When I first read online about "finding the best ways to spend your time and money”, I thought that these were good ideas. I also immediately thought I should tell all my friends (of which I had… none. sorry little Ada.) If only you knew how bad the world is out there, why wouldn't you give to the causes you’re told would stop it? I tried to present the argument at my high school’s after-school philosophy club. No one bought this (I had no theory of mind).

This year at university, I thought I’d learnt from my mistakes. I decided there was not enough planning involved in my presentation, so I made plans. I thought the reason people weren't convinced was that they found the presentation suspicious. So me and some new compatriots I met in the spring made plans that October, and took to the activities fair. We made signs and prepared decorations and spent some money to print flyers. We had mosquito nets and books on display.

We did not get anyone really interested, over a whole day of tabling. (I guess I still didn’t have a good theory of mind).

If there was one thing I would tell my past self regarding community building for EA, it’s that generally people cannot be reasoned with. They are not a coherent utility function - they are made up of linear piecewise approximations that change based on where they find themselves in a given moment. There is no consistency in sight. Even where there is an overall gestalt, it's made of proxy-metrics. Grades; financial stability; family approval; clubbing, sports and hot people; old habits; prestige. The vast majority of humans in the recruitment pool of a potential student society are not aligned to anything in any meaningful sense.

Culture entrepreneurship is difficult. Entrepreneurship itself is difficult, but at least there are playbooks. Let me know if there are any of a similar calibre written for culture, without the financial incentives. Maybe political playbooks?

EA is a “social technology”. It has helped me make friends with people who are very kind, have moral integrity, and good at machine learning. Incidentally they have also made a bunch of progress on the welfare of the most neglected beings on earth. However, the movement can be super siloed. And I get it. The siloing is a function of the ideas itself - “Keep EA Weird!” - and the type of people who stick around in EA. It's hard to ask them to have a background and know the relevant people in disparate fields, many of whom are not like them. The minimal form of EA fills a niche only present in a very small number of people - those versed deeply enough in the analytical tradition of philosophy to understand numbers and weep, and those who have the resources to do something about it.

KINETICS

I don't know how to tell whether someone is that kind of people. Try the pitch on them and see if it sticks? But there is no meaningful selection pressure for rationality independent of answering questions in standardized tests, and those selected to be academically rational will not often extend that to any other part of their life. There’s no appeal, except to ask people to extend the trust that we have really tried being honest at doing the best we can for the world and working from the ground up, and that being dispassionate about the analytics before being passionate about the causes is a method which really, actually works.

Maybe with a model of what ideas (associations) exist in each person’s head, and the valence associated with the hooks of the EA pitch, I could make a better estimate of how to do this kind of outreach?

THERMODYNAMICS

According to survey results, the total addressable psychological market of the Effective Altruism movement is ~6% of the university population1. Imperial has 20,000 students, multiplied by 6% is 1200. How reachable they are depends entirely on the shape of the social graphs that form in the first week. If this 6% are far away from each other, there is no word of mouth. Imagine an application process that filters 2.5x for proto-EAs. That means an initial reading group cohort with 15% * 7 people which is ~ONE proto-effective-altruist: really on the edge between worth and not worth running. And that's if everything goes right, and not counting ways this psychological proto-EA might be held back by frictions in the actual world (family, culture, finances) from being able to get into EA to the extent they might want.

ENDING ONE

Entrepreneurship can be made easier if you have a great product that sells; the product and the relationship between you and your club members can be messier to define in community building. Each market is different and we don’t want to be outcompeted by other clubs that are substitutes to giving student what they want. The first step to knowing what they want is knowing them. The second step is knowing the local menu of options they are offered. Without good intuition, both of these steps takes repeated sampling and verification.

Not everyone is built for this. Like a minority but significant part of the EA demographic, I am not naturally built for this.

However, people can change a LOT with time. I look forward to keeping on attempting this challenge. At steady state, there exists a solution where an alive and contributing “EA-adjacent" Imperial society exists. The only thing now to work out is the kinetics of how to get there.

So, I guess visit us at EA Imperial if you like to do good, or to be right, and especially both. EA is about using underrated facts about the world and pushing you into high impact, fulfilling careers. If you were a math competition kid, or care about animals, or want to do engineering that changes the world - we are your people.

…is what I would have said.

ENDING TWO

I look down at my hands. I read the above section again.

“But do I really believe this, and want to put it into everybody’s inboxes, telling everyone this is what I believe?

I’m a kid who’s just started living this last year. ‘Not being the best for this role’ is an understatement. I can feel there’s more basic things, which successful socialising build off of, which I don’t quite know how to do.”

I think about a long term plan of what I want my life to look like. I want to have cosy gatherings. I want to have films and homemade food, I want to look forward to events rather than having a slight distaste at the upcoming one in a sterile venue probably not going well each time in the forseeable future. I have limited willpower, and there’s gotta be a better way than that.

i do think it’s important to maybe...graduate ea-student-group-centrism at some point?

and to that end, i wonder if

speedrunning it

is a thing one can do?

like what speedrunning it might look like

The friend is right: among a lot of things not working, the broad interest space, principles-first thing is one of them. Unknown principles are a bit too weak to bring busy people together. When we had one board game night discuss AI, a specific topic, that might have been the most coherent gathering we had, and I still didn’t have that much fun. Instead, we probably want a skill to work on together, some utility to provide through practice at that skill, that at least some people will want to make time to gather for each week.

“general nonfic substack writing group / prediction trading group might work; doing research for policy proposals?”

hair-pully but someone's got to do it lol

Which I guess is at least a calmer, more stable place to start.

Thanks to Lydia for being able to measure the final ideas off of, Ivet for inspiring this essay, Celeste for demonstrating the feasibility of writing on the train.

No Silver Bullet - Jason Crawford

So fire safety was achieved through the combination of:

• General-purpose technologies: engines, electronic communications, electric light and heat

• Specific inventions: fire pumps, fire hose, fire alarms

• Infrastructure: municipal water supply, telephone lines

• Standards, testing and certification: of electrical products, fire preventing and fire-fighting equipment, building materials, etc.

• Law: building codes and other fire safety codes

• Education and training: in fire departments, among the public

This is a general pattern. Safety requires:

• both prevention and “cure”

• both technical and social solutions

• among technical solutions, both products and systems

• among social solutions, both education and law

So too should this apply to air safety; road safety; biosafety. Everything robust is a defense-in-depth. If you have a policy goal of “preventing bad thing X from happening”, you will want to add layers of protection in different ways that don’t fail together. Social, cultural, legal, technological.

Status as a Service (StaaS) - Eugene Wei

Community building for EA has been on my mind. How networks form and grow is the core problem you try to solve if you do CB. From past reading (and past experience), I tend to look at belonging to a community as a conveyer of status - I’ve been on the grind to get cultural capital this year, and convert it into social capital, so that I can finally get things like a good job, a girlfriend, etc. and feel like I deserve it. And it feels natural to me that yeah, I wasn’t told this, but this is pretty close to how it works.

This piece was written in 2019 so it’s probably influenced, through other writers, how I ended up thinking. Looking backwards in time, the ideas were approached from an interesting cosine wrt. where I’d approach it from now. Quotes:

Social capital is, in many ways, a leading indicator of financial capital, and so its nature bears greater scrutiny. Not only is it good investment or business practice, but analyzing social capital dynamics can help to explain all sorts of online behavior that would otherwise seem irrational.

In the past few years, much progress has been made analyzing Software as a Service (SaaS) businesses. Not as much has been made on social networks. Analysis of social networks still strikes me as being like economic growth theory long before Paul Romer’s paper on endogenous technological change. However, we can start to demystify social networks if we also think of them as SaaS businesses, but instead of software, they provide status.

Can I use the social network to accumulate social capital? What forms? How is it measured? And how do I earn that status?

On starting new networks:

This is the classic cold start problem of social. The answer to the traditional chicken-and-egg question is actually answerable: what comes first is a single chicken, and then another chicken, and then another chicken, and so on. The harder version of the question is why the first chicken came and stayed when no other chickens were around, and why the others followed.

Students at top universities pack their days, and prioritise and exploit well, and don’t have free time every week to hang out with new people they don’t know well. I was thinking before reading this that an EA social club offers a way to get better at some (broader) competition - of getting into trading, or ML research jobs, prestige by proximity to an affluent bay tech / academic Oxfordian culture, or of feeling you are a good and rational person. But you can’t compete with Algotrading Society for the first-line attention of people who want to algorithmically trade.

So, to answer an earlier question about how a new social network takes hold, let’s add this: a new Status as a Service business must devise some proof of work that depends on some actual skill to differentiate among users. If it does, then it creates, like an ICO [Initial Coin Offering], some new form of social capital currency of value to those users.

This is not the only way a social network can achieve success. As noted before, you can build a network based around utility or entertainment. However, the addition of status helps us to explain why some networks which seemingly offer little in the way of meaningful utility (is a service that forces you to make only a six second video useful?)[(Is a service that asks you to help people far away, or animals, or both who do not even currently exist)] still achieve traction.

However, broad-based skills I think people can work on together (that EA could be a good RL environment for) are those of writing - including reading and some of the situational awareness that comes with it; betting and making money on your bets; and desk research habits on selected issues. Notably, this may not include mechanistic interperability, engineering research, or biological lab research - few people may ever come in at once with the ability to do these, and the r-value may be close enough to <1 to make it difficult to cold-start this competition without a magically large base population and high proportion selected who can do these things. A possible solution is to make friends among people who can do CS, or wet-lab (iGEM), and are passionate enough to know other people also into their thing (have a tight enough social network from which to pull enough friends to cold-start a positive-sum competition in that skill, as directed to doing good in a numerically efficient manner). I wouldn’t know how to do that though, I don’t even have many good friends.

Is the world actually that vulnerable? - Croissanthology

Looking at existing preparedness against climate, pollution, solar flares, even pandemics: no, it’s not.

The moral of these stories is that civilization is often more resilient than I had assumed. For most tail risks which are too big for markets to insure against, the answer to “what would happen if it came about” is less “oh we are so deeply unprepared it would be a knockout for humanity” and more “yeah we're 30% prepared, and in a pinch we can whip up another 50% via the indomitable human spirit”.

The Adolescence of Technology - Dario Amodei

Speaking of current preparedness against risks of powerful AI, from maybe the top organisation in the field on this issue.

the myth of mechanical engineer ai - em0sh

What needs to happened for AIs to do mechanical engineering? Lesson applies to other engineering disciplines who design and produce physical parts that fit together and do what you want it to, across a large space of “things you want your assembly to do” - mass-produce chemicals, go to space, transmit power, house people or computer infrastructure, etc.

