Zack_M_Davis

Zack_M_Davis' Comments

A Problem With Patternism

the same question Yudkowsky uses in his post on cryonics in the sequences, although I can't find a link at the moment

You may be thinking of "Timeless Identity". Best wishes, the Less Wrong Reference Desk

[Site Meta] Feature Update: More Tags! (Experimental)

I want a language or philosophy of language tag (examples: "37 Ways Words Can Be Wrong", "Fuzzy Boundaries, Real Concepts", some of my published and forthcoming work).

I want a disagreement tag (examples: "The Modesty Argument", "The Rhythm of Disagreement", a forthcoming post).

Takeaways from safety by default interviews

AI researchers are likely to stop and correct broken systems rather than hack around and redeploy them.

Ordinary computer programmers don't do this. (As it is written, "move fast and break things.") What will spur AI developers to greater caution?

Benito's Shortform Feed

Alternatively, "lawful universe" has lower Kolmogorov complexity than "lawful universe plus simulator intervention" and thereore gets exponentially more measure under the universal prior?? (See also "Infinite universes and Corbinian otaku" and "The Finale of the Ultimate Meta Mega Crossover".)

Benito's Shortform Feed

Why appeal to philosophical sophistication rather than lack of motivation? Humans given the power to make ancestor-simulations would create lots of interventionist sims (as is demonstrated by the populatity games like The Sims), but if the vast hypermajority of ancestor-simulations are run by unaligned AIs doing their analogue of history research, that could "drown out" the tiny minority of interventionist simulations.

Can crimes be discussed literally?

If X are making claims that everyone knows are false, then there's no element of deception

"Everyone knows" is an interesting phrase. If literally everyone knew, what would be the function of making the claim? How do you end up with a system that wouldn't work without false assertions, and yet allegedly "everyone" knows that the assertions are false? It seems more likely that the reason the system wouldn't work without false assertions, is because someone is actually fooled. If the people who do know are motivated to prevent it from becoming common knowledge, "It's not deceptive because everyone knows" would be a tempting rationalization for maintaining the status quo.

When to Donate Masks?

if you donate masks today I expect them to be used much more quickly than if you wait and donate them when things are worse.

But this (strategically timing your donation because you don't expect the recipient to use the gift intelligently if you just straightforwardly gave when you noticed the opportunity) is kind of a horrifying situation to be in, right? If you can see the logic of the argument, why can't hospital administrators see it, too?—at least once it's been pointed out?

SARS-CoV-2 pool-testing algorithm puzzle

This reminds me of that cute information-theory puzzle about finding the ball with the different weight! I'm pretty dumb and bad at math, but I think the way this works is that since each test is a yes-or-no question, we can reduce our uncertainty by at most one bit with each test, and, as in Twenty Questions, we want to choose the question such that we get that bit.

A start on the simplest variant of the problem, where we're assuming the tests are perfect and just trying to minimize the number of tests: the probability of at least one person in the pool having the 'roni in pool size S is going to be , the complement of the probability of them all not having it. We want to choose S such that this quantity is close to 0.5, so . (For example, if P = 0.05, then S ≈ 13.51.)

My next thought is that when we do get a positive test with this group size, we should keep halving that group to find out which person is positive—but that would only work with certainty if there were exactly one positive (if the "hit" is in one half of the group, you know it's not in the other), which isn't necessarily the case (we could have gotten "lucky" and got more than one hit in our group that was chosen to have a fifty-fifty shot of having at least one) ...

Partial credit???

Load More