Weekly Roundup #4
Relatively light roundup this week, likely due to Elon Musk buying Twitter plus the incoming midterms sucking all the oxygen from everything else. I do have a Twitter/Musk post in progress tentatively scheduled for Monday.
Prediction Markets Would Probably Be Great
Journalist Ben Collins says ‘very good example of how illogical betting markets are’ and gives classic example of prediction markets being completely logical and doing good probabilistic reasoning.
Metaculus timelines for AI winning a gold medal in the International Math Olympiad keep compressing, down this year from 2036 to 2026. This is certainly super scary and exciting on the AI front. It is also disappointing as a former aspirational math competitor, as it reveals that the IMO problems are far more formulaic and hackable than we realized. When AI makes super rapid progress in some areas (e.g. math competitions) while stalling in others (e.g. self-driving cars) relative to expectations, a lot of that is learning about the underlying task space.
People Are Trying To Destroy the Internet
Joe Manchin taking his turn? Proposing to force filing of ‘suspicious activity’ reports for essentially everything. From a few weeks ago, forgot to include before.
Sports Go Sports
It is real, and it is spectacular. Worth watching.
He learned this from playing NASCAR on Game Cube growing up. It works in real life.
Competitors mostly thought it was super cool, which is correct.
The question is, it was super cool that time. What happens when it happens on a regular basis? Looks like there is currently no appetite for a rules change until the off season. I think this is a mistake. We honor such crazy one-time moves by saying ‘yes, that was super cool and super legal, with great execution. Kudos. Now let us make sure that never happens again.’ If it happens a few more times before the season is over, that lessons the moment, and also seems dangerous and like it would rapidly get kind of dumb.
On China stocks last week, I noticed my confusion that stocks were down after the party congress. This shows how long it has been since I have been trading. Back in the day, I would not have missed what a reader reminded me about, which is that stocks are not permitted to fall during a party congress, so one should expect them to fall after a party congress. Also, others said that while it was expected the congress would be bad for stocks, it did go worse than one would have expected, which I did not realize.
The New York Times remains loosely banned here. It is no longer a strict ban, where I will not use a source a link to things like a Twitter post that leads to a NYT article. However, to the extent feasible, they are not welcome here, and this week provided a reminder of why during discussion of crazy reactions to Elon’s Twitter moves (which mostly I have put in their own post.)
Kelsey tries her best to be nice about it.
I would be less nice. They set out to use take position as America’s most trusted newspaper and use it to go after America’s technology industry while pretending to be doing objective reporting. They did their best to sabotage the people looking to give us nice things out of some combination of political motives, an attempt to protect their business model, a hatred for technology, and spite.
The Intercept on DHS’s attempts to ‘fight misinformation.’ Report of which TechDirt reports is absolute garbage. Kids, let’s not fight, both the DHS and the Intercept can be up to no good at the same time.
Uber Eats fees have always been ridiculous when added up, here in New York I find it absurd that anyone would use them over some combination of Caviar (for my purposes by far the best), DoorDash, Seamless and Slice. They don’t put it all into the price because of how it would look if they did. FTC is looking into such practices but as of writing this their website is down.
Good News, Everyone
Is it a good list? Hard to say. I’m inclined to say not especially. There are some glaring omissions. My favorite part of looking at this year’s list is seeing ‘previous position’ with a constant stream of ‘new entry’ and ‘down 66 places’ and so on. For example, their previous list didn’t even have Day of the Tentacle (1993) on it, now it’s rightfully (or at least highly defensibly) at #14. No one’s list will be correct or complete. What such lists do provide are good sources of ideas. In general, choosing from a sampling of classic games people love is better than looking for New Hotness.
In the short term, Grand Theft Auto games decrease crime. Long term direction is less obvious for this particular game, neither result would surprise me. For games in general, I am reasonably confident they reduce crime unless one is able to tell a ‘this crowds out work which makes us poorer which leads to crime’ long term story.
Scott Alexander reviews The Malleus Maleficarum. Felt like good old school Scott.
Zohar Atkins doing a weekly reading of Genesis on Threadable, free to all, starting next week. Self-recommending if relevant to your interests.
While I Cannot Condone This
What Stanford seems to hate less is ‘random person lives in dorm basement for a year and stalking a student for much of that time until caught taking a television.’
There is something great about adjacent countries having different answers by multiple orders of magnitude here. This is strong Czeck and Hungarian dominance.
What’s the problem? It’s all OK, Google, makes perfect sense.
Scott Alexander wrote a post on the Jhanas I found somewhat interesting, and there was a fun comments highlight post after largely thanks to a large faction of commenters who disbelieve the existence of Jhanas where one can experience pure pleasure. In the section ‘is this good?’ there is this.
On the other hand, Peter Gerdes:
As a hedonic utilitarian I think it might actually be immoral for these individuals to not spend much more time in Jhana.
The fact that it's only themselves they are hurting doesn't really seem to change the matter. Unless their job/life is so beneficial to others as to make up for the difference they are choosing to fill the world with more suffering and less pleasure than they could by spending more time in that state.
Also I gotta get my act together and quit lazying out on meditation.
I am very much not a hedonic utilitarian. If I was one, I would notice that this was leading to this kind of preference falsification, and that this was a problem for my belief system. If you think ‘failure to sit around experiencing contentless mental pleasure’ is a moral failure, your moral system seems obviously broken. Sure, it might be a tactical error to fail to sit around experiencing that. Not my call. It’s definitely not an obligation. Peter, on the other hand, is saying that it is, independent of whether it is what anyone involved in such experiences wants.
Whereas I would say, it is important to ensure that your moral system is not making this mistake. I am not sure if it is worthwhile or not to work to gain access to the Jhanas. I do have confidence that the vast majority of people with this access, who choose not to spend that much time in them, are being wise, and that one should not require any hacks to recognize this.
Magic: The Gathering update, all this ‘excellence’ is bad, you know it’s bad, right?
(For those who are asking ‘huh?’ this is saying that Magic: The Gathering keeps making cards that are more and more powerful for what they cost to use, and while this makes the individual cards cool and powerful when you first see them, it is long term bad for the game.)
There is not that much support for democratic norm violations in a new survey, and Republican support for them in the survey is actually lower than Democratic support. The only exception is widespread Democratic support for using executive orders.
However it seems like this survey asks the wrong questions. It is asking symmetrically about norm violations. I don’t care if you are for a norm that baseball players don’t steal signs. I care if you object when your team steals signs from their team.
Thus, when considered abstractly and fairly, ‘reduce polling places’ is unpopular.
Then the officials go out and reduce polling places in out-party areas. A lot. Do people then go around saying “look at this 94% unpopular policy and norm violation, I should tell their base about this they will be mad?” Of course not. No one thinks that would work.
Wealth taxes seem to cause very large adjustments in reported wealth, including some combination of housing prices, migration and evasion.
Who pays for your credit card rewards? People who get tempted by rewards to overspend and then not pay off their balances, resulting in a regressive redistribution of $15 billion. And here I was worried it was getting baked into prices. It turns out, no, it’s the good old ‘some people who use credit cards do not realize they should not keep balances on their credit cards.’
Losing a family member is bad for mutual fund returns and for CEO performance, acting largely through risk aversion. Which suggests CEOs should take more risk and also that CEO risk taking is highly valuable and thus you should be willing to pay in order to get it. On the mutual fund side, the ability to underperform in this way implies the ability to overperform by doing the opposite, and once again that the efficient market hypothesis is false.