- coding is only text. LLMs are good at this. most of mechanical engineering is multi-modal and messy

- this multi-modal problem is really difficult. even something simple like a single part is really complicated -- it's not even just 3D! you have the application layer instructions (CAD feature history), the BREP (step file, actual 3D), the CAM (here be dragons), the drawing (still lives on paper, even today), etc. one simple task, many modes…

- we haven't even gotten to the "real" engineering like stress analysis, material selection, etcetera. at first, i thought this would be easy because there is a ton of training data on engineering analysis, but i've realized recently so much of engineering analysis is experience and not knowledge. i rely on my eyecrometer more than my calculator

Most of these multi-modal tasks will probably be solved by a mix of general model capabilities and specialised training for it. Because of the connection between parts’ physical details and engineering performance, robotics will probably have to be solved before we get AIs which can engineer 90% of things from end to end. Until then, experienced humans will be necessary in the loop. But for how long? 5 years, 10 years, 25 for the hardest cases?

Going Founder Mode On Cancer - Elliot Hershberg

The founder of GitLab, Sid Sijbrandij, in 2022, was diagnosed with bone cancer, osteosarcoma. He went through the standard of care, but in 2024, he reached the end of it and the osteosarcoma was not fixed. What do you do at this point?

Over the last two years, Sid has assembled a veritable SWAT team to navigate—and in many cases create—his care journey.

Many of the ingredients resemble GitLab.

Sid developed a complex information management system (his care notes) to manage streams of complex information (his diagnostics) that guide the use and creation of complex information products (drugs).

Cancer is a disease of information. Loss of genomic information threatens to destroy the self. It may have met its match with Sid.

He used consultations with experts, as many diagnostics as he was able to do at every interval, and repurposed and individual therapies - because those are the options in at the end of the standard of care flow chart.

So far, using the Individual Patient Expanded Access - Investigational New Drug Application (IND), known as Form 3926, Sid has gained access to five experimental medicines where no trial would have otherwise been available to him. In each case, the FDA has accepted his application within 48 hours. “The FDA wants me to live,” Sid says.2

Surprisingly, hospital institutional review boards (IRBs) have been more challenging to navigate than the FDA.

In the spirit of GitLab, you can see everything about his treatment, open-source.

The BBN Fund - Eric Gilliam

To fulfill their technical visions, FROs need contractors to carry out engineering-heavy work. Who takes these contracts? Companies like Bolt Beranek and Newman.

BBNs pursue ambitious North Star technical visions with a mix of contracts and grants. The shape of technical ambition will vary. It can range from an FRO-like technical goal, such as developing methods to cheaply map mammalian brains, to building a center of excellence in an area of R&D in which academia struggles, such as bespoke biotech instrumentation or chip building.6 Some BBNs may even pursue the speculative work necessary to help establish new subfields of research, e.g. computational law. What unites BBN founders will not be the shape of their technical visions, but their capacity to fund them by solving real problems for paying customers en route to these visions.

It is now time for Stage 2 of the [modern BBN] experiment: building a “Convergent Research for BBNs.” The BBN Fund’s objective will be simple: seed a modern ecosystem of BBNs and work to maximize their overall technical ambition.2 If successful, we will forge a new pathway for today’s best applied, ambitious researchers to pursue ambitious R&D agendas — as Convergent Research has done with FROs.

Yoyo 2027

Celeste 2027

Look, all my friends have made plans for 2027. These are aggressive. Because of my experiences seeing people physically build big things, and with myself, I am a lot more conservative with timelines. So 2030. 5 years time, like those 5 year plans central national committees make.

I think I am further than these people from the frontier (of technical ability or cultural capital/social capital), yet I don’t intend to hurry getting there more than I need to. We’re here in the (digital) takeoff, yet I’d like to take it slow. Taking getting credentials unnecessarily fast, lasering in on them (in the wrong way), has already hade my life more difficult and less rewarding than it strictly needed to be. We were always in post-scarcity anyway.

- Get into computers

I realize two things over time. 1. Everything is computers. 2. Every one of my friends can do computers. And it really works out for them.

I am applying to UCL and KCL for Computer Science, I know you don’t need a degree to be good at computers; in fact doing a CS degree is mostly orthogonal to actually being able to do computer. HOWEVER, it is a good way for me to buy time - with which I can get better at computer, on my own terms, and end up with a passible credential to show in the end.

- Pass as a woman??

“You already pass enough to have 60-70% of all the social benefits of passing” - not enough! I know we won’t get to six sigma levels, but 90% or bust!

- Lovely adorable wife

I have heard that attracting a (long-term) partner is a T-shaped enterprise. As in, when you make a bar chart of different skills and good qualities of a person, it’s necessary to be good enough at all of them, and a lot better than most people on one or two. This makes sense to me. Some of the things I have in mind, where not all of them are strictly necessary, but different subsets of them probably are to different potential partners:

- dressing + grooming + accessorizing

- voice training

- exercise

- cooking, DIY work

- housework

- electronics: PCBs, radios; or low-level computer stuff

- music

- home bio/chemlabbing (lab technician work)

- woodworking / metalworking

- driving

- writing

- high level computer system management (kubernetes or GPU cluster management)

After this, I shall work on micromarriage generating activities (conferences, social gatherings, writing and reaching out to people from the internet).

- Employment I enjoy

Model for this was also T-shaped skills.

- project management

- teamwork

- to fill!!! there is a specific document i have somewhere which my old personal tutor sent us.

- industry-specific skills such as assays, computer simulations, bioinformatic packages, coding for data analysis etc. etc.

Concurrently work on microemployment generating activities (conferences, job applications (targeted, research-based), writing and reaching out to people from the internet).

However, that was until I got sent this - which tells you, keep on doing contract work. Write your CV for idiots, apply across your city. Contract work is lucrative/hour, and you will not be treated as disposable. Do contract work until you land the job you’re excited about through routes which do not include the standard application.

- With potential for advancement

Pick the right industry for a career. It’s always better if you’re in a growing, rather than dying, industry.

- Software, biomanufacturing (the kind with growth potential), managing data centers, trading, robotics?

Pick the right company in the industry. You always pick habits up from your company — there are good companies which you learn from, and there are bad companies which poison you over time.

- various AI labs (many of which pay well) (many of which are not “sexy”)

- Vow, any other cultivated meat lab. However, they will be priced these days as food processers, which means not high, which means not very high potential for advancement??

- companies that trade (many of which pay well) (many of which are not “sexy”)

- I am probably not going to get in a trading firm on my first job in tech

Pick the right role in the company. Very ideally: not sexy, is a niche, pays well

- devops, SRE, something else? unclear. pending more investigation.

- Giving me financial security

If everything goes right, financial security should fall out of the part where I do or make something people want, sustainably.

- In a city I enjoy being in

Over time, aim to get good enough to be relocated to San Francisco, Sydney, San Diego or other sunny place beginning with S.

Someone tells me you CAN do projects on a tourism visa (3 months) in the US if you’re really in demand, by very small, niche orgs in particular.

- able to pursue hobbies i enjoy (that ive been wanting to pursue for a while)

Like interior decoration, crafts, cooking, keeping a pet? Normal human stress-relieving activities.

- needs a semblance of work-life balance

- which involves picking the right industry, company and role

- and be willing to negotiate for time off. Which, with financial security and niche security, should be possible.

- contribute to something that will stay after myself

At the first time of writing down this heading, I couldn’t rigorously define what this meant. Helping people? Contributing to an ideal?

Now it’s more clear: participate in a tradition of knowledge. Learn important, relevant things that have come before me. Process them through acting in the world and refining the knowledge. Record it in a medium and through to people. This way, my experiences - the information gained in my senses, the best patterns of thinking I can produce - can still contribute to something after I am gone. Thereby is the project.

As I can see right now, this comes down to:

being able to read and write in and out of my job

- Depending if I want to see my knowledge being used in my working career: probably without good reason don’t work in a top secret environment or for a national government etc.

- be able to write about what I do in my job + experiences gained

- keep on writing and publishing, so that I can write quickly + well.

- meet new people, and learn from them and test on them to iterate on my ability to communicate to and from them. Work on my rhetoric and pragmatics.

- this sets me up to take mentors later in my career and life. And maybe kids, but who knows. If I want kids, I want them at 27.5, in the middle of the next 5-year plan.

- Live

self-explanatory if you’re with me so far.

Summarizing what Celeste said.

- Celeste said that mimetic desire, she defines as being derivative and non-unique, of being [closely] clustered with others. She has a negative opinion on this.

- I disagree with this evaluation as a generalisation. I think mimetic desire is so much more than a drawback of your “average, median human”. I think it underlies most of what makes us successful as individuals, and broadly as a species - about as much as writing does. The women and men of the Renaissance had mimetic desire, aimed at the ancient Greeks and Romans. State leaders in developing Asian countries looked towards the Soviets and the West. I think that was both necessary to organize them together to a vision which they all found worthy of emulation.

Much like prices are an information processing system, so is social status. When you are born, you do not know what makes a good life. There is no way in hell you can stably derive this from first principles, just look at all the different philosophers who have tried. Information on “how to enjoy your life” is distributed among the millions of individuals who are outside of your immediate access. Information on “how to be a good person”, which is about winning in group-level selection (whether that of families, tribes, nations, species or wider groups) is distributed among all the instances of that group (e.g. different nations) that exist. Each individual person or group-level being selected can only sample an epsilon-tiny space of all the states there are. And the state space is SO big -- you need all the samples you can get.

Here is where mimetics (with an i, as opposed to memetics with an e) comes in.

Look at the people around you. Notice, or ignore because it doesn’t matter whether you know or not, that they are physically living in a similar environment as you are - your local one. This means they very likely face the same constraints and pressures as you do, and need to solve them.

Notice which ones are existing (if you can see them, they exist) and succeeding - have lots of resources, have lots of friends. Look at what they are doing.

They exist, so they must have solved for survival. You need to do that too. They have also seemed to solve for thriving and slack. You also want this! Why don’t you just try what they’re doing? Sampling is hard, and the wholesale package of “what they are doing” seems to be working out. If you didn’t “want” to do what they’re doing, you’re letting all this valuable information go for free. If you did, you parallelize your information gathering from the environment, get to learn from other people’s mistakes and rewards they found, embodied in their revealed actions. Get to learn from other people’s social learning from their local networks, so on and so forth, for as many degrees as information compression is efficient.

To be honest, this is how I have made some of the best decisions I have made yet - by choosing my targets as people “who are like me”, and are happy and succeeding. This is how I am getting socialized into being a proper person after a long time of not being that. The day-to-day of this looks like becoming an owner of a framework laptop and a flipper zero and a red checkered flannel overshirt. Writing this blog post. And going on estrogen, of course1.

As the English and Art teachers say, you have to master the rules before you learn how to break them. Mimetic desire is the first step to becoming as good or eventually better than the targets of your envy.

What if you say that: the average human doesn’t seem to do well at this. They seem to either want to spend time at the local alcohol hole and sports stadium all day.

Or be anything but college student #29573 that sends instagram reels to his friends and gets drunk on the weekends and doesn’t have a single unique thought.

