Cool! This website looks good as well but I haven't tried it yet: quadraticvote.co
Hi! I really liked this post and think there should be more discussions about these algorithms on here. I just wanted to point out what he writes in the penultimate paragraph:
One particular cause dear to me personally is what I call "entrepreneurial public goods": public goods that in the present only a few people believe are important but in the future many more people will value. In the 19th century, contributing to abolition of slavery may have been one example; in the 21st century I can't give examples that will satisfy every reader because it's the nature of these goods that their importance will only become common knowledge later down the road, but I would point to life extension and AI risk research as two possible examples.
What he says here is almost the same as what we mean when we talk about the search for "Cause X". In this lingo, cause X is a public good which is currently underfunded.
Maybe a mechanism like quadratic funding could even help us with allocating funds to researchers working on niche topics? An example could be insect suffering: Suppose that many people think that there is some chance that insect suffering might be a big deal, but very few people care deeply about it (I guess insects are hard to relate to because they are so different from us).
It would cost very little to all those people who think there is a small chance of importance, because initially the funding is cheap. Yet due to the nature of the algorithm, research into insect suffering would receive a strong subsidy.
Am I getting this right?
Cool! I knew gwern but wasn't aware of his experiments, thank you.
Does anybody have examples of pre-registered self experiments or thoughts about the usefulness of self experiments?
I am only aware of Alexey Guzey's experiment on sleep.
Recently I have become quite interested in improving my running with heart rate training and read a lot about the related physiology. I would like to try out Zone 2 Training by using heart rate as a proxy and experiment with this in a structured way.
An idea about AI in medicine:
Fecal microbiota transplants have been gaining prominence as we discover that many diseases seem to have a relation with the gut microbiome. There seems to be therapeutic potential for certain gut infections, inflammatory bowel disease, obesity, metabolic syndrome, and functional gastrointestinal disorders.
What if you could use new high-throughput sequencing techniques like Nanopore to figure out what kinds of bacteria constitute the microbiome of tens of thousands of people?
Combine this with tons of computational power and the latest machine learning algorithms to find relationships between certain illnesses and symptoms and those people's microbiome. Maybe this would allow finding the perfect fecal transplant donor to reverse the relevant symptoms.
I like the framing of "optimum population trajectory", that's an idea I haven't encountered before. Thanks!
Hey Max, thank you for the links! I guess now I have some quality reading material over the holidays :)
Good thinking! Attention spans are short enough these days so 15min seems plenty :)
Hi! Just watched the video and I think it's super well produced, good job.
I was surprised how you managed to summarise a lot of the knowledge I already had on this topic in under 15 minutes.
Furthermore, I have talked to lots of people who mentioned that EA seems underrepresented in video format. YouTube seems like a good opportunity here that could be used more. Regarding this, I think this is valuable work. Also I like that you didn't give it the Effective Altruism "stamp". There should be more discussion on when it's appropriate to publicly advertise EA as a concept and when it's better to just introduce a topic we're concerned about without explicitely relating it to EA. This is related to the idea of external movement building:
Hi Aaron, great question.Let's get the obvious out of the way: For people who are still in university, in study-intensive subjects, it's a great advantage to use a Spaced Repetition System like SuperMemo. To me it feels empowering not to have to worry about forgetting. It's a common experience to feel very frustrated to study so much for an exam, only to forget most of it afterwards. This doesn't happen to me anymore, because I just know the algorithm will take care and as long as I do my daily repitions, my knowledge will get transfered into long-term memory.
Another obvious use case is learning languages. An SRS can greatly help you to learn a language much faster and this seems to be the most common usage.
One not so obvious advantage is about creativity/innovation. In my understanding, creativity has a lot to do with connecting ideas from different fields, ones you wouldn't initially notice as being related to each other. Imagine you study two different domains, e.g. Biology and Economics. Actively remembering important information from both of those might result in two at first glance disparate ideas appearing in your mind in close succession. This is what leads to creativity, you making the connection between those. This is less likely to happen if you store your information mostly externally, e.g. in Evernote.
To answer your question more straightforward: So far, I have found it most useful for studying medicine in University and learning French/Spanish.