How long to AGI and possibility of it destroying the world?
2031 median [2026 (lower 25%) –2043 (upper 25%) – Metaculus on @March 28, 2023
If / when A
How long to AGI
Female vs male
Female exclude often Women as Worriers who Exclude
Science progress – is it possible to ask big / general questions?
You can answer big question but you must be prepared you may be 100% wrong
We generate more and more primary data, it gets harder and harder to go to the bottom and read everything. The sort of questions you’re able to ask and answer in this way get narrower and narrower and narrower. Most make fun of this trend among academics, and rightly so, to get to these ridiculous tiny questions that some people ask. Yet in a way, that’s the only way to be a valid scientist. –
Ian Morris: Yes, this is the big methodological question for the kind of work that I started doing, because in a lot of ways, asking questions on this scale drives you to being very anti-historical. Even though they’re historical questions, methodologically, you become anti-historical.
Ian Morris: What happened with historians and a lot of other academic disciplines is in the 19th century, you get these new skills that are getting developed, which are basically about going to the very bottom of the problem, and being historical, being scientific about what you’re doing. People start saying, “The only way we can really know anything scientifically about the past is by going to the archives, where all the primary documents from the past are stored, and reading all the primary documents that are relevant to the question you’re asking — every single one ever produced. You go to the very bottom of the well and then you have a scientifically valid answer to the question.”
Ian Morris: This is an absolute breakthrough. This transforms the way we do history. It’s one of the biggest ideas in the history of scholarship. It’s phenomenal. But it does generate a problem, which is that as we generate more and more primary data, it gets harder and harder to go to the bottom and read everything. The sort of questions you’re able to ask and answer in this way get narrower and narrower and narrower. Most make fun of this trend among academics, and rightly so, to get to these ridiculous tiny questions that some people ask. Yet in a way, that’s the only way to be a valid scientist.
Ian Morris: The challenge is, how do we hang on to the seriousness of modern scholarship while asking the bigger questions that people actually care about? When you’re starting to answer these things, you’re walking this tightrope all the time, that you cannot follow the standard, traditional historical practice of reading absolutely everything relevant. I cannot read every document, study every individual artefact ever found, study the entire history of the planet. It’s just ridiculous to think that.
Ian Morris: So you start having to behave more like social scientists do, or even natural scientists do, which is you take things on trust. You say, “There are other scholars out there. I’m never going to be an absolute master of mediaeval Chinese poetry, but there are a bunch of absolute masters out there, some of them at my university. I have to take things on trust.” But the problem is, if you are a serious scholar, you know that all other serious scholars disagree about absolutely everything.
Ian Morris: This is why I always worry a little bit when I read big history books written by people who didn’t start off themselves in old-fashioned, traditional academic disciplines. If you haven’t done that, you just don’t know the kind of knife fights that go on in the long grass over these tiny little details. If you don’t at least understand how the arguments have been waged, you’re not in a position to say, “OK, here I’ve got three world-famous experts disagreeing about the details of the domestication of maize in Mexico 8,000 years ago. Which am I going to believe? Whose story is more plausible?” You’re just not in a position to judge that, unless you at least know how the arguments get waged.
Ian Morris: But the ultimate answer to your question is that you have no guarantees: you never know when you’re getting it right. That’s because nobody ever knows when they’re getting it right. There’s a great saying they have in the natural sciences that I think we sometimes forget in the humanities: “All science is revisable, no knowledge is ever final.” You’ve just got to embrace that. You do your best, knowing you’re never going to be right.
The longer in time the more scientist you need to have to solve problems.
Fragment from Where is Todays Bethovens? by Holden Karnovski where he explains
Why science is wrong so often why Replication Crisis?
I actually just wrote a paper about this for the journal Evolutionary Psychology,
I actually just wrote a paper about this for the journal Evolutionary Psychology, which had this special feature on why evolutionary thinking has not been more successful in humanistic fields. The point I was trying to make in this paper in Evolutionary Psychology is that evolutionary history — looking at the big term and thinking about it the way a biologist would think about biological evolution — is never going to take off until the people doing the little history and the people doing the big history are at least on the same page, methodologically and theoretically. Then the people doing the little history will be more willing to say, “OK, I can use this as a test of some of these ground theories” — which is what you do if you are an anthropologist. You go off on a new study.