Well, I think this is because most people have bad taste in who they want to become. Imagine the life of college student #29573. Maybe he fantasises his girlfriend wasn’t so like herself. And of getting his way in the national electoral theatre. And of getting a stereotypical, highly-paid job in the local high-paying sector. This way he has a stable, high position in the local status hierarchy, among his friends and enemies (who are both his competitors).

Um, ok. I’m glad the local hierarchy works for you. But if you happen to read enough reports from outside your local environment, or hear from enough travellers, you could start to be aware of what a better world there can be out there. There is also the “openness to experience” factor, which could be an innate dial. But I think most of the time, people just don’t know, or happen to be so invested in the first sphere they are born into before seeing anything from outside it that they don’t desire to know anymore. It’s not relevant for them. And then there, the information line gets dropped.

Taste is so important to develop. Statistically, you need to sample widely from types of people you want to emulate before finding one that a) is the best available and b) is fitted for you. This is what Knowing Thyself helps with.

Aside: one big reason that autism is a disability is that it removes your ability to access this way of figuring out what is desirable to others, and therefore to build and calibrate a mimetic desire model at the rate of a non-autistic.

One thing which I fixate on is getting information where it needs to go. Mimetic desire is a large part of how the world runs. I posit that it powers (motivates) a large part of social learning, which is one of the most efficient ways of transferring information from where it is to where it needs to be, as carriers of valuable traditions of tacit information live and die and gradually get replaced.

To Celeste: if you over-index on originality, you can re-invent the wheel; and we all know that the type of people EAs are loooove reinventing the wheel. Maybe you are good at inventing wheels. I am not good at this, and prefer standing on the shoulders of former trailmakers before me, at least for now.

Love,

Ada

tldr: Somehow it feels like the US is the promised land. It is destiny manifest. The stories it tells itself holds it to be so, and that’s part of the reason it’s also true.

## industrialized ideals

I loved the cars. My Bostonian friend, a leftist, doesn’t like the cars. “you’ll learn” - but still, the roads are so wide, and the cars are so many, that I found them - a reflection of the system that builds these roads, and trades for these cars, beautiful. I loved the public spirit of the Massachusetts State House, the Museum of Fine Arts, the Boston Public Library. It feels-- utopic, because the city was centrally planned. It gives the feeling of somebody up high having provisioned for us, rather than the haphazard restructuring of old rooms, leftover train tracks, leaking pipes that is the UK. The solid stone and brick buildings of Boston tell this story. That of things being worse in the past, but that of being thought for by the men of old. The effect of order gained from the result of planning on greenfield sites treats my brain in just the right way. It inspires awe. I don’t know if my friend knows how good they have it.

The Boston Public Library was built, decorated worshiping science, commerce, progress. Exhibits of the Boston School of painters in the MFA draws beauty from curated scenes - use upper-class domestic women, paint them in a casual, mid-action manner. Contemporary objects. Older methods. A breath of fresh air to me. I can relate to these women, see my friends in them, those upper upper middle class writers of trading algorithms or millionaire retired Anduril early employees.

I can relate enough, through my friends, to aspire to be those women. I carry my printed Works in Progress magazine. I see the ideas reflected back by the stone, and the oil on canvas, to the words and the design style in my hands. It’s pretty obvious once you see them both.

## every day was the end of yesterday

The progress studies movement advertises itself to be a bringer of change to the moment, looking at the progress of the past. Productivity is not going up as fast as it was in history. We are (some of us more than others) drowning in legislation and bureaucracy. If things go on like it has for the past 50 years (roughly since 1971), we’ll have lost a real generation of any change, and stasis, in a way, IS like death. “Where is my flying car?” - a tagline that organized the creation of the beautiful, timely artifact I hold in my hand. But look back 100, 200, 300 years. Over the first industrial revolution as a whole, yearly real GDP growth was 0.5% - today’s is 2% on a normal year, or a full four times that.

The truth is, it is not the end of the world. Or if it is, it has been the end of the world many times. WWII - nuclear armageddon - overpopulation - the oil crisis. Race relations, climate change, AI, or Chinese industry will not bring the United States to its knees. It is more resilient than that - one of the strongest egregores, or family of egregores, which we have ever wrought. I can see it in the flags.

They are everywhere from the buildings to the train stations to lawns to the upstart factories and machine shops. The US has the command of the edge of technology, and it starts and stops at this edge, but despite this starting and stopping the edge does not seem to be going away. And technology has been the key to national sovereignty for a long time. The great metropolises of the East Coast has stood, and grown, and adapted, for 250 years. And they do still.

## the us <=> utopia

Think of what utopia is like for me. A vast expanse, free to travel across. A common language, so you can talk to the locals wherever you go. The most beautiful sights, the highest technology, the most dedicated art of every genre. The economic surplus to make this all happen. The freedom to choose and group with your own people. Massive, artificially flavour-optimised portions of food (ok. China has this too). It’s called artificial because the flavouring is an artifact. There’s an art to it.

The US was founded on some ideals. when it was founded, the average citizen didn’t have much (they were poorer than african nations are today). Average life expectancy was under 45. The country was lucky to inherit a relatively stable legal tradition to build upon, an “open canvas”, and the clearest minds on governance of the day.

The US was so successful it re-wrote the idea of what utopia can be. Sometimes, San Francisco is held as heaven is to some of my weird, techy friends. New York is the greatest city in the world; Seattle is “transgender Mecca”1. It is probably the country with the most active socio-political manifestoes per capita these days, edging out the UK. (And you need to count the films and science-fiction novels2 that act this way too.) People test ideas of how to live their lives, the ideas diffuse, the ideas make other people. Hard power wielded well means soft power means that everybody across the world not only knows your issues3, but actively reframes their own issues along your lines, even without any reason to. Vendor lock-in for schemas on the societal level, for better or for worse. Per-capita GDP so high that people fight over cultural issues rather than the all-encompassing struggle for material well-being. Or maybe it’s the civic culture they inherit. Or both.

## hill-climbing in high dimensions

Boston has been succeeding. MIT, Harvard are there; so are many other good universities and colleges, with students each writing their own stories, setting the standard for the world over. It is old - these institutions have tradition - but it is experimental. Was always experimental. The US is the cultural and technological research lab of the world. The existence of a wealthy and powerful united states leads to so many positive externalities it’s not funny. And here, you often get to taste the best, first. Nowhere is perfect. But the US - specific parts of and depending on the whole - seem to asymptotically approach perfect, as much as recursive self-improvement asymptotically brings us to the singularity at 2% closer per year.

So it gets better. It keeps getting better. The material wealth, the ability to lose everything in a hurricane, move across the country and build back again - the cultural slack it takes to be casual with your language and your actions, to live in (if not by income) a socially flat, friendly and open world. The existence of cities like Boston lay an example. It says to the Old World that things can be different. That these qualities, this life - can be ours.

To the united states. You are truly something special.

With love,

Ada

David Friedman’s Substack

Externalities, Population and Climate

All societies face the coordination problem: In order to do anything complicated you have to somehow get millions of people to coordinate their activities. To make automobiles you need steel. To make steel, you need iron and carbon. To make iron you need iron ore and coal…

Read more

3 years ago · 62 likes · 47 comments · David Friedman

1. Maybe climate change could be on-net positive for humans when all its effects are taken into consideration: specifically, opening up more polar areas for settlement, fertilising crops, and reducing deaths from cold as it increases deaths from heat. The overall verdict is that its net effect on welfare, like many changes with wide-ranging effects is unknown.

2. the Chinese government have been unilateral actors on its own population (as it always is) and has “saved the world” twice - once from overpopulation, and once from climate change through electrotech investments.

3. David Friedman assumes technological progress is exogenous and assumes the effect will be massive over the modelled timescale (3x current output over the next 3 centuries!), which may well be a correct median estimate but it takes effort to make happen.

from https://www.huskymachining.com/ through Making liquid nitrogen from scratch (an absurd amount) - NileRed

Experimental History

Thank you for being annoying

“Do what you love” is the most dangerous sentence in the English language…

Read more

9 months ago · 703 likes · 91 comments · Adam Mastroianni

I think this is actually the way most good-hearted people work: they’re motivated not by warm fuzzies, but by cold pricklies. They help because they can’t stand the sight of someone in need.

They help because they’re annoyed.

Another piece in Experimental History’s unofficial “theory of motivation” series.

Animal Photo Art References Search - x6udpngx

Find animal pictures with their head pointing in any direction!

The Fork in the Road: AI Horseless Carriages or Collective Intelligence for Science

1. The social processing of a group of non-localised scientists

2. More bite-sized, clear and modular records of science, as it’s done live — “nanopublications”

3. This creates a better source of ground truth to “how science is done”, letting AI collaborate with scientists meeting them where they are to empower their work.

Even though this piece is written in an entirely different cultural dialect (the Bluesky dialect), it’s about many of the same things the rest of us are concerned about in Metascience or Progress Studies, as well as the importance of Karlsson’s/Gwern’s Scenes

see also: Cosmik awarded $1M grant from Open Philanthropy and Astera Institute for new social knowledge network for researchers

celeste-land

The passing problem

Important preface: I’m not contesting the fact that passing is something trans women descriptively want. They obviously do. This desire logically follows from gender dysphoria and a desire for normalcy: you want nothing more to be like (and treated like) that one pretty girl from your class or whatever…

Read more

3 years ago · 6 likes · 8 comments · Celeste 🌱

Why I Transitioned: A Case Study - Fiora Starlight

pronotre

Two Types of Homeostasis

Contents…

Read more

7 months ago · 3 likes · 1 comment · Lydia Nottingham

Best essay of November 2025

The point of all the resources invested in you is that you should do something exciting with your life, lest you fall into the trap of Level 1 (not improving) or Level 2 Homeostasis (improving, but only as much as the most likely path all along from an outside view).

Lydia has vivid mental models of what’s going on, and out of the couple of writers I’ve read currently at Inkhaven, she is the best one.

Also: Lydia’s field notes from EA Global: New York City in October, which I saw her and met various other people at. Looks likely representative for the state of a slice of the AI safety field in late 2025.

A pretty, colourful graphic on how people use ChatGPT.

em0sh

A Letter to Software Engineers Wanting to Help Mechanical Engineers

Why I’m Writing This…

Read more

a year ago · 5 likes · 2 comments · em0sh

Request for Startups from Em0sh (more accurately, a request for specific better tools for mechanical engineering on the computer, and tool integration)

em0sh

Defining Open Source Hardware

TLDR If I can’t edit, see or contribute to your calculations and CAD, it’s not open source…

Read more

2 years ago · 1 like · em0sh

Bentham's Newsletter

We're Not The Center of the Moral Universe

The Mormons say that God instructed Joseph Smith to have a bunch of hot, underage wives (they don’t usually phrase it that way). I’m skeptical. While it seems rather unlikely that God would be in favor of such an arrangement, it seems quite likely that…

Read more

7 months ago · 21 likes · 10 comments · Bentham's Bulldog

Beware suspicious convergence of moral priorities.

