One of Aaronson's arguments in the article boils down to the idea that running a full universe simulation (without cheating) on a universe with the same physics as ours may just not be possible computationally; it seems physically plausible that you need a universe to compute a universe.
If that's true, then the simulators would need to be running in a different kind of universe than ours... in which case "ancestor simulation" doesn't really make sense.
Well, yes, quantum physics says that's true. Our universe is the minimal requirements necessary to run our universe. There could be tricks, for instance maybe you only simulate at fine grain the universe near an observer. But still, to simulate our granularity you need a simulation of equal (or greater) granularity. There's the trick, it's conceivable that the simulator exists in a universe with more dimensionality.
I can't speak for FaceTime, but I've found much better success with all of these platforms if everyone in the meeting is wearing headphones. A lot of the latency is from echo cancellation, and that turns off if there's no echo.
At my last company, I let everyone on my team expense comfortable headphones with a boom mic, and it made a huge difference on Google Meet. I've found Zoom to be similar. I can't speak to Slack's latency issues.
I'm also a big fan of Tuple[0] (the pair coding app) for extremely low latency screen sharing / pair coding and that was a huge advantage too.
Eh. I studied math at an elite university with a lot of people who ended up at these secretive quant hedge funds. They're undeniably brilliant, and as per GP also extremely motivated and hard working (brilliance alone doesn't cut it!).
They're also heavily dependent on collaboration with others, availability of data, the existence of the right tools, and so on. Unlike in comic books, raw intelligence alone doesn't get you very far. But put a bunch of very smart people together for a decade+ with a lot of money on their hands and you can definitely make magic happen.
Not OP but in my experience it’s more likely to Just Work with fewer glitches and at a higher video resolution. I’d like to get away from Zoom but with consulting work it’s the only thing I’ve found that typically works with clients on the first try and with the fewest artifacts.
I have not had the same luck with the video calls as you. People have mentioned “new Teams” which is perhaps better and more stable. When I used Teams last year it was not at all comparable to Zoom
It’s weird to have been working on a paper for almost a year and have it launch into this environment, but uptake has been good. My hope is that we will continue to see more nuance around different kinds of alignment risks in the near future. There’s a wide spectrum between biased statistical models and paperclip maximizing overlords, and lots bad but not existentially catastrophic things for the public to want to keep a pulse on.
Thanks! Looks like good work. I hope this idea continues to get traction:
> In some sense, we’re already living in a world of misaligned optimizers
I understand this is an academic paper given to nuance and understatement, but for any drive-by readers, this is true in an extremely literal sense, with very real consequences.
Higher interest rates mean that investors can earn money by buying bonds, which are lower risk. Riskier investments (stocks, real estate, cryptocurrencies) are less attractive to investors as a result, unless they can provide even more return to compensate for the risk, which is unlikely.
I can't speak for the Japanese restaurants, but I had surprisingly delicious salmon nigiri just from a Seattle grocery store chain. Great flavor, and very inexpensive. I think it was a platter of 12 nigiri for what I'd expect to pay for a supermarket salmon maki in New York. This would have been at a QFC around 2015.
It sounds like you have a good path on your technical skills, so I'm going to suggest something else: take some humanities classes. Pick topics outside of technology that interest you, and learn how to communicate about them. Take as many seminar discussion and writing-heavy classes as you can. Hopefully they are there even at a technical university.
From a career perspective, programming skill is relatively easy to come by. Programming AND speaking AND good writing will put you on a better career path. You won't get stuck after a job or two when you know the tools of the trade but not how to handle things outside the compiler.
But also, just as a person who has to make it in this world, it's good to have more ways of understanding people and systems around you. Humanities students are enriched by taking science and math classes, and getting a new way of seeing things. Technical undergrads who learn to tolerate ambiguity and learn some history are rounder humans.
Thanks. Unfortunately picking classes is not something that happens in the university system in my country, and even if it was, I'm in strictly technical one, so I have to pass on that one.
But thank you, good writing is certainly a skill that is going to be useful - good idea to work on that.
If you liked this article, the book Lost in Thought[0] also references Bennet and goes into more depth on why spending some of your time learning for its own sake leads to a more fulfilling life.
I’d love more content like this on hn. Done right, tech jobs can afford a lot of leisure, and we don’t have generally good guidance from our culture on how to spend leisure time in a fulfilling way.
Thank you for this recommendation!
I’d love to see more content like this too. Also I’d love to see more exploration of how those of us with young kids or other caring responsibilities can get a better balance and still keep pursuing learning for its’ own sake.
Look all the way over at "Second dose", the age-standardized mortality rate is 1.1 per 100,000 vs 5.4 per 100,000 for the unvaccinated. 5x protection against death.
Table 1, in the linked XLS file for data through 24 September. You can see that it was originally closer to 0.1 (50x protection against death) but it's gone up to 1.1 over the last few months, I'm guessing as immunity has started to wear off.
Something else to consider is that while a person may die while having COVID-19, it does not mean that is all that they died from.
People in general may have several health problems, which is especially likely for an 80+ year old person.
Some healthcare activities have also been postponed since the outbreak, by the patients and also the hospitals, which may have led to less healthy people.
This is indeed what Table 8 shows. The vaccinated skew elderly, and while slight more vaccinated died, they were at a much lesser risk of dying for COVID. According to that data, over 32 times less likely for their age.
If that's true, then the simulators would need to be running in a different kind of universe than ours... in which case "ancestor simulation" doesn't really make sense.