Anthropic reasoning only goes this far. Even if I accept the silliness in which zillion of Earths are destroyed every year for each one that survives… the other planets in the solar system could also have been destroyed. And the stars and galaxies in the sky would all be devoured by now, no? And no anthropic reasons would prevent us from witnessing that from a safe distance.
Here’s a fun game: Try to disprove the hypothesis that every single time someone says “Abracadabra” there’s a 99.99% chance that the world gets destroyed.
Here’s a fun game: Try to disprove the hypothesis that every single time someone says “Abracadabra” there’s a 99.99% chance that the world gets destroyed.
We haven’t been anthropically forced into a world where humans can’t say “Abracadabra”.
Here’s a fun game: Try to disprove the hypothesis that every single time someone says “Abracadabra” there’s a 99.99% chance that the world gets destroyed.
We haven’t been anthropically forced into a world where humans can’t say “Abracadabra”.
Oh, but a non-trivial number of people have mild superstitions against saying “Abracadabra”. Does this not constitute (weak) anthropic evidence?
This is totally testable. I’m going to download some raw quantum noise. If the first byte isn’t FF I will say the magic word. I will then report back what the first byte was.
Initially this was anthropic evidence for normality, until people would have had time to replicate the experiment. Suppose the word was that dangerous, and the first byte had been FF. By now, all the people replicating the experiment have destroyed those universes. Only the universes where the experiment failed to show FF on the first try are still around.
At what point would you have accepted that saying “Abracadabra” does destroy the world? How would you have felt about that? And what service have you been using? I only know about random.org. Thanks.
ETA:
HotBits generates random numbers from radioactive decay.
I used this one. After two FFs I would have decided I was in a simulation which some Less Wrong poster had set up post-singularity to screw with us. Those kind of Cartesian Joker scenarios are way more probable than “Abracadabra” destroying the world…
Just two FFs? That doesn’t seem all that improbable even forgetting all thought of world destruction. After about 100 FFs I would suspect that there was a problem with my experimental procedure (eg. internet quantum byte source broken). That too would be testable. (“I’m not going to say Abracadabra this time. FF? FF? Now I am. FF? FF?)
Well two FFs by chance is 1 in 65536. And my prior for “I’m in a simulation” isn’t that low. You’re right about the service being broken or fraudulent and really right about needing to test what happens if I don’t say Abracadabra. But you definitely don’t have to wait for 100 FFs!
Well two FFs by chance is 1 in 65536. And my prior for “I’m in a simulation” isn’t that low.
That isn’t the number to consider here. The relevant prior is “I’m in a simulation and this particular simulation involves the abracadabra trick”. That number is quite a bit lower!
You’re right about the service being broken or fraudulent and really right about needing to test what happens if I don’t say Abracadabra. But you definitely don’t have to wait for 100 FFs!
True enough. I estimate that I’d start testing after 4 or 5. :)
That isn’t the number to consider here. The relevant prior is “I’m in a simulation and this particular simulation involves the abracadabra trick”. That number is quite a bit lower!
Yeah. Hmm. I don’t really have a stable estimate of that probability. Of course, it’s not like like I would have stopped after two trials, but at that point I’m poring myself a drink. Worth noting that by coming up with the hypothesis I drastically increased its probability and then by mentioning it here I increased it’s probability even further.
I estimate that I’d start testing after 4 or 5. :)
Would you mind attempting to narrate any internal dialog you’d imagine yourself having after the 3rd? Lol.
And no anthropic reasons would prevent us from witnessing that from a safe distance.
I accept this counter-argument.
Try to disprove the hypothesis that every single time someone says “Abracadabra” there’s a 99.99% chance that the world gets destroyed.
This is unlikely because it is wildly incompatible with everything we know about physics, not because we have never observed it to happen. It is unlikely because it has an extremely low prior probability, not because we have any (direct) evidence against it.
On the “abracadabra” example? The overwhelming majority would come from the possibility that any time anything whatsoever happens the world is “destroyed”, for some weird, maybe anthropic use of the word “destroyed” I don’t understand compatible with me still being here.
If we limit it to “abracadabra” and nothing else, that’s complex enough that < 1/trillion just picking it out of hypothesis space (lots of combinations of sounds that could destroy the world, lots of things that aren’t combinations of sounds).
Just the world? Well, all you need is a good rocket ship so you aren’t on it anymore, and take a look.
If you mean destroy the MW branch in which it’s said, then Nick Tarleton’s answer works—that rule would make the choice to say ‘Abracadabra’ far smaller in probability than saying similar things that don’t destroy the world. People saying that one thing would be greatly suppressed relative to, say, “Alakazam” or “Poof” or “Presto Change-o”, and it would quickly leave the lexicon.
Perhaps rather than just causing a black hole, it causes a tear in space-time that expands at the speed of light. By the time you see it, you’re already dead.
Of course, there’s still the fact that early worlds would be weighted much more heavily, so this is probably about the first instant that you exist. And there’s the fact that, if that’s true, the LHC wouldn’t decrease the expected lifetime of the world by a noticeable amount.
Perhaps rather than just causing a black hole, it causes a tear in space-time that expands at the speed of light. By the time you see it, you’re already dead.