The Works in Progress Newsletter

More pieces we would like to commission

This April we published our first list of pieces we would like to commission. Some of these ideas have already turned into great articles, such as on the wonders of modern drywall and an 80,000 year history of the tomato. Others are planned for upcoming issues, including our…

Read more

7 months ago · 123 likes · 16 comments · Works in Progress

Request for Articles from Works in Progress.

There are a couple of pieces about pharmaceuticals. Also, Works in Progress is looking to fill ads, for free! Join the slack channel to learn more.

Clever Carnivore’s culture media cost – which has been at the $0.07 level for two years now – undercuts its competitors.

“Our path to approval should be relatively smooth because the groundwork for evaluation has already been laid” - CEO Virginia Rangos

AI Safety for Fleshy Humans - Nicky Case & Hack Club

(Nicky Case is a she now? wow.)

Putting the “You” in CPU - Lexi Mattick & Hack Club

How does your computer work on the low level?

Anti-TechBio - Ron Boger and Dennis Gong

The Mountain is High, the Lead Investor is Far Away - Rhodium Group

Principle-agent problems! at the chinese local banking office.

Molecular Cause of Death - Dennis Gong

A recent GWAS of genetic variants associated with human height showed that all we needed to understand the missing heritability was more samples.

It Can Think!

How can we get enough data to train a robot GPT?

It’s no secret that large language models are trained on massive amounts of data - many trillions of tokens. Even the largest robot datasets are quite far from this; in a year, Physical Intelligence collected about 10,000 hours worth of robot data to train their first foundation model, PI0. This is something Andra Keay…

Read more

a year ago · 31 likes · 2 comments · Chris Paxton

MOD 171

Why Psychology Hasn’t Had a Big New Idea in Decades

A guy named Thomas Kuhn gave us an idea: Every scientific field has its own paradigm — a set of assumptions, methods, and commitments that shape how scientists think. Paradigms frame what counts as a valid question, suggest what methods make for a meaningful study, and determine how problems are solved…

Read more

a year ago · 141 likes · 25 comments · Ethan Ludwin-Peery

Seeing like a Bank - Patrick McKensie

A Visual Guide to Genome Editors - Evan DeTurk, Asimov Press

Progress in in-vitro oogenesis.

Construction Physics

Why Are Homes in Western States So Expensive?

Earlier this month I came across the following graphic, originally posted on Reddit in 2022, showing the state of housing affordability in the US…

Read more

a year ago · 177 likes · 59 comments · Brian Potter

Remember Étienne Fortier-Dubois’s Tech Tree? Well, the appeal system is real too. Here’s how it looks for the United States:

Click through to the original embed to interact!

A single-cell journey from mechanistic to descriptive modeling and back again - Hyperparameter Space

computer should be friend - Tess, loginasroot

in a world where everything's computer, computer should be friend. computer should make you feel warm and fuzzy. when you wake up in the morning you should look forward to operating computer. computer should love you, and you should love computer.

Measured Life

Why Does Philanthropy Not Have Clear Stages of Investment?

Investment in the for-profit world follows a logical progression with distinct stages and comprehensive data tracking. Angel funders serve a clearly different purpose than Series C investors, as illustrated in numerous industry charts…

Read more

a year ago · 19 likes · 1 comment · Joey

What Some Animals Endure Before We Eat Them - Nicholas Kristof, The New York Times

Theo's Take

On Loving Animals

Anyone who’s had a pet knows how human they can seem. Cats are capricious and conniving, yet they’ll dotingly care for you when you’re sick—language barrier notwithstanding. Dogs are jolly and loyal, territorial and emotional, like a your sidekick or partner in crime. What is true about household pets holds for other animals. Pigs can…

Read more

a year ago · 23 likes · 5 comments · Theodore Yohalem Shouse 🔸

A straightforward blog post introducing all the fundamental ideas about why we offset and how we become good animal advocates - for a counterintuitive, but real, notion of good.

Aidan's Soupstack

The ethical cheat code you need: Offsetting your non-vegan diet

So, you're not vegan. But maybe you tried. Maybe you stood in the grocery store, cradling a carton of oat milk, whispering "this is my new life now." Then you spotted cheese, remembered joy, and crumbled like feta. Or maybe you got ambushed by a charcuterie board one week in and never looked back. Either way, whenever a cute animals hits you in the feel…

Read more

a year ago · 26 likes · 6 comments · Aidan Alexander

Here's the wild truth that even the most committed vegans don't always consider: offsetting has some comparative advantages over diet change alone. Think about it:

1. Your impact ceiling is limitless. Go vegan and you spare roughly 255 animals a year. Impressive! But donate enough and you could save 1,000 animals. Or 10,000. There's no upper limit to how much good you can do, if your wallet is willing.

2. You can make amends for your past. Diet change only helps animals going forward. It can't help the ones already affected by your old cheeseburger habit. Offsetting? It's like moral time travel. You might not be able to literally help the same animals your past self impacted, but you can do the same amount of good today — the next best thing. Even vegans can use this to clean their pre-enlightenment slate.

3. You're funding systemic change. Individual dietary choices, while admirable, are just that — individual. Donation dollars can fund lobbyists fighting for animal welfare legislation, corporate campaigns pressuring entire industries to change and scientists cooking up real meat in labs so you can enjoy your steak but skip the slaughterhouse. You're not just taking your business elsewhere, you're actively transforming the system.

4. It's sustainable for most people. Let's be honest: the five-year retention rate for veganism isn't great. Many people try, slip up, and abandon ship entirely. But a set-and-forget monthly donation? That's something most people can stick with for the long haul. And a consistent donor over decades will save more animals than someone who goes vegan for six months then gives up.

Ethics Offsets - Slate Star Codex

The Electrotech Revolution

Rewiring the energy debate

The debate on the future of energy is a mess. Ask a dozen pundits about the energy transition and you’ll hear a familiar chorus of gloom: electric vehicles stall, net-zero targets unravel, emissions climb, and the fossil fuel industry is emboldened. The world, we’re told, is failing to decarbonize. Energy transition? What transition? With Trump back in …

Read more

a year ago · 112 likes · 21 comments · Daan Walter, Sam Butler-Sloss, and Kingsmill Bond

Austin Vernon's Substack

The Outsize Impact of AI Logistics

AI's Impact on the Physical World…

Read more

a year ago · Austin Vernon

Best Essay of July 2025. Optimism, numbers, specific industry structure, it’s all there.

chat.html - full-featured browser llm interface from Mariven

Hedonic News - link aggregator powered by Exa.ai Websets. Made by Mariven

Token for Token

There Are No New Ideas in AI… Only New Datasets

Most people know that AI has made unbelievable progress over the last fifteen years– especially in the last five. It might feel like that progress is *inevitable* – although large paradigm-shift-level breakthroughs are uncommon, we march on anyway through a stream of slow & steady progress. In fact, some researchers have recently declared a…

Read more

a year ago · 241 likes · 20 comments · Jack Morris

“The Bitter Lesson” but for data. Each advancement is only as good as the quality*quantity of data it unlocks the system to learn from.

This discovered equivalence is really profound because it hints that *there is an upper bound to what we might learn from a given dataset*. All the training tricks and model upgrades in the world won’t get around the cold hard fact that there is only so much you can learn from a given dataset.

If our ideas aren’t new, then what is?

One obvious source of information that a lot of people are working towards harnessing is video. According to a random site on the Web, about 500 hours of video footage are uploaded to YouTube *per minute*. This is a ridiculous amount of data, much more than is available as text on the entire internet. It’s potentially a much richer source of information too as videos contain not just words but the inflection behind them as well as rich information about physics and culture that just can’t be gleaned from text.

See Why Tacit Knowledge is More Important Than Deliberate Practice - Commoncog, also about using information embodied in videos to learn what is not learnable over text.
The flip side of this paradigm is that once we figure out how to learn from embodied data, there would be a straight shot to general physical intelligence (even if not fully robust, like we are not.)

Nicholas Decker talks about India’s dilapidated courts, with sources that estimate it costs up to 10% of GDP. You think Californian permitting is slow? As of 8th June 2025, ~40 properties’ worth of permitting for rebuilding houses that burnt down in the LA Palisades fire (January) has been issued, out of ~6800. At the current rate, all permitting will be done in 53 years.

The Indian government estimated that, at current capacity, it would take 324 years to clear all of the cases. That was in 2018. Since then the number of pending cases has doubled.

The process by which a lack of court capacity impact GDP is also explained: misallocation of labour in firms, where firms do not fire, do not hire, and do not outsource (and specialise) to the extent that they would if there was a functional legal system to resolve land, labour and firm-firm disputes. The lack of judiciary capacity also makes it relatively hard for multinationals to enter and operate in the country. This delays the transfer of effective technologies and management practices into India - in practice, other countries would poach them away.

Leopold Aschenbrenner compiles economics papers from Chad Jones, on long-run economic growth. Highly recommended to longtermists: Jones’s approach, focused on the value of “new ideas”, is general enough to apply to the future and much more concrete than “superintelligence will make all predictions moot”, because it’s a model based upon historical data. Jones was one of the authors of “Are Ideas Getting Harder to Find?”, a canonical starting point in progress studies discussion.

No Magic Pill writes about tooling for semiconductor fabs, with one tool as example. Good for industrial literacy.

Common Tech Jobs Described as Cabals of Mesoamerican Wizards, in a style much like Slate Star Codex.

Colour theory, but the physics of it.

Scientific journals and how they changed over time, in order to answer the question “Why Is ‘Nature’ Prestigious?”

A different way to look at the value of art. Personally, I think the value of art can and does increase over time, as we explore more of art-space and find better aesthetic fits for every observer - it’s not just a function of novelty. (I say this mainly because I listen to 10 songs on my Spotify on repeat, and rarely vibe with new ones.)

All these are from the archive of Étienne Fortier-Dubois.

Will AI step in as consultants in decision-making about preclinical assets? Looking at the various decision points for go/no-going a typical asset, and where AI could or could not fit - from Owl Posting.

You’ve heard of broad-spectrum antivirals. Broad-spectrum vaccines. Broad-spectrum antivenoms.

But have you heard of broad spectrum cancer treatments? Some avenues of investigation that might make more of them - like the original chemotherapy and radiotherapy, that target characteristics of a wide range of different cancers - feasible in the near future. from Sarah Constantin.

Big news - an improved methodology finds more happiness than previously reported, hidden in the same scale - due to rescaling of the 1 to 10 ladder, into territory equivalent to being beyond the first 10.

