once McGonagall sees Harry disappear, he can’t undo it.
Sounds like UDT might be applicable here. Here’s a time-traveling version of Counterfactual Mugging:
Harry appears to McGonagall and tells her, “If you give me 1 Galleon now, I’ll go back in time and hand you 100 Galleons an hour ago.” Suppose McGonagall does not recall being handed 100 Galleons an hour ago. What should she do?
Here’s my analysis. Suppose McGonagall decides not to give