I feel vaguely disapproving of anthropic reasoning when it rewards elaborate and contrived scenarios over simpler ones with similar characteristics.
Anthropic reasoning only goes this far. Even if I accept the silliness in which zillion of Earths are destroyed every year for each one that survives… the other planets in the solar system could also have been destroyed. And the stars and galaxies in the sky would all be devoured by now, no? And no anthropic reasons would prevent us from witnessing that from a safe distance.
Here’s a fun game: Try to disprove the hypothesis that every single time someone says “Abracadabra” there’s a 99.99% chance that the world gets destroyed.
We haven’t been anthropically forced into a world where humans can’t say “Abracadabra”.
Oh, but a non-trivial number of people have mild superstitions against saying “Abracadabra”. Does this not constitute (weak) anthropic evidence?
Warning, spoiler alert:
Abracadabra.
You’ve just murdered 99.99% of all Earths. Our Everett branch survived for anthropic reasons.
This is totally testable. I’m going to download some raw quantum noise. If the first byte isn’t FF I will say the magic word. I will then report back what the first byte was.
Update: the first byte was 1B
...
Abracadabra.
Still here.
Initially this was anthropic evidence for normality, until people would have had time to replicate the experiment. Suppose the word was that dangerous, and the first byte had been FF. By now, all the people replicating the experiment have destroyed those universes. Only the universes where the experiment failed to show FF on the first try are still around.
Which means we have to cut down on the worlds where FF didn’t happen. Say it with me everyone.
Abracadabra, Abracadabra, Abracadabra, Abracadabra, Abracadabra, Abracadabra...
If everyone who reads this comments says the word say, thirty times, we should be good, right?
At what point would you have accepted that saying “Abracadabra” does destroy the world? How would you have felt about that? And what service have you been using? I only know about random.org. Thanks.
ETA:
HotBits generates random numbers from radioactive decay.
QRBG Quantum Random Bit Generator
I used this one. After two FFs I would have decided I was in a simulation which some Less Wrong poster had set up post-singularity to screw with us. Those kind of Cartesian Joker scenarios are way more probable than “Abracadabra” destroying the world…
Just two FFs? That doesn’t seem all that improbable even forgetting all thought of world destruction. After about 100 FFs I would suspect that there was a problem with my experimental procedure (eg. internet quantum byte source broken). That too would be testable. (“I’m not going to say Abracadabra this time. FF? FF? Now I am. FF? FF?)
Well two FFs by chance is 1 in 65536. And my prior for “I’m in a simulation” isn’t that low. You’re right about the service being broken or fraudulent and really right about needing to test what happens if I don’t say Abracadabra. But you definitely don’t have to wait for 100 FFs!
That isn’t the number to consider here. The relevant prior is “I’m in a simulation and this particular simulation involves the abracadabra trick”. That number is quite a bit lower!
True enough. I estimate that I’d start testing after 4 or 5. :)
Yeah. Hmm. I don’t really have a stable estimate of that probability. Of course, it’s not like like I would have stopped after two trials, but at that point I’m poring myself a drink. Worth noting that by coming up with the hypothesis I drastically increased its probability and then by mentioning it here I increased it’s probability even further.
Would you mind attempting to narrate any internal dialog you’d imagine yourself having after the 3rd? Lol.
“Um. WTF? Is this even working?”
(Yes, since the test is so trivial I might even click through a test after 2. I just wouldn’t start suspecting modded sims.)
Really?
Well chance is 1 in 65,536. Is there some hypothesis I’ve neglected?
The person running the qrng server decided to screw with you.
Damn!
I accept this counter-argument.
This is unlikely because it is wildly incompatible with everything we know about physics, not because we have never observed it to happen. It is unlikely because it has an extremely low prior probability, not because we have any (direct) evidence against it.
I should like to know Yvain’s prior on this.
On the “abracadabra” example? The overwhelming majority would come from the possibility that any time anything whatsoever happens the world is “destroyed”, for some weird, maybe anthropic use of the word “destroyed” I don’t understand compatible with me still being here.
If we limit it to “abracadabra” and nothing else, that’s complex enough that < 1/trillion just picking it out of hypothesis space (lots of combinations of sounds that could destroy the world, lots of things that aren’t combinations of sounds).
Just the world? Well, all you need is a good rocket ship so you aren’t on it anymore, and take a look.
If you mean destroy the MW branch in which it’s said, then Nick Tarleton’s answer works—that rule would make the choice to say ‘Abracadabra’ far smaller in probability than saying similar things that don’t destroy the world. People saying that one thing would be greatly suppressed relative to, say, “Alakazam” or “Poof” or “Presto Change-o”, and it would quickly leave the lexicon.
Indeed—none of us would have ever heard it.
Perhaps rather than just causing a black hole, it causes a tear in space-time that expands at the speed of light. By the time you see it, you’re already dead.
Of course, there’s still the fact that early worlds would be weighted much more heavily, so this is probably about the first instant that you exist. And there’s the fact that, if that’s true, the LHC wouldn’t decrease the expected lifetime of the world by a noticeable amount.
I feel vaguely disapproving of anthropic reasoning when it rewards elaborate and contrived scenarios over simpler ones with similar characteristics.
There are some interesting replies here.