I’d love to enter a competition like this.
Hmash
Karma: 636
There’s also the possibility that a maximum doesn’t exist.
Suppose you had a one-shot utility machine, where you simply punch in a number, and the machine will generate that many utils then self-destruct. The machine has no limit in the number of utils it can generate. How many utils do you select?
“Maximise utility” has no answer to this, because there is no maximum.
In real life, we have a practically infinite number actions available to us. There might be a sense in which due to field quantisation and finite negentropy there are technically finite actions available, but certainly there are more actions available than we could ever enumerate, let alone estimate the expected utility for.
In practice, it seems like the best way to actually maximise value is just to do lots of experimental utility-generating projects, and greedily look for low-effort, high-reward strategies.
I had a crack at doing the Fermi Paradox calculations using vanilla JS for benchmarking. Took maybe 5 minutes to build reusable probabilistic estimation functions from scratch. On that basis, it doesn’t look to me like it would be worth the effort of learning a new syntax.
However, what took me almost all day was trying to get a nice visualisation of the probability distribution I came up with. I would like to be able to zoom and pan, hover over different x-values to get the PDF or CDF as a function of x, and maybe vary model parameters by dragging sliders. IMO, this is the real advantage of a probabilistic reasoning web-app.
After like 6 hours, I came up with a janky prototype which has zooming and a hover tooltip on a CDF.
Very messy code here: https://github.com/hamishhuggard/interactive-CDF/blob/main/fermi.html
PS: I hear QURI is hiring? Can I use this as a work trial? :P
Follow-up: Crowdsourced Criticisms
this one weird trick creates infinite utility
:D
Yeah, hard to know what to do with that. I’ll make it clear in the post that it is an acknowledged mistake that has been apologised for.
I have a few novel ideas about how to make infinite ethics problems go away (by solving or dissolving them, depending on your perspective), but they would take hours or days to write down. How valuable would it be for me to do this?
Third option:
I object to the very existence of this survey
No, infinite ethics is not a serious problem and doesn’t deserve criticism.
Yes, infinite ethics is a serious problem and deserves criticism.
Survey: do you think infinite ethics is a serious problem?
If you agree that EA should:
Be more accomodating of people who want to work on climate change
Please upvote this comment (see the last paragraph of the post).
I ended up significantly reworking the section. Any feedback on the new version?
Thank you and good points.
but then the policy suggestions seem to endorse every criticism. (Maybe you do agree with all of them?)
I guess what I was attempting was to steelman all of the criticisms I heard. Trying to come up with a version I do agree with.
I will change the title to “Be more respectful of climate change work”
Great, thank you.
I will update the bullet point with a link to your comment.
If you agree EA should:
Have more quiet spaces at conferences
Please upvote this comment (see the last paragraph of the post).
If you agree EA should:
Have better mental health support
Please upvote this comment (see the last paragraph of the post).
If you agree EA should:
Have more money transparency
Please upvote this comment (see the last paragraph of the post).
Lol. Not bad for 60% joking.
PS, here’s the code actually deployed: https://hamishhuggard.com/misc/fermi.html