I think one of the difficulties we have with big history at the moment is not enough people who are primarily working at the small scale, who want to use the small scale to try to answer the big questions. Or at least to nuance and clarify the big questions. There is this sort of “us and them” attitude. Like, say, Steve Pinker, the evolutionary psychologist writes this book, The Better Angels of Our Nature, about the decline of violence. The general response to that among humanists has been to tear it down, rather than to say, “Can we test it? If the hypothesis fails to meet the standards and gets falsified by my test, what does that tell us? Does it actually mean everything you’re saying is wrong, or does it mean we need to push the theory in different directions?” — the way somebody would do in physics, for example.
“The kind of questions you ask does matter a lot here. One of the classic things for the ancient Greek historian is this question of why did the Peloponnesian War break out? This war between Athens and Sparta started in 431 BC. The historian Thucydides, who lived through it, has this very famous bit at the beginning of his history, telling us why the war broke out. This is great for historians. You can assign the relevant readings to your students. You can discuss it in seminar. It is really great. One of the things that students will normally say is, “We can never answer this question, because we just don’t have enough information about what different people were thinking.”
So I like to pair this with some readings about the discussions of why the US invaded Iraq in 2003, where we are drowning in information. Or actually an even better one, which I do sometimes, is the outbreak of World War I in 1914, where of course the government documents have been declassified even. We are just drowning in information. You could never ever read it all. Yet the question remains the same as for the Peloponnesian War. The answers remain the same as for the Peloponnesian War.
It’s that sometimes the sheer quantity of information is not the issue. It’s the question — why did a specific war happen? — that we are probably unlikely ever to be able to answer, at least to everybody’s satisfaction.That is the kind of thing I think we can talk about quite plausibly here. Not the specific wars, but the general trends: What made a great power war more likely? Why is it we are now living through a time when a great power war is spectacularly more likely than it was 20 years ago?‣
Then after 1870, increasingly they said, “Oh my goodness, this is just getting scarier and scarier.” It’s like a gazillion reasons why Archduke Ferdinand did not have to die that day in the summer of 1914. It’s heartbreaking when you get into the details. So many things, if one little thing had gone right, if the bodyguard had been standing on the correct side of the car, he wouldn’t have died. World War I would not have happened. Yet the odds that something like World War I was going to happen anyway within the next 20 years was just becoming overwhelming. Every time you rolled the dice, it got more likely that you’ve got a World War I-ish outcome.
–onIan Morris(Historical methodology [01:02:35])80000 hours show
During a health symposium @StanfordMed I learned from a cardiologist colleague that even if we get our 180min of zone 2 cardio per week, the benefits are largely (or entirely) erased by sitting >5 hours per day. Solution: standing (1:1 ratio w/sitting) & 3-4, 10min walks per day.
World is a highly efficient place.
Efficient market hypothesis
"If it were so easy, why wouldn't everyone else have done it already?" Thus if something seems easy and obvious to do, you must be missing something in it's difficulty However if an idea seem ridiculous on the surface, you can rest assure that less people have thought about it deeply.
Specifically for economic markets this is the idea that any predictable nature in the market will be exploited to maximum potential by everyone who knows about them. If you know a stock is underpriced, it makes sense to buy it until it is priced to equilibrium. This is how markets work.
- Naval claims this. And a followup conclusion is that there are rarely openings, opportunities. e.g. biology the way your hand skeleton, tendons work is highly efficient and it would be hard to design it from scratch to improve it
- I think this comes from the lack of imagination. There are wells of efficiency. Biology really optimize hand for the purpose of having four finger one direction and one in another. If we would want artificially build hand with that constraints using "human parts" and adding all functions this hand does currently it would be hard to come up with better design. But this view lacks imagination. How about adding 3rd hand or more fingers? I believe there are wells of efficiency – a limited areas that nature or culture excel in innovation and efficiency. But there are huge hole in between these silos there are huge breaks.