We are getting better at measuring empirical well-being! - Charlie Harrison, via EA Forum

Foundational models have a ways to go in mechanical modelling yet. - Adam Karvonen

Bell Labs was the first of a series of big, centrally funded industrial laboratories in the 19th century. From the 1950s, other corporations formed their own subsidiary labs, chasing opportunities like Bell’s invention of the transistor was. But through the last quarter of the century, these labs have been gradually integrated closer with existing business pressures, or spun down. It seemed that Bell Labs, as first mover, captured most of the best scientific work and prestige from that work, starting a “bubble in the meta-idea of economically unjustifiable investments in pursuing basic research” - given that “there’s a well-known phenomenon that technological progress is often driven by bubbles”. The ideal is romantic for the scientists, it worked for Bell, it led to overwhelming positive technological externalities - but the Bell Labs model didn’t prove a high enough rate of return to go on forever. - Construction Physics

There is about a 50% chance across all categories of important inventions of not being unique. This means 50% were a unique effort for its time. - Construction Physics

Best parts from The Gentle Singularity, Sam Altman, June 2025.

The rate of technological progress will keep accelerating, and it will continue to be the case that people are capable of adapting to almost anything. There will be very hard parts like whole classes of jobs going away, but on the other hand the world will be getting so much richer so quickly that we’ll be able to seriously entertain new policy ideas we never could before. We probably won’t adopt a new social contract all at once, but when we look back in a few decades, the gradual changes will have amounted to something big.

If history is any guide, we will figure out new things to do and new things to want, and assimilate new tools quickly (job change after the industrial revolution is a good recent example). Expectations will go up, but capabilities will go up equally quickly, and we’ll all get better stuff. We will build ever-more-wonderful things for each other. People have a long-term important and curious advantage over AI: we are hard-wired to care about other people and what they think and do, and we don’t care very much about machines.

A subsistence farmer from a thousand years ago would look at what many of us do and say we have fake jobs, and think that we are just playing games to entertain ourselves since we have plenty of food and unimaginable luxuries. I hope we will look at the jobs a thousand years in the future and think they are very fake jobs, and I have no doubt they will feel incredibly important and satisfying to the people doing them.

The latest analysis of human impacts on on wild animal suffering - Bentham’s Bulldog

Everyone knows pain follows a power law intensity. I include this article to ensure that now everyone knows that everyone knows it. - Ways of Seeing

Three priority questions to investigate in animal welfare market dynamics.

Using the market structure hammer on the nail of animal farming.

Both above by Robert Yaman at The Optimist’s Barn.

Just so much good stuff in Yuxi on the Wired’s Structure and Interpretation of the Chinese Economy. Best Essay of June 2025.

State of Autonomous Vehicles - Scenarios for Widespread UK Adoption - Tym Syrytczyk.

Interesting things from this report:

• Chinese robotaxi firms (Apollo, WeRide and Pony) are setting up to be super export oriented. Making as many foreign city operation deals as they can, from the UAE, Turkey and Switzerland to various western European countries and South Korea.

• Waymos recoup their costs at current usage rates (30 trips/day) in around one year, even costing $125,000-$150,000 per vehicle.

• “the first 100 Waymos now cover 20% of all Uber rides in Austin.”

Several “big if true” ideas to know in global health and development. Noise pollution - a bit like air pollution; Flooring - a bit like sanitation; and biodiversity conservation. The part on biodiversity conservation (vultures) is the only idea I’ve seen on the EA forum before. - Deena Mousa, Under Development

The integrity of OpenAI is not doing very well under Altman (10k words of quality reporting) - The Midas Project & The Tech Oversight Project.

Very long post regarding the nature of current LLM systems, its history, and predictions about alignment. The predictions seem to flow obvious, but actually track very well. “Why are all LLMs convergent on ChatGPT?” - oh, that’s why. The piece is opinionated and the middle part has the most opinions and is a slog. - nostalgebraist

Comments on Compassion, by the Pound - Reducing Suffering

The SolidWorks Model of Simulation - Corin Wagen

Owl Posting

A primer on molecular dynamics

Introduction…

Read more

2 years ago · 34 likes · 1 comment · Abhishaike Mahajan

Owl Posting

Better antibodies by engineering targets, not engineering antibodies (Nabla Bio)

Note: Thank you to Surge Biswas (founder of Nabla) for comments on this draft and and Dylan Reid (an investor into Nabla) for various antibody discussions! Also, thank you to Martin Pacesa for adding some insight on a paper of his I discuss here (his comments are included…

Read more

a year ago · 30 likes · 4 comments · Abhishaike Mahajan

Beautiful examples of modelling as an ideal. Who cares if it’s unphysical, if it’s useful?

Owl Posting

Can AI improve the current state of molecular simulation? (Corin & Ari Wagen, Ep #1)

Introduction…

Listen now

2 years ago · 53 likes · 9 comments · Abhishaike Mahajan, Corin Wagen, and Ari Wagen

Informative on how do people use simulation right now in the life sciences - i.e. what is the current state?

Owl Posting

A socratic dialogue over the utility of DNA language models (Part 2 of 2)

Introduction…

Read more

a year ago · 25 likes · 5 comments · Abhishaike Mahajan

The indifference engine - ponnekanti

Why takeoff? Why now? Three hundred thousand years then talking machines.

Ponnekanti is the first place I’ve heard of “computational psychiatry”. Which sounds so promising, as a field! But the journals and top links are super brain-based - a lot of imaging and category deconstruction from the images and so on. Is anyone doing comp-psych, but for images and textual descriptions? For example, evaluating likelihoods of psychiatric conditions from freeform text, or even given a big set of all the images you’ve taken in your life? I would think there is not non-zero signal in these things way, which has recently been enabled for machines.

Scalable tissue biofabrication via perfusable hollow fiber arrays for cultured meat applications - Trends in Biotechnology

Hollow fiber bioreactors (for cultivated meat) have been improved by arranging the perfusion fibres in a grid. Previously the fibres have been arranged at random, due to the way the fibres were positioned randomly to each end when the HFBs were made.

This resulted in the largest piece of single-cut cultivated meat we’ve yet made (11g, chicken-nugget sized").

Hidden away in the evaluation section:

Electrical stimulations were performed to verify the contractile function of the fabricated chicken meat tissue.

With active perfusion, the contractile length of the tissues was significantly improved compared with control tissues.

Contraction… of the perfused tissue was visible… with an amplitude of ~5 μm.

The muscle is (at least a little) functional when you run a current through it!

Texture profile analysis (TPA) and free amino acid (FAA) analysis were performed.

The measured [mechanical] stress was significantly higher for the perfused tissues [when under mechanical strain - i.e. it is tougher].

The amount of FAA and the fraction of sweetness and umami were [also] higher for the perfused tissue [i.e. it will taste better].

Which just makes it seem like having a functional circulatory system (running a perfusion, with evenly distributed circulatory vessels) is an obvious improvement for all aspects of tissue health.

This was featured in NBC and The Guardian.

Stem cell-derived mouse embryos develop within an extra-embryonic yolk sac to form anterior brain regions and a beating heart - bioarxiv

Progress in artifical wombs.

After building embryoids for 10 years, we succeeded in generating structures with brains, hearts, neural tubes, somites with a yolk sac. Both original submissions using two different approaches. - Magdalena Zernicka-Goetz, X

Cleared for Takeoff? Cultivated Meat’s Commercial Flight - Faraz Harsini, via Linkedin

Comparing cultivated meat to aviation in terms of product attributes and how those determine the development timeline and market structure.

The Century of Biology

Hub-and-Spoke Biotechs: BridgeBio

Welcome to The Century of Biology! This newsletter explores data, companies, and ideas from the frontier of biology. You can subscribe for free to have the next post delivered to your inbox…

Read more

a year ago · 76 likes · 14 comments · Elliot Hershberg

If you want to watch The Big Short (I assume. I don’t remember much from the Big Short). But in text. And in biotech. And there is a happy ending (instead of the aftermath of a financial crisis).

This formula seems like something I would find in effective altruist spaces - which of course it is, since both BridgeBio and EA have some of their roots in quantitative finance.

Spectech Newsletter

Health Technology is Weird

Serious people often try to draw lessons from the quest for better therapeutics or health technology for research more generally. From how to fund and commercialize technologies to the value of patents to self-sustaining institutional structures…

Read more

3 years ago · 7 likes · 3 comments · Ben Reinhardt

Why is drug discovery structured unlike other technologies?

The downstream effects of therapeutics’ tight correspondence between discovery, patent, and product is that in the world of therapeutics:

1. Academic research, which is particularly good at discoveries, is directly connected to products.

2. The lines around patent infringement are reasonably clear.

3. Because there’s a close correspondence between patent and product, patents get in the way of innovation relatively infrequently (Their effect on drug prices is another issue.)

4. Because of #2 and #3, patents actually serve the purpose they were intended to and incentivize work by enabling an inventor to capture the value of their work.

5. Because of #4, it’s common and straightforward for the discoverer of a drug to become wealthy off of its sales.

None of these are the case for most other technologies.

Musings by Sepia

A company you have never heard of is revolutionizing cell-free protein synthesis

Cell-free protein synthesis is the production of proteins “outside of the cell” using only DNA, cell-free reagents, and buffers. It has always held promise for its speed and simplicity. However, two common complaints of the technology have been its inability to make complicated proteins, and its inability to scale…

Read more

2 years ago · 22 likes · 1 comment · Zachary Sun

Musings by Sepia

Observations from BIOCHINA2025

There’s been a lot of talk about the rising tide of China’s biopharma community, but not as much about its counterpart in synthetic biology. To learn about this ecosystem, I’ve been traveling to China and Taiwan almost every month since the end of 2024 to get an on-the-ground viewpoint, have toured more than a dozen f…

Read more

a year ago · 24 likes · 1 comment · Zachary Sun

Torches Together

Did China’s One Child Policy Fail To Actually Lower Birth Rates?

"The One Child Policy failed to meaningfully lower birth rates…

Read more

a year ago · 2 likes · 2 comments · Torches Together

The Century of Biology

On Modality Commoditization

Welcome to The Century of Biology! This newsletter explores data, companies, and ideas from the frontier of biology. You can subscribe for free to have the next post delivered to your inbox…

Read more

a year ago · 108 likes · 5 comments · Elliot Hershberg

Essential Technology

A Beginner’s Guide to Scientific Roadmapping

Science needs more catalytic projects so that we can build essential technologies…

Read more

a year ago · 51 likes · 1 comment · Adam Marblestone

✨introducing: Chicken Credits✨ - Louis Arge, via Twitter (X)

Incautious Optimism

Electric aviation: High hopes, low battery.

Doot-deet-doot-deet…

Read more

a year ago · 15 likes · 3 comments · Incautious Optimism

Onboarding Strategy for EA Groups - guneyulasturker🔸, via EA Forum

DYNOMIGHT INTERNET NEWSLETTER

Limits of smart

dynomight.net/smart…

Read more

a year ago · 64 likes · 35 comments · dynomight

The Anatomists

He touched the thyroid

In the 1800s, surgical success was very much defined as getting the patient off the table. The really ambitious surgeon aimed for getting patients out of the hospital. Kocher had changed what was considered possible— but he hadn’t changed the definition of success. Not yet…

Read more

a year ago · 5 likes · 2 comments · Laura Mazer

Q: How did medicine historically progress? A: Surgeon by surgeon.

