Hopefully Interesting

  Metaculus is a website where you can make bets on the future.  There are some websites like this, like Polymarket , where you make bets in real money but Metaculus isn't like that - generally the only thing at state is one's pride and what you can learn from the collective wisdom of everybody making predictions on a question.  I say generally because occasionally someone will bankroll a tournament on some questions they're particularly interested in getting good predictions for, I'll hopefully have another blog post on one of these soon, but that's very much the exception. Sometimes it's fun to look through the predictions I've made and see where I most disagree with everyone else.  For me the top entry is my optimism about robots compared to the rest of the site. Here I am giving robotic cooks by 2030 67% percent odds when most other people have it at 25%.  Why am I so optimistic here?  Mostly because robotic foundation models have been making great str...

 The recent 80th anniversary of the dropping of the atomic bomb on Hiroshima brought to mind some thoughts I'd had when I was watching the Oppenheimer movie a little while ago, and so I thought I'd finally get around to writing them up.   Ultimately Truman thought that dropping the atomic bombs was the best way to bring about Japan's unconditional surrender and its hard to disagree about that.  The most obvious alternative at the time would have been a land invasion of Japan.  One can certainly have "clean" land combat that takes place far away from most civilian residences and which is a matter of combat between professionals.  That wasn't what had been seen recently in the Pacific war though.  The recent Battle of Manila fought as the Allies liberated the Philippines saw the death of around 20,000 soldiers but 100,000 Filipino civilians.  In taking Okinawa 100,000 soldiers died and about 100,000 of island's 300,000 civilians, some because they h...

      With the rise of generative AI we've all seen a lot of discussion about whether AIs could be conscious, now or in the future.  While many of these discussions are productive I've seen others that look more like the proverbial youths arguing about the question  "If a tree falls in the forest and no one is around to hear it, does it make a sound?"   That's not a question you can just answer, but by realizing that "sound" can mean both an auditory experience and a sequence of vibrations in the air and so dissolve the question , satisfying our curiosity without answering it directly.     I don't see us having any hope at this moment of dissolving the much harder questions around artificial consciousness, but I do hope that recognizing that the term "consciousness" can mean different things in different philosophical and scientific systems can make discussion more productive.  I see three major schools of thought there which I'll call ...

 What rating should you give the driver who just dropped you at your door via Uber or Lyft.  Well, unless they messed up in some noticeable way you give them a 5 out of 5 rating unless they messed up somehow.  That means you give the same rating to a driver who was just passable as to a driver who was outstanding, but you don't want to unfairly sink the career of the passable driver when everybody knows a 4 star rating is terrible. In Japan things are different.  3 stars is considered normal so you can give that for a normal ride, and then you can give 4 or 5 stars for an exceptionally good driver.  And then you have Eastern Europeans who are reputed to tend to give low ratings except when service is unusually good. Normally this doesn't matter too much unless you're an American Uber driver who happens to get a Japanese rider who sinks your rating with a 3, but it can have worse consequences.  Until a GPT large language model stopped speaking Croatian becau...

It might not seem likely from the title, but Bryan Caplan's The Myth of the Rational Voter: Why Democracies Choose Bad Policies is an effective defense of democracy (previously reviewed by me here ).  It's easy to come up with cases where most voters naturally favor policy proposals that experts in the matter think are crazy.  The author of the book is an economist and it's easy to come up with lists of things economists agree on but which voters are skeptical of.  But you can find examples from plenty of other disciplines, for instance in toxicology where you would find that a large majority of voters would agree that our tap water should not contain any arsenic at all, but experts typically think our current limit of 10 parts per billion is fine.  Even in cutting edge semiconductor manufacturing where a stray atom of arsenic in the wrong place can ruin a chip they're only able to reduce contamination to 1 part per billion or so. In ancient Athens they didn't jus...

There's a famous logical paradox put forward by Carl Gustav Hempel regarding observations and logical inference.  Lets say we have some statement we're interested in like "all raven's are black".  In logical terms you could write that as "if something is a raven, then that thing is black" or "Raven → Black".  And by formal logic this is equivalent to the contrapositive statement that "if something isn't black, it's not a raven" or "not Black → not Raven".  So far so good. But if we were to go out looking for evidence that the original statement was true then most people would readily accept that looking at a raven and finding that it's black is evidence towards the idea that all ravens are black.  But looking at instances of the contrapositive, say a white object that happens to be a rabbit it seems a bit odd to count that as evidence towards the proposition that all ravens are black, even if it's an instance...