- There is that is highly inefficient and constraining (see example QWERTY vs. DVORAK)Foundation-Paralysis Effect 🎨
- Peter Thiel in his book "From Zero to One" sees the same thing in a technology. e.g. in a startup world if there is a possibility for a business its probably already existing. It's really hard to do a new big startup because as soon as the technological opportuinty opens up somebody will figure out and go into it.
- I believe there are hundreds of new social media startups possible right now with a technology we have currently but it's really hard to pull them off (idea, design, marketing). to explain why startups are possible in the world of large companies uses the following metaphor. When small planes are flying into big cloud it's not that they have radars. They cannot see where they are flying into. The reason it's safe to do is that this is humongous space and it's highly unlikely to crash with another plane.Paul Grahamsays that it's important and difficult excercise to distinguish weather some startup idea is impossible (yes, as soon as new technology opens new possibilities it is filled with new applications) or nobody thought about it or was able to bring a successful product. Patrick Collison asked why is this so hard to make a good app. There are so many bad apps. And I think that's the core of it. Why good apps are good is not obvious. There are many things synced, working that look minimalistic and obvious but took a lot of effort to arrive at.Paul Graham
In technology winner takes it all
Not sure if following is true. There are a lot of technologies that are existing in parallel. Maybe this applies more to technologies such as electrical circuits, algorithms.
If there are two competitors in the technology (does it more apply to technologies such as algorithms usually the one that is a little bit more effective tend to slowly snowball into a bit larger, a bit larger and eventually only
Nature or nurture
"Since the 1950s, the nonprofit Holt International has helped American families adopt tens of thousands of children from Korea and other countries. Parents would sign up, get approved, and get the next available child who fit their general criteria. The process was essentially random, which gave scientists an opportunity. They could compare genetically unrelated children who were assigned to the same parents: The more the parents influenced the children, the more these adopted brothers and sisters would end up alike.What the scientists found was that the family a kid was raised in had surprisingly little impact on how that kid ended up. Unrelated children adopted into the same home ended up only a little more similar than unrelated children who were raised separately. The effects of nature on a child’s future income were some 2.5 times larger than the effects of nurture.”
"Other researchers have done further studies of adoptees and twins, with similar results. As Bryan Caplan notes in his 2011 book, Selfish Reasons to Have More Kids, parents have only small effects on their children’s health, life expectancy, education, and religiosity (though studies have found that they have moderate effects on drug and alcohol use and sexual behavior, particularly during the teenage years, as well as how kids feel about their parents).”
This is beautiful moment.
I want to share how high am I on the hierarchy ladder.
Is taking notes on net positive?
It may be pretty effective strategy for things you naturally tend to agree, things that expand your knowledge in a predictable manner.
That said these area of knowledge, perhaps in a smaller extend, is gonna be subjected to constant forgetting process as well.
But won’t work with thing you don’t understand, didn’t click, didn’t connect as an epiphany with different areas of knowledge, are on the verge of grasping. These ones will be discarded even though they have a high potential of being an important building blocks for knowledge. A high potential for changing or calibrating your views – probably the most important feature of learning.
Is this area actually expanding your understanding more effectively than the things you more naturally agree with?
Things that are counterintuitive or anti-conscious (actively being not accepted as truths e.g.
Some learn through ANKI cards I think having a wiki type of blog like mine is an incredible way to come back to the definitions. My social brain makes me reevaluate what I put in front of the public.
Add the most efficient is to: write things doesn’t click (because otherwise I would just forget it) and juxtapose contradictory arguments.
How to square The grass is greener somewhere else bias vs. My-side bias?
On aggregate is it more about sum of all parts or specific combination of ingredients?
- Universal: Cooking all things together. Cooking each thing seperately and adding one by one. No difference in temeperature or length. How much value can you get out of the two?