Scientific Discovery

We don't have to sit back and just watch the horror unfold

This month, I signed a pledge to donate at least 10% of my lifetime income to effective charities. And this week, I decided to donate 19% of my annual income, with more than 80% to a charity working in Nigeria to complete a water and sanitation project. The project was being funded by USAID, but that funding was suddenly canceled by the Trump administra…

Read more

a year ago · 67 likes · 6 comments · Saloni Dattani

“Most painful condition known to mankind”: A retrospective of the first-ever international research symposium on cluster headache - Alfredo Parra🔸, via EA Forum

The Simple Heart

I Was Wrong to Condemn Cage-Free Eggs

In 2016, I walked into a cage-free egg farm and accomplished what was, at the time, my greatest success as an activist. I was leading an investigative team with Direct Action Everywhere (DxE), and we documented tens of thousands of birds crammed in industrial sheds, covered in feces, and (in some cases) literally being eaten alive. When we published foo…

Read more

a year ago · 82 likes · 26 comments · Wayne Hsiung

Experimental History

New paradigm for psychology just dropped

I’ve complained a lot about the state of psychology, but eventually it’s time to stop whining and start building. That time is now…

Read more

a year ago · 431 likes · 123 comments · Adam Mastroianni

The Optimist's Barn

Individuation is a Key Challenge in Animal Welfare Technology

Technologies in animal agriculture (not just around welfare) can be separated into two broad categories: ones that operate at the level of individual animals, and ones that operate at the group level. Examples of group-level technologies include things like a barn’s ventilation, lighting, and heating systems; incubators in a hatchery which distribute he…

Read more

a year ago · 5 likes · Robert Yaman

The economics of fish farming and fish welfare in Europe, a systematic review - Ren Ryba, Animal Ask

Liveware

Actions per minute

One day in mid-March I saw the future…

Read more

a year ago · 5 likes · Alex Telford

Starcraft and the current state of AI Agent Management.

Construction Physics

Japan and the Birth of Modern Shipbuilding

During WWII, the US constructed an unprecedented shipbuilding machine. By assembling ships from welded, prefabricated blocks, the US built a huge number of cargo ships incredibly quickly, overwhelming Germany’s u-boats and helping to win the war. But when the …

Read more

a year ago · 114 likes · 13 comments · Brian Potter

Bentham's Newsletter

Long-Run Human Impact On Wild Animal Suffering: Much More Than You Wanted To Know

1 Controversy…

Read more

a year ago · 20 likes · 30 comments · Bentham's Bulldog

Memory is upstream of agency.

ordinary instants

upon reflection

I spent a lot of time staring at my own reflection in 2020. Zoom was one reason, of course: I’m embarrassed to admit that even today, my eyes tend to drift toward my own pixelated thumbnail. Yet another was that I had just started immersing myself more seriously in the art of drag. Before the pandemic, it had been nearly impossible to find enough time o…

Read more

3 years ago · 15 likes · john d. zhang

Level Zero Health - linkedin

Company which started in London, commercializing continuous hormone monitoring devices (as per Feb linkpost). They’re hiring!

Rough Diamonds

Book Review: Affective Neuroscience

After years of clumsily trying to pick up neuroscience by osmosis from papers, I finally got myself a real book — Affective Neuroscience, by Jaak Panksepp, published in 1998, about the neuroscience of emotions in the (animal and human) brain…

Read more

a year ago · 17 likes · 2 comments · Sarah Constantin

The Scholar's Stage

Why Republican Party Leaders Matter More Than Democratic Ones

The Republican and Democratic parties are not the same: power flows differently within them. The two big political news items of this week—the happenings of the Republican National Convention and the desperate attempts of many Democrats to replace their candidate before their own convention next month—reflect these asymmetries. Nevertheless, many discus…

Read more

2 years ago · 122 likes · 20 comments · T. Greer

Democrats debate identities, Republicans debate ideas. Necessary knowledge in an age where the Republican Party holds the presidency and both houses.

Startup Strategy for Commodity Products - Austin Vernon

How do we scale a commodity, such as steel, cement, energy storage, or god forbid - cultivated meat??

The answer: simplify, iterate, and simplify again.

Best Essay of March 2025

Desystemize

If You're So Smart, Why Can't You Die?

1. When a coffee shop makes a bagel, it’s a pretty good bet they can make a croissant as well. Not every shop that has one has the other, but they’re pretty strongly correlated. We call this correlation “baking…

Read more

a year ago · 138 likes · 14 comments · collin

Against intelligence explosions using various angles of approach. This puts into words some of my common assumptions: that like other products of science and industry (see: petrochemicals, digital computing) AI will permeate society and affect it broadly over a scale of decades, not necessarily years or months like a runaway reaction.

Update: one angle to look into that does not promise against an intelligence explosion is self-recursive algorithm improvements without hardware improvements.

Defender’s Corner

A Beginner's Guide to Culture Science

I’ve stumbled onto a scientific field that, as far as I can tell, is still very nascent. There are many working in it, but they all use different terms & language. It’s historically been very difficult to study this rigorously and openly, but I think this is changing now. I think the missing piece is active participation from the people being studied…

Read more

a year ago · 62 likes · 12 comments · Defender

A compilation of recent writers about (real, social) memetics. One of these days I’ll be able to get a real analytical trace of my intellectual and socialcultural heritage, like a sort of 23andMe style-thing. One that’s much more interesting than just genetic ethnicity, considering how I live way more in the noosphere than any physical place.

Update: 23andMe is apparantly going bankrupt? and all their (ID-matched??) DNA data is about to be auctioned??

word-cluster-matrix - Github

An (experimental?) implementation of a mimetic influence tracking tool that works from real time text. This is cool because fields - I’m thinking of mainly biology - are built from their tools.

Refactoring: Culture As Branch Of Government - Slate Star Codex

Memetics Timeline - Tim Tyler

Memetics as it existed in history, last updated 2012.

Venice

“Private and Uncensored” AI platform that routes your requests to decentralised GPUs.

SesameAI

Check out the state of the art in voice-to-voice AI. It is unbelievable good, and even ranks as slightly more human than human speakers in testing.

R1 (Zero) is on OpenRouter! With Grok 3 and now this, it’s easier and easier to get uncensored AI.

Update: DeepSeek V3 (Base) is also on OpenRouter!

Mr. and Mrs. Psmith’s Bookshelf

REVIEW: A Means to Freedom, by H.P. Lovecraft and Robert E. Howard

A Means to Freedom: The Letters of H.P. Lovecraft and Robert E. Howard 1930-1932 (eds. S.T. Joshi, David E. Schultz, and Rusty Burke; Hippocampus Press, 2017…

Read more

a year ago · 65 likes · 16 comments · Jane Psmith

The key framing in this piece is about America 1, 2 and 3. Related to the idea of how Western Europe summoned the timeless, rootless idea-monster of rational liberalism and had it took over itself, leaving many old, real “European” traditions dead that the average person will not remember.

AugmentOS

The open source smart glasses operating system used in several hackathons this month. I’ve never seen this before — it seem like it will be so useful soon!

Farmkind’s Petty Protest Compassion Calculator

Animal suffering offsets for you and your friends. A reminder that money is the unit of caring.

Looks to the Moon

Three Hundred Ways It Can Hurt to Be a Man — Index

This series is now available as a single pdf document for easy reading. Find it here…

Read more

4 years ago · 11 likes · 1 comment · Elodes

Or (with apologies to Elodes) : Three Hundred Reasons to Transition MtF Before Puberty

Many of the social reasons - especially the part where being a non-traditionally-conforming woman is better in a WEIRD society than a non-traditionally-conforming man - are pretty consistently derivable. This especially speaks:

Ultimately what’s much more important than a role’s inherent quality (insofar as that’s a coherent concept), is how well a role suits you. To many feminine men, it does them no good at all whatever privileges come with masculine roles for men, because those aren’t roles these men were ever interested in filling anyway.

It is almost as if the roles are an infinite game, but all the moves you were going to play were already dictated by chance in advance. But what if you can’t handle specifically the downsides of the role you prescribed? What if you’re just asymmetrically bad at it, compared to the other one? What if different physical and social parts of the role are perceived together and because you don’t like it on net, you end up reasonably hating every single bit of the role you play and how you’re treated?

I still swear by giving children full information and full choice, and this piece is a great proof-of-concept of what that in theory could look like (specifically the part where this is a counterbalance against a subjective culture that the author himself belongs to). Even if right now it’s a messy, overlapping yet self-contradictory list of rephrases and qualitative codes.

HTM is robust to noise, and has high capacity (it can learn multiple patterns simultaneously). When applied to computers, HTM is well suited for prediction,^[1] anomaly detection,^[2] classification…

There is a January 2025 paper on this method being used for water treatment plant online monitoring. They achieve 92% F1 (F1 is a geometric mean between precision and recall), which is equal to the state of the art.

How does this architecture compare to others time-series anomaly detection methods Time-k-means? Simple linear regression? I’m not experienced or well-read enough to say.

Think-in-Memory: Recalling and Post-thinking Enable LLMs with Long-Term Memory - arxiv

A way to get make “agents”, or at least personal assistants, that feel persistent? Sounds simple enough to pull off…

(1) before generating a response, a LLM agent recalls relevant thoughts from memory, and (2) after generating a response, the LLM agent post-thinks and incorporates both historical and new thoughts to update the memory. Thus, TiM can eliminate the issue of repeated reasoning by saving the post-thinking thoughts as the history.

Reading about adding memory to LLMs and finding out so many old architectures for AI exist. Copycat, LIDA, etc. Who is doing the work of redeveloping them, either running them with modern data and CPUs or parallelizing them and using them on modern accelerators? Because it’s very weird all the smartest models you can talk to today are based on one architecture: transformers. There’s no way that that’s the only one that works. Can an expert fill me in on why some of these ideas from the 1960s-1990s won’t scale?

“The history of technology is full of brilliant ideas that arrived decades early—only to become viable when upstream progress in materials, computation, or economics finally caught up.” - chatgpt (eigenrobot’s)

i.e. reality has a surprising amount of detail, and there is always plenty of room at the bottom for improvement.

The Bitter Lesson - Rich Sutton

A wearable aptamer nanobiosensor for non-invasive female hormone monitoring - Nature (2023)

A tiny piece of electronics and fluidics fit around a finger - someone should definitely get into and manufacture these. The strongest case is that this can measure so many more things than estradiol - this publication’s technology was repurposed from the base technology itself. There is such a massive market opportunity here.

Related? : The Enchippening (Sarah Constantin)

HRT: inside the complex global supply chain behind a $20bn market - The Guardian (2022)

the API (estradiol) is chemically transformed from similar compounds found in soybeans to be more potent (which in this case means to be bioidentical). The batch the Guardian investigated were grown in Northeast China, processed in China and further processed (“cooked”) in the Netherlands. You can see some of the machines that they use in the cover photo.

Shop the Look - Amazon.co.uk

After a weekend of difficult coding a custom pipeline at the computer, this was in fact the best solution I could find to shop pieces from Pinterest photos.

Affective Medicine

Good Reasons for Bad Feelings

The first two years of medical school is a firehose of content, and there’s only a few ways to organize it. MS1 traditionally covers “how the body works” with anatomy and physiology, while MS2 gets to “how things go wrong and what to do about it” with pathology, microbiology, and pharmacology. The alternative is a systems-based approach where, for examp…

Read more

4 years ago · 17 likes · 1 comment · AffectiveMedicine

In humans, “strong selection for extreme mental capacities may have given us all minds like the legs of a racehorse, fast but vulnerable to catastrophic failure.” This fits nicely with speculation about schizophrenia and autism emerging out of rapidly expanding human capacities for language, theory of mind, and other cognitive advances.

Keeping in mind schizophrenia, autism, and bipolar disorder all exist on a scale, are heritable, massively polygenic, strongly selected against in most if not all environments, and yet stable in every population everywhere at ~1% prevalence.

Affective Medicine

Why People Die by Suicide

A CDC report this year earned extensive media coverage with data on the poor mental health of U.S. high school students. In addition to sky-high rates of sadness and hopelessness, 30% of girls endorse seriously considering suicide, up from 19% in 2011, with 13% reporting a suicide attempt in the past year. Subsequent commentary has focused on potential …

Read more

3 years ago · 19 likes · 13 comments · AffectiveMedicine

I feel seen in this piece - especially at the intersection of “I am alone” and “I am a burden”. Thankfully, now I feel better - not as much of a burden anymore.

The relative benefits for environmental sustainability of vegan diets for dogs, cats and people

Dogs and cats consume 20% of all animal protein in the US and 9% in the world. Offsetting that much animal protein consumption will not have insignificant impact on all farmed animals and the environment as a whole.

The Personal Analytics of My Life - Stephen Wolfram

Stephen Wolfram is someone else who sleeps from 3am to 11am.

A History of the Future - No Set Gauge, part 1 of 3

Best Essay of February 2025.

Hymn of Breaking Strain

The careful text-books measure
  (Let all who build beware!) 
The load, the shock, the pressure
  Material can bear. 
So, when the buckled girder
  Lets down the grinding span, 
The blame of loss, or murder, 
  Is laid upon the man. 
    Not on the Stuff—the Man!

But in our daily dealing 
  With stone and steel, we find
The Gods have no such feeling
  Of justice toward mankind. 
To no set gauge they make us— 
  For no laid course prepare—
And presently o'ertake us
  With loads we cannot bear: 
    Too merciless to bear.  

The prudent text-books give it 
  In tables at the end–
The stress that shears a rivet 
  Or makes a tie-bar bend—
What traffic wrecks macadam—
  What concrete should endure—
But we, poor Sons of Adam
  Have no such literature,
    To warn us or make sure! 

We hold all Earth to plunder—
  All Time and Space as well—
Too wonder-stale to wonder
  At each new miracle;
Till, in the mid-illusion
  Of Godhead 'neath our hand,
Falls multiple confusion
  On all we did or planned—
     The mighty works we planned. 

We only of Creation
  (Oh, luckier bridge and rail!) 
Abide the twin damnation—   
  To fail and know we fail.
Yet we–by which sure token
  We know we once were Gods—
Take shame in being broken
  However great the odds—
    The Burden or the Odds.

Oh, veiled and secret Power
  Whose paths we seek in vain,
Be with us in our hour
  Of overthrow and pain;
That we–by which sure token
  We know Thy ways are true—
In spite of being broken,
  Because of being broken,
    May rise and build anew.
    Stand up and build anew!

— Rudyard Kipling, 1935

Center for Educational Progress

Our Education Reading List

Introduction…

Read more

a year ago · 14 likes · 8 comments · CEP

The advertisement OpenAI put on the Super Bowl! The most expensive advertisement you can place, on a $/s basis.

I Met Paul Graham Once - Phill Mendonça-Vieira

Ada Udang di Balik Batu (There is a Shrimp Behind the Rock)

The phrase means, in Indonesian, that there is an ulterior motive.

why can’t we all live like Anson?

Levers for Progress - Kelvin Yu and Anson Yu

An explainer for the history of technological progressivism. Pretty 10/10

LERF: Language Embedded Radiance Fields

Find things in real, camera-input 3D space based on natural language queries!

Moth Fund

the lost charisma of capitalism

My main learning after a full year of Moth Fund has been to bet on unabashed capitalists. Having grown up in an area of Northern California dominated by hippies, this was a breed of human I’d barely been exposed to. Where I’m from, capitalism is seen as evil and a system no “morally pure” person succeeds within…

Read more

2 years ago · 20 likes · 3 comments · Molly Mielke

Moth Fund

the earnest ambitious kid's guide to investors

PG had a great post back in the day called the Hacker’s Guide to Investors that detailed how to understand, relate to, and work with early-stage venture capitalists. I’ve been thinking about this a lot …

Read more

2 years ago · 56 likes · 7 comments · Molly Mielke

Hierarchical Taxonomy of Psychopathology

A Standard-Model-like categorization for all psychiatry conditions we know of.

Neural Networks, Manifolds, and Topology

Oh, when they said neural networks are linear algebra this is what they meant!

Curius / Bookmarks for the extremely curious

A chrome and firefox extension that allows you to share bits and pieces of articles you read with all of your friends in real time! I would recommend!

Hello, process improvements in the paper milling industry!

Globally, separations account for 10-15% of carbon emissions and up to 70% of the cost of producing a raw material or chemical

Even if their membranes don’t replace phase creation separation entirely, their system can still be installed AND save paper mills money. Via Separations is cool.

Seven chemical separations to change the world - Nature, 2016

Why is separations a big deal? Well, in 2016, as with today:

Processes involved, such as distillation, account for 10–15% of the world's energy consumption.

The world’s. Period.

The authors list seawater uranium extraction and carbon capture, rare earth mining and desalination as fields in want of research. But also some classic, petrochemical separations:

Alkenes from alkanes

…A major hurdle is scaling up the membranes — industry might require surface areas of up to 1 million square metres. Deployment on this scale will require new manufacturing methods as well as advances in materials' properties.

Benzene derivatives from each other

Implementing [new technologies] will require that their viability be proved on successively larger scales before commercial implementation. Constructing a chemical plant can cost US$1 billion or more, so investors want to be sure that a technology will function before building new infrastructure.

Chatgpt tells me that 9 years later, none of these alternative separation processes have been commercialised at scale. But I assume it is happening, and I hope it happens soon.

The Polymerist

Deep Dive into DuPont

I’ve had a lot of people ask me to do a deep dive on a public chemical company and I’ve resisted for a multitude of reasons, but I think it’s because I have no idea how to approach it. I’m not a public markets analyst. I do not want to pretend to be a public market analyst—I wouldn’t even know how to pretend. This is not investment advice. This is just …

Read more

2 years ago · 14 likes · 6 comments · Tony Maiorana

(Ohio and the chemical industry have an abusive relationship).

Construction Physics

Book Review - the Sources of Increased Efficiency

A major goal of this newsletter is to understand at a fundamental level how a production process improves, in the hope that this will make it clear how to improve construction productivity and lower construction costs. We’ve previously looked at the…

Read more

4 years ago · 17 likes · 3 comments · Brian Potter

This happens to be another post on DuPont. DuPont is a big company, but if I had a nickle every time this month I read something about it, I’d have two nickles, and that’s not a lot, but it’s kinds weird it happened twice.

I hope to not over-index on this book review, because the book Brian Potter reviews is incredibly detailed. On the flip side, this book is incredibly detailed, which means if you read it you can probably have lots of fun getting immersed in all the details.

Technological improvements here seem to again be made of a stack of S-shaped curves. Counting by total efficiency increase, incremental changes were the most important; but without more fundamental changes that originated from the lab rather than the production line, process staff ran out of increments to improve, especially during the later stages when the market for their product (rayon yarn) was saturated and there was little financial prospect in more big investments into improving the process. It seems that new S-curves’ existence were made possible by R&D, while the “moving the process up those curves” part was done by operations staff and engineers.

Without major changes, the supply of possible improvements eventually seems to “run out”

Often process changes improved problems in several different areas at the same time:

Consider improvements that made it possible to produce thicker, heavier yarn. This increased plant output (since production was measured in pounds of yarn), reducing cost indirectly. But it also reduced cost directly, since the thicker yarn broke less often, and so required fewer repairs (which were laborious to perform). And because there were fewer breaks, the resulting yarn was also higher quality.

For these incremental improvements:

In a similar vein, improvements generally required the close interaction of many different groups. Process improvements were often suggested by the plant staff, but this frequently required consultation with the engineering department to help work out how to implement them… It was hard to improve a process in isolation, without the collaboration of everyone involved.

This was back in 2022, so Brian Potter also goes and sees what lessons he can takeaway to building construction efficiency. Read his post for the full details.

Alchemy At Scale

Big and Slow: Why Chemical Startups Are Rare

Companies love to talk about value creation, but never mention value capture. Maybe capture sounds a bit too mercenary for public-facing business speak, but the concepts of creation and capture are inseparable: companies aren’t just tasked with creating value, they’re tasked with capturing the value they create, and not everyone is dealt the same hand…

Read more

3 years ago · 16 likes · 10 comments · Darius Mortazavi

Why is value capture important, and what can you do about it?

How to get money for your deeptech/synbio startup.

Ideas on how we can do industrial policy right, applied to biomanufacturing.

Rough Diamonds

What's Behind the SynBio Bust?

Not too long ago, synthetic biology was fashionable…

Read more

a year ago · 54 likes · 6 comments · Sarah Constantin

Short post-mortems on Ginkgo, Zymergen and Amyris. For each company respectively, the summaries are:

Ginkgo

Developing new processes costs time and skilled labor; and it’s hard to get production costs down and sales up fast enough to achieve profitability in a few years.

Zymergen

You need to be onsite doing continual testing and adjustment of the process to get it working well. Vendor relationships in manufacturing need to be very tight and interdependent; your vendors are your coworkers.

Amyris

It simply doesn’t make sense to biomanufacture just anything. You only want to biomanufacture products that can’t be made more cheaply some other way.

Checking some prejudices on materials decarbonization - Ian McKay, Orca Sciences

My prior here was that biology wins at making complex compounds, chemistry wins at making simple things cheaply. This view came from looking at how things like the amino acids are produced; where chirality doesn’t matter (glycine and methionine) they’re made chemically, where it does (all the rest) they’re made either biologically or semi-biologically with immobilized enzymes.

Telescopic Turnip

Mechanisms too simple for humans to design

As we all know, humans are terrible at building butterflies. We can make a lot of objectively cool things like nuclear reactors and microchips, but we still can't create a proper artificial insect that flies, feeds, and lays eggs that turn into more butterflies. That seems like evidence that butterflies are incredibly complex machines – certainly more c…

Read more

a year ago · 7 likes · 4 comments · Malmesbury

This post wins my Best Post of Jan 2025 award.

(The implications for catgirls are too sad to discuss.)

NOOOOOO!!! This is an Engineering Problem!! Please… (inhales copium) PLEASE…

Raye's Substack

Zen and The Art of Semiconductor Manufacturing

I. BEGINNING…

Read more

2 years ago · 54 likes · 9 comments · raye

You can just do things - including getting a job at TSMC and documenting your semi-coherent semi-mystical semi-psychotic-break in the tradition of Robert M. Pirsig about it.

The Long, Slow Death of Global Development - American Affairs, Winter 2022

Asian success, in short, has obscured a bleaker picture in the rest of the world.

https://x.com/OurWorldInData/status/1755939422116184280

The ugly sides of two approaches to charity - Julia Wise

“47,500 people is about five times the population of the town I grew up in. . . . It’s useful to imagine walking down Main Street, stopping at each table at the diner Lou’s, shaking hands with as many people as you can, and telling them, ‘I think you need to die to make a cathedral pretty.’ And then going to the next town over and doing it again, and again, until you’ve told 47,500 people why they have to die.”

I like Sagrada Familia much more than Notre Dame. I have cathedral opinions because I’m in the small fraction of the world who had money to visit Europe; most people will never get to see either cathedral. Photo: Forbes Johnston

Psychiatry at the Margins

Gender Identity, Biological Sex, and Psychopathology: Back to Basics

The following post is adapted from the answers I gave to J.D. Haltigan in a Q&A we had for his substack, The Multilevel Mailer. In these answers I clarify how I think about the relationship between gender identity and biological sex and why I resist seeing trans identity as psychopathological or as a denial of biology…

Read more

a year ago · 82 likes · 1 comment · Awais Aftab

Neurophysiology and evolutionary history do not determine what we ought to do.

As it turns out, sex difference and autism are not actually the same thing.

Digital phenotyping from wearables using AI characterizes psychiatric disorders and identifies genetic associations - Liu, Jason J. et al., Cell

We show that an interpretable AI framework can objectively classify adolescents with psychiatric disorders more accurately than previously possible.

Predictive maintenance for mental health in Western countries when?

Small Potatoes

Three productivity tips for the restless (like me)

Thanks for reading Small Potatoes! Subscribe for free to receive new posts and support my work…

Read more

2 years ago · 381 likes · 47 comments · Paul Bloom

A much more realistic picture of how you might be able to get things done.

Ways to Stand Out When You Apply for a Startup - quarter—mile

Artificial Analysis

A practical Our World in Data for comparing AI models and API providers. Very easily compare important model specs: capability, cost, speed and latency. Essential information for building a wrapper.

Did you know o1-mini is the fastest model in terms of token generation speed, o1 is still the highest quality (if only slightly), and GPT4o-mini is the priced the cheapest, by a factor of 2? Maybe OpenAI still has a moat after all.

Rough Diamonds

The Great Data Integration Schlep

This is a little rant I like to give, because it’s something I learned on the job that I’ve never seen written up explicitly…

Read more

2 years ago · 179 likes · 31 comments · Sarah Constantin

At Imperial College London, me and some friends are working on a project to integrate existing engineering drawings to computers. This article explains why we have been having such a hard time. It also shows that we’re not alone.

Anywhere you read about “Industry 4.0”, you will think about this.

https://pathways.gfi-apac.org/value-chain-cultivated-meat/

New Good Food Institute page is up, and it looks beautiful! They lay out the entire production process of all 3 types of ‘meat alternatives’. My favourite is the cultivated meat page, with its nice blue accent colour. This is because I am biased.

https://progress.fiftyyears.com/

A table, showing underfunded areas for venture capital. This is arbitrage! High quality arbitrage! They are giving it to you for free!!

Construction Physics

How China Is Like the 19th Century U.S.

I spend a lot of time reading about manufacturing and its evolution, which means I end up repeatedly reading about the times and places where radical changes in manufacturing were taking place: Britain in the late 18th century, the US in the late 19th and early 20th centuries, Japan in the second half of the 20th century, and (to a lesser extent) China …

Read more

2 years ago · 200 likes · 23 comments · Brian Potter

What has been, will be again. There is nothing new under the sun.

“The parents of migrants had terrible instincts. At every stage, they gave bad advice; they specialized in outdated knowledge and conservatism born out of fear… on the job front, their advice was invariably bad.”

Being in a separate country from most of your family, having language barriers, and being a continent away has a certain way of making you feel like no-one else is experiencing the problems you are. When I read this, some recognition pinged from within. I’m relieved I’m not the only one who feels like this.

The most interesting thing about these parallels, to me, is that the U.S. and China in many ways were starting from very different places. Prior to its opening up, China’s economy was entirely state-owned and state-planned, and its economic expansion was coupled with unwinding much of the state enterprise machinery, letting small businesses form and markets bloom.

The U.S., on the other hand, was on the other end of the spectrum. Prior to its economic expansion it had an incredibly weak state, and economy driven by very small enterprises. Its development was accompanied by the creation of large, powerful companies and institutions, and moving away from the “invisible hand” of the market and towards the “visible hand” of exchanges of goods and services mediated within very large organizations.

China’s success came from finding ways to mobilize its huge number of people and hasn’t necessarily been focused on operating at the frontier of efficiency. The U.S., on the other hand, despite its comparatively large population, had a chronic shortage of labor, and much of its development was focused on developing less labor-intensive manufacturing technologies like the American System. China built its success on the back of inexpensive labor, and it remains a middle income country. (emphasis added)

I know no-one cares, but it would be lovely for advances in automation to solve our problems here. I hope “新质生产力” pivots around to this. And for the entrepreneurs, govt. officials, scientists and engineers - we have so much more to do.

By default, capital will matter more than ever after AGI

which is this tweet below but way grander.

Happy Future

Quantity Has a Quality All Its Own

Intellectual Crossroads…

Read more

2 years ago · 2 likes · Rob L'Heureux

Speculative metaphysics - thinking about emergent, large-scale phenomena as “error correction”.

Happy Future

Technology Development and the Collapse of IDM 2.0

Previously: Silicon Statesmanship 1, 2, 3, 4…

Read more

2 years ago · 17 likes · 1 comment · Rob L'Heureux

Have you ever wondered exactly who is responsible for driving Moore’s Law and how they are organized and fit within the greater semiconductor manufacturing firm? The answer is the technology development division of foundries. Here is an overview of what they work on, in the context of Intel’s old business strategy and why they failed.

Happy Future

All Progress Goes Through Cost Accounting

Previously: Roots of Progress Fellowship…

Read more

2 years ago · 17 likes · 4 comments · Rob L'Heureux

Happy Future

Kinds of Abundance

Former US president Franklin Delano Roosevelt argued that the US is defined by four freedoms: freedom of speech, freedom of worship, freedom from want, and freedom from fear. Material progress and policies focused on abundance are centrally concerned with the latter two, freedoms from want and fear. The question, then, is how do you know when enough is …

Read more

2 years ago · 19 likes · 6 comments · Rob L'Heureux

Will all our drugs come from China?

The latest things that have come from there: new, but already phase-1 validated candidates, which Western biotechs are now keen to shop for and license.

It is in early-stage (phase I), oncology, and cell and gene therapy where Chinese companies are particularly active.

Within the last few years, China now on par with Europe (both behind the US) in new clinical trials started.

All the World’s a Trading Zone, and All the Languages Merely Pidgins

I’m not so sure that difference is as big as it used to be. In modern pluralistic societies we’re all franken-humans, built from cultural and subcultural fragments. Unlike a hunter-gatherer tribe or old-timey farming village where everyone shares virtually all of their culture, modern western societies are experiencing unprecedented cultural fragmentation and constant recombination of symbols and meaning.

With people living in their own little one-person culture, all communication becomes cross-cultural, all social spaces start to resemble trading zones and all our languages start to resemble pidgins.

Miniaturization of Nervous Systems and Neurons

When biology tries to make its neural networks smaller.

Models All The Way Down

Pretty explainer of the curation of datasets used to train large image foundation models.

Personality: The Body in Society - Melting Asphalt

This is going to be an important piece of canon.

What is code? - Paul Ford

Not being a CS person, this was an amazing read and teaches me some of the background info that is so, just, necessary to know when you use a computer in this day and age. What is a ‘framework’? What does Javascript do?? How does software development generally look, from the inside? It unlocks some of these terms that I didn’t even know were important.

Alexander Hamilton’s Report on Manufactures and Industrial Policy - Richard Sylla

I didn’t know how incredible this man truly was. These quotes are not even all the good bits. Open the link and read it, I beg you.

Hamilton designed his policies to create a sound system of banking and finance for the capital needs of the government and American entrepreneurs; to use infrastructure, innovation, and technology diffusion to speed up economic growth and diversification; and to support industries crucial to US national defense in a world dominated by marauding European empires.

Manufactures is one of the first extended commentaries on the Wealth of Nations…

In late eighteenth century Great Britain, what much later would be termed "the Industrial Revolution" snuck up unannounced. By contrast, in the United States, Hamilton and a few others foresaw the Industrial Revolution and espoused public policies to accelerate it. (emphasis added)

????????

Tariff discourse:

Hamilton thought that, with a few exceptions, it was good policy to exempt raw materials used by manufacturers from import duties.

And he wanted technology transfer from other countries to his own, proposing patent protection for the transferer. Or,

As an alternative, toward the end of Manufactures he would suggest (Hamilton 1791, pp. 338–40) the establishment and funding of a Board to encourage and pay for the transfer of important foreign technologies and the migration of workers skilled in them to the United States. Hamilton (1791, pp. 308–09) lamented the "selfish and exclusive policy" of other countries that sought to prevent technological transfers.

H1-B discourse.

How does he do this? HOW does he do this???

Conservatively, this is a case of a man who wrote so much that of course he covered everything that was actually important in the subsequent, 200-year rise of the United States. But the records show that he spent effort pushing for every single one of these policy points, in his day, to his congress, in 1791.

I am reminded of this quote from the musical:

God help and forgive me,
I wanna build something that's gonna outlive me.

and in reality:

back in 1791, Hamilton had a prescient vision of how America's economy with governmental backing could shift from an agricultural economy and become a major manufacturing nation. His vision was not far off from the way that future actually unfolded.

Websites, Done Cheap

If you ever want a forever website for 15 dollars. I love its design, it just feels reliable.

The Map is Mostly Water

Small Spaces

Our peasants love their villages. The Romans were ardent patriots, when Rome was a mere township. When it became more powerful, their patriotism was not so keen. A city that was mistress of the world was too vast for the hearts of its citizens. Men are not made to love immensity…

Read more

2 years ago · 104 likes · 6 comments · Simon Sarris