You're right again I think. As far as dislike of utilitarianism not entirely without cause in some cases; while "make ethics math" is a really good idea it seems surpisingly difficult to formalize without wierd artifacts - as a not insubstantial volume of posts on this site can attest. I imagine at least some of that resistance goes away as soon as someone perfects a formalism that doesn't occasionally suggest outlandish behavior and has all the properties we want.
Yeah, that's pretty on the nose. Even suppose you trust your philosophers and ethicists work through the merits of all the possible ethics frameworks we could use. Let them pick the best one, specify how different utilities should be framed; they'd still never be the right people to implement it in any specific decision. Real world ethics problems are still 95% other problem domains and 5% ethics.
The interview does beg more questions than it answers though. Obviously consequentialist ethics have some traction among philosophy experts. Is bioethics different for some reason? Are the vocal people shouting down these (obviously correct given consequentialist ethics) ideas on twitter and in the news even in any relevant field? Does the consensus of the field, if any, bear any relation to public policy whatsoever, or are experts merely being cherry picked to toe the party line as needed and lend credibility after a decision is made?
Speaking specifically to the difference between the newer and older batch of papers. Neither are good. In my admitedly breif skim, the older ones have an extra layer of dissonance for the same reason 20 year old TV and movies can come across as unexpectedly cringey.
These papers were mostly unoffensive and not that terrible in contrast to expectations. At the same time, I do not get any impression of relevant expertise either such that I feel good about this group being in a privileged position regarding any kind of ethics decision. They aren’t bad, just… not good enough.
I do notice (from comparing to the circa 2000 batch of papers) that value drift makes older papers seem much much worse than they would've seemed at the time. I expect 80s or 90s era papers would produce the kind of revulsion many folks were expecting.
Effective blinding is definitely more involved if one's both experimenter and test subject. It's not impossible but an assistant would help a lot. Controlling for placebo effect does seem one of the big issues at this scale.
This is insightful. The areas where strong evidence is common are largely those areas we don't intuitively think of as governed by probability theory and where classic logic performs well.
It seems like someone could take this a little further even and show that the limiting case for strong evidence and huge likelihood ratios would just be logic. This might be fruitful to unpack. I could see it being the case that our instincts for seeking "certainty" make more sense than we give them credit for. Gathering enough evidence sometimes allows reasoning to be performed using propositional logic with acceptable results.
Such logic is many orders of magnitude cheaper to evaluate compute wise vs probabilistic reasoning, especially as we get into larger and larger causal networks. There’s an obvious tradeoff between the cost to obtain more evidence vs more compute – it’s not always a choice that’s available (e.g., spend time investigating vs. spend time thinking/tinkering with models) but is often enough.
When I think about how I’m using the reasoning skills I’ve picked up here that’s roughly what I’m having to do for real-world problems. Use probabilistic reasoning to resolve simpler more object level propositions into true/false/maybe, then propositional logic to follow the implications. Fail back to probabilistic reasoning whenever encountering a paradox or any of the other myriad problems with simple logic – Or just for periodic sanity checking.
Or more completely: In the absence of malice or extreme negligence there's nothing criminal to punish at all and money damages should suffice. Given a 100x lower occurrence of accidents this should be insurable for ~1% the cost. The default answer is drivers remain financially responsible for damages (but insurance gets cheaper) and driver can't be criminally negligent short of modifying/damaging the car in an obviously bad way (e.g. failing to fix a safety critical sensor in a reasonable amount of time that would have prevented the crash. Alternately, bypassing one or more safety features that could have prevented the crash). Car companies would be smart to lobby to keep it that way as letting every car accident become a product liability thing would be much more expensive.
While it does seem there was a certain amount of shotgun aproach following a few different lines of reasoning, that critism is difficult to square with actually reading the paper. It looks like the peptide selection was largely empirical and cited. The decisions about how to actually package that info into a vacine is largely educated guesswork (as you say theory, supported by computer modeling).
"Mapping of linear B-cell epitopes by binding antibodies in convalescent
sera to a library of peptides representing viral antigens. A strong signal in a
linear epitope mapping study does not guarantee that the epitope peptide
in the context of a vaccine will trigger the production of an antibody that
binds to this epitope within the context of the virus. However, it is a good
indicator that this is at least possible."
Or as I understood from elsewhere: present antibodies from recovered people to every possible short peptide sequence and see which ones they actually attacked. Make the inference that people with less severe infection had better antibodies than those with more severe symptoms in the event antibodies differed. Package a selection of promising looking pepties into a vacine; choose enough that there's likely multiple effective peptides even if 2/3rds of the choices are duds.
I think it even adds to the horror that this senario is compatible with being a Great Filter that doesn't generate a meaningfully goal oriented successor that would do anything after destroying or stagnating us. The goal oriented Mesa Optimizer is effectively trapped inside a system that's objective is simplicity and stagnation.
I've noticed in consistently good moderation that resists this kind of trolling/power game:
Making drama for the sake of it, even with a pretense, is usually regarded as a more severe infraction that any rudeness or personal attack in the first place. Creating extra work for the moderation team is frowned upon (don't feed the trolls). Punish every escalation and provocation, not just the first in the thread.
Escalating conflicts and starting flamewars is a seen as more toxic than any specific mildly/moderately offensive post. Starting stuff repeatedly, especially with multiple different people is a fast ticket to a permaban. Anyone consistently and obviously lowering the quality of discussions needs to be removed ASAP.