:D mfw OUR CORE VALUES are
other people's values other people's values
other people's values other people's values
(ಠ_ಠ) mfw jenn attributes the development of civil discourse to the woke
I regret writing that as if I were confident of my opinion. Hinduism is the major world religion that I know the least about, and I knew less about it 17 years ago.
(I've read more of the Koran since posting this, and retract my statement about Islam being like Christianity. I now see Islam as a political movement masquerading as a religion, more like the Iroquois Confederacy than like Christianity.)
I didn't say Hinduism doesn't make moral claims, or impose duties. I don't consider "vegetarianism is virtuous" to be a fact about the world. I don't remember why I said Hinduism isn't about making claims about the world. It makes philosophical claims about the nature of existence, which you might say is making the ultimate claims about the world.
Probably I meant that to be a Hindu, you don't have a long list of very specific and theoretically falsifiable facts which you must believe, such as that Jesus was born of a virgin, suffered under Pontius Pilate, was crucified, and rose again on the 3rd day, etc. Christianity has a lot of claims that specific events happened at specific times and places, for specific purposes; and if you fail to believe any one of them in fact happened, for the reason given, some people say you aren't a Christian. Catholicism has a LOOONG list of things you're required to believe, called the Catechism, and many of them are impossible for most Catholics to understand. There were many centuries in which you could get burned to death for publicly denying any one of thousands of points of dogma which were officially stamped as dogma in a variety of ways (scripture, church council, certain Papal declarations, being in the Catechism, for instance). The Bible doesn't say that the Sun moves around the Earth, and I don't think it was even official dogma; but Galileo would have been burned if he kept on denying it.
Hinduism makes claims about the world, and has epics containing lots of events which maybe you are supposed to believe actually happened. I've read that some Hindus would get upset at someone who denied Rama was a real person. But I don't think anybody has ever been burned to death for it.
Re. attitude towards the gods, what I've read about Hinduism said that Hindu theologians or philosophers usually see the gods as manifestations or symbols rather than as people who live one timeline, being at just one place at one time, while the masses see them as people living within time. I have the possibly bad habit of using the name of a religion to denote the rigorous theology rather than the folk practice. Probably I do this because I was raised Christian, and the theological Christians, like Catholics and Calvinists, must continually distinguish between "real" Christians who adhere to all the dozens or thousands of points of doctrine, and "phony" Christians who just go to Church on Sunday.
I was with you until this:
"Ultimately, the civilization systematically destroys the ability of its unreasonable men to compete for the slots in the society where rationality is required to maintain the society's energy and the society looses the ability to respond coherently to threats and collpases."
The problem isn't that the unreasonable rationalists can't compete for the slots of power. The problem is that they can; and they take over all those slots, and have no conservatives around to tell them to try to understand the purpose of fences before tearing them down.
Re. "If I didn't know the history of connectionism, and I didn't know scientific history in general - if I had needed to guess without benefit of hindsight how long it ought to take to go from Perceptrons to backpropagation - then I would probably say something like: "Maybe a couple of hours? Lower bound, five minutes - upper bound, three days." :
Using gradient descent was figured out quickly. The problem was that it wasn't useful to build multiple layers using backprop, because ,as you mentioned, perceptrons used a linear activation transfer functions, so each layer of the network was doing a matrix multiplication. If it's matrix multiplication all the way down, you can't solve categorization problems which are linearly separable, which is approximately all of them.
What was hard, was thinking non-linearly. This is even more surprising to me, since Minsky's Perceptrons explicity pointed out that linearity was the problem! An entire field just rolled over and died because someone said "You can't do that with linear functions", and nobody thought of using a nonlinear function!
But there is something extremely difficult for humans in thinking non-linearly. For instance, all of the models used to "disprove" group selection assume that the contribution of an allele to a group's reproduction rate is linearly proportional to the fraction of group members with that allele. (You'd have known this over a decade ago if you'd read my wiki post on group selection before deleting it.) This even though group selection under that linearity constraint is mathematically the same as kin selection, and even though Ed Wilson said loudly for years that eusociality in insects doesn't correlate with haplodiploidy; it correlates with having a communal defensive structure. Such a structure is useless unless it is finished and large enough to protect the colony, so it provides very little value unless the number of individuals working on it is above some threshold. (Though I worked out some of the math in 2013 or so, and it is not trivial to satisfy the conditions needed for group selection; you need a specific kind of nonlinearity. It would not be the default condition.)
This post admits that EFA sometimes works, that the examples we see in college usually work, and that "Listing out all the ways something could happen is good, if and only if you actually list out all the ways something could happen, or at least manage to grapple with most of the probability mass."
People who make this argument think they have managed to grapple with most of the probability mass. So this article justifies the method it is calling out as especially bad.
Even the most moral people - in fact, especially the "most moral" people - do not incorporate the benefits to others, especially future others, into their utility functions.
I clicked on the "I changed my mind" icon to indicate that I changed my mind about this old view of mine, not that my old view changed my mind today. Oops. I can't delete it.
"Someone with a cleft palate has 25 times as great a chance of having a child with a cleft palate, as someone without a cleft palate does."
I know many people whose lives were radically changed by The Lord of the Rings, The Narnia Chronicles, Star Wars, or Ender's Game.
The first three spawned a vast juvenile fantasy genre which convinces people that they're in a war between pure good and pure evil, in which the moral thing to do is always blindingly obvious. (Star Wars at least had a redemption arc, and didn't divide good and evil along racial lines. In LotR and Narnia, as in Marxism and Nazism, the only possible solution is to kill or expel every member of the evil races/classes.) I know people on both sides of today's culture war who I believe were radicalized by Lord of the Rings.
Today's readers don't even know fantasy wasn't that way before Tolkien and Lewis! It was adult literature, not wish-fulfilment. Read Gormenghast, A Voyage to Arcturus, The Worm Ouroboros, or The King of Elfland's Daughter. It often had a nihilistic or tragic worldview, but never the pablum of Lewis or Tolkien.
Ender's Game convinces people that they are super-geniuses who can turn the course of history single-handedly. Usually this turns out badly, though it seems to have worked for Eliezer.
Calling cancer a disease is like calling aging a disease. We definitely want to call it a disease, because otherwise it couldn't get federal funding. But a doctor is unlikely to see two cancer cases in her lifetime which have exactly the same causes. Cancerous cells appear to typically have about 100 mutations, about 10 of which are likely to have collectively caused the cancer, based on analysis of the gene networks they affect. Some of the genes mutated are mutated in many cancers (eg BRCA1, p53); some are not.
The gene networks disrupted in cancer are generally related to the regulation of the cell cycle, DNA repair, or apoptosis. Any set of mutations that damages these networks sufficiently may cause cancer, but the specific way cancer develops will depend on the precise mutations. So when we ask "what causes cancer", we're not asking a question that has a specific answer, like "what causes AIDS"; we're asking a question which is more like asking "what causes my car to stop running". DNA damage may cause cancer, just like shooting enough bullets at your car may cause it to stop running.
Today we can distinguish cancers with about the level of resolution that we might say, "This car stopped running because its tires deflated", "This car stopped because its oil leaked out", "This car stopped because its radiator fluid leaked out." To fix the car, you'd really like to know exactly which of many hoses, fuses, or linkages were destroyed, which is analogous to knowing exactly which genes were mutated. (My analogy loses accuracy here because car-part networks can be more-easily disrupted, while gene networks can be more-easily pushed back into a healthy attractor by a generic up-regulation or down-regulation caused by some drug. Also, you can't fix a car by removing all the damaged parts.)
It's been obvious for many years that curing cancer requires personalized medicine of the kind mentioned in this post, in which what the FDA approves is an algorithm to find a custom cure for any individual, not a specific chemical or treatment. I'm very glad to hear the FDA has taken this step.
I expect a generic algorithm to cure cancer will require cell simulation, and probably tissue and biofilm simulation to get the drugs, siRNAs, plasmids, or whatever into the right cells.
"Multiple organizations working on parts of the same problem can achieve more collectively than one big charity alone."
Do you really mean "working on multiple parts of the same problem can achieve more than working on just one part of the problem", or do you really believe you can achieve more collectively just by breaking up one big organization into many little organizations?