doing/saying anything important, or you don’t have a significant audience. Any important work that gets attention will draw criticism (Threads, Twitter)
The tragedy is that bioethics ought to be a serious, interesting field. There are big, important questions that deserve deep thought. But bioethicists have made a mockery of the field by extreme hand-wringing and waffling over very easy questions (Twitter)
Episode 7 of Age of Miracles: “What will the energy mix look like in 2050? We’re not nuclear maxis. We’re energy abundance and human flourishing maxis. Getting there is going to take a mix of nuclear, solar, wind (?), geothermal, hydro, fusion, and… fossil fuels. Yes, fossil fuels.” (@packyM). And episode 8 is focused on fusion (also via @packyM)
“Global poverty is commonly measured by counting the number of people whose consumption falls below a given threshold. This approach overlooks an enormous component of people’s economic well-being: public goods” (thread from @amorygethin based on this working paper)
“Who would want to work on [a startup to build open-source parametric CAD software] with @Spec__Tech?” (@Ben_Reinhardt)
“People who work in biotech: Would you pay for access to a new database of clinical-stage failures by target, and the reasons for failure (incl. metadata on phase, program milestones and dates, company, etc.)?” (ROP fellow @Atelfo)
“Is anyone working on an AI tool to do basic quality checks on scientific papers? Check for prereg; cross-check prereg outcomes with paper; check for appropriate power calc; check for appropriate stats; basic fraud checks” (@sguyenet)
“What’s the best answer to why more good universities aren’t being founded in the last 50 years? (e.g. the 1890s alone had Stanford, UChicago, & Caltech.)” (@nabeelqu) (My comment: “FWIW 1890s was a special time. US badly needed top tier research universities and did not have any. Europe outclassed them especially Germany. So there was an urgent need that we don’t have now”)
On not romanticizing the past
“We are so impossibly wealthy in the modern era” (@kendrictonn, commenting on the nail manufacturing link below). “All these things that used to be expensive enough to notice and care about. It’s dizzying when you start to think about it.” Good thread, read the whole thing
“In the Middle Ages ordinary people in Europe could not afford plates and the practice was to eat off yesterday’s stale bread, called a trencher. By the 14th century most people used wooden trenchers, and by the 17th century pewter (dangerous!) or earthenware plates” (@DavidDeutschOxf)
“This Thanksgiving I am particularly thankful for the existence of centralised heating and hot water at one’s fingertips. Our hot water was temporarily discontinued due to a leakage, and it’s the best time to remember how good we have it compared to our ancestors!” (@RuxandraTeslo)
“Project Lyra explores the possibility of sending a spacecraft to chase ‘Oumuamua using an Earth gravity assist to reach Jupiter, which robs the craft of almost all its speed, causing it to drop to the Sun for an Oberth maneuver” (@tony873004, check out the trajectory visualization)
“Let’s say you have a humanoid robot with AGI. How much power does the robot need on board to do continuous inference? What is a good power source that has appropriate density and portability?” (thread from @elidourado). See also related thread from Eli on “the energy requirements of an ‘intelligence explosion’ scenario,” with replies from @anderssandberg where he points to this paper
“Even though it feels like there are a million health metrics you can track through wearables etc, health is actually UNDER tracked. The average machine/software/etc has so many more available stats” (@auren quoting Sam Corcos)
“I think I’m changing my mind about prediction markets. Whether it’s the House Speaker drama, or the OpenAI story, I’ve noticed they’re definitely becoming part of my news consumption” (@TheStalwart)
Austin City Council votes to legalize 3 homes per lot. “Incredibly proud of the work our AURA volunteers and partners put in over the past six months to make this win possible. Austin is taking meaningful steps toward housing abundance” (ROP fellow @RyanPuzycki)
“The OpenAI governance crisis highlights the fragility of voluntary EA-motivated governance schemes. So the world should not rely on such governance working as intended,” says Jaan Tallinn to Semafor (via @semafor). Semafor describes this as “‘questioning the merits’ of running companies based on effective altruism,” although others have pointed out that it might instead be questioning the reliabliity of self-regulation
“In 1989 Wang Huning envied the daring he found in America. He celebrated a culture willing to engineer its way out of any problem, its drive to slowly build towards utopia.” But now, “on almost every metric Wang discusses for his ‘spirit of innovation’ and ‘spirit of the future’ the Chinese government and society seems to embody these things more than America does” (thread from @Scholars_Stage, summarizing this essay)
“TIL that there are countries where ~10% of the population applies every year for the US visa lottery… Crazy to think that if the US opened its borders, 10% of a country would pack their bags and leave” (@sebasbensu)
“It took a very long time to get here, and we had to change dozens of laws but now you can just go for a 17-story tower of housing in Palo Alto” (@kimmaicutler)
“Remember folks: if no one in your organization was around the last time something was done, you don’t have institutional experience, just institutional records” (@culpable_mink)
“Every … invention mocked in the same cartoon came to pass within one hundred years, including the railroad, the airship, the airplane, the automobile, the powerboat, and the industrial factory.” (from “Technological Stagnation Is a Choice”, via @balajis and @packyM)
The community of interests, the unity that exists between men, stems first and foremost from scientific progress, from discoveries, inventions, and the proliferation of new machines.
Thomas Edison, quoted by Matt Ridley (via @jonboguth):
I am ashamed at the number of things around my house and shops that are done by animals—human beings, I mean—and ought to be done by a motor without any sense of fatigue or pain. Hereafter a motor must do all the chores.
… from the beginning America’s religion has been the religion of work. In other societies, heredity, class, social status, ethnicity, and family are the principal sources of status and legitimacy. In America, work is. In different ways both aristocratic and socialist societies tend to demean and discourage work. Bourgeois societies promote work. America, the quintessential bourgeois society, glorifies work. When asked “What do you do?” almost no American dares answer “Nothing.”
The increases are not caused directly by the regulations. Rather, they result from interaction of NRC and industry staffs in an effort to obtain the goals of “zero risk and “zero defects” in an adversarial and legalistic regulatory environment. The pursuit of “zero risk” means that all failure modes have the same importance regardless of their probability of occurrence. The pursuit of “zero defects” encourages continuous expansion of the number of alternatives that must be analyzed and continuous refinements of the requirements. The legal need for “evidence” encourages development of complex analyses rather than simple ones. It also submerges the fact that all engineering analyses are really approximations to actual phenomena because such subjective pos-itions are difficult to defend in court.
The advantage which “precedent” has in the legal environment encourages the adoption of unrealistic and expensive approaches because it is easier than suffering the uncertainty associated with trying to have a new approach accepted. The continuation of this situation for a decade has resulted in the tradition that if a method is more complex, more difficult and more time consuming, it must be safer! The development over the last decade of seismic design criteria and design approaches with respect to piping systems is an example of the situation.
The classic division of authority and responsibilities between engineer, manufacturer and constructor has been distorted by the fear of losing a decision in a regulatory or judicial hearing which would cause political damage to the regulatory agency or financial damage to the utility. Consequently, there has been an ever increasing reliance on academic analytical techniques rather than experienced judgment. The concept of “good engineering practice” is discouraged during construction. Analysts and designers on both regulatory and industrial staffs inexperienced in the realities of hardware and construction practices are making impractical design decisions which have been exceedingly damaging to nuclear industry economics.
Ironically, some of the decisions that have been made in the name of improving safety margins for low probability events, may have reduced the safety margins for high probability events. The UE&C piping study uncovered numerous areas where the tolerances requested in the piping design documents might be appropriate for a machine shop oriented manufacturing operation, but are totally unrealistic for field construction.
Maps and charts
“Mechanization, the harnessing of energy, is man’s answer to slavery” (Buckminster Fuller). Map from 1940, when the world had just over 2 billion people, and machines doing the energy equivalent of 37 billion people’s worth of work for them (Twitter, Threads):
“An incredible 2001 paper re-analysed every then-available dataset on historical European homicide rates. This turned up an amazing trove of estimates. Firstly, England. Note that it’s a log scale – 1300s England was a war zone!” (via @bswud, see the whole thread for more):
“SpaceX is tracking to launch over 80% of all Earth payload to orbit this year” (@elonmusk)
“5th December 1952 marked the start of the horrific Great Smog of London. The smog triggered the process that led to the Clean Air Act in 1956” (@jimmcquaid)
The “journey of the miraculous GLP1 weight-loss drugs: 120 years ago: first basic discoveries in gut hormones; 37 yrs ago: GLP1 identification; 25 yrs ago: discovery that GLP1 decreases appetite” (via @DKThomp)
“Here’s how many new US federal laws were applicable to nuclear power plants from 1954 to 1984” (@whatisnuclear)
“The US increasingly leading in oil production” (@StefanFSchubert)
“Nvidia datacenter revenue, by @Thomas_Woodside” (@paulg):
Job openings
Prizes
Other opportunities
RIP, Charlie Munger
Munger giving the USC Law Commencement 2007 (via @garrytan):
I heard the news from @_TamaraWinter; coincidentally, Stripe Press just republished Poor Charlie’s Almanack.
Bio announcements
AI announcements
Bio ∩ AI announcements
Energy/climate announcements
UK news
Risk, safety, and tech
More thoughts from me
Podcasts
Video
Papers
Other articles and links
Queries
Click through and answer if you can help:
On not romanticizing the past
More social media
Politics and culture
Quotes
Maria Montessori (via @mbateman):
Thomas Edison, quoted by Matt Ridley (via @jonboguth):
Samuel Huntington, Who Are We? The Challenges to America’s National Identity (h/t @SarahTheHaider)
On nuclear construction cost increases, from Crowley and Griffith 1982, “US construction cost rise threatens nuclear option” (via @whatisnuclear):
Maps and charts
“Mechanization, the harnessing of energy, is man’s answer to slavery” (Buckminster Fuller). Map from 1940, when the world had just over 2 billion people, and machines doing the energy equivalent of 37 billion people’s worth of work for them (Twitter, Threads):
“An incredible 2001 paper re-analysed every then-available dataset on historical European homicide rates. This turned up an amazing trove of estimates. Firstly, England. Note that it’s a log scale – 1300s England was a war zone!” (via @bswud, see the whole thread for more):
“SpaceX is tracking to launch over 80% of all Earth payload to orbit this year” (@elonmusk)
“5th December 1952 marked the start of the horrific Great Smog of London. The smog triggered the process that led to the Clean Air Act in 1956” (@jimmcquaid)
The “journey of the miraculous GLP1 weight-loss drugs: 120 years ago: first basic discoveries in gut hormones; 37 yrs ago: GLP1 identification; 25 yrs ago: discovery that GLP1 decreases appetite” (via @DKThomp)
“Here’s how many new US federal laws were applicable to nuclear power plants from 1954 to 1984” (@whatisnuclear)
“The US increasingly leading in oil production” (@StefanFSchubert)
“Nvidia datacenter revenue, by @Thomas_Woodside” (@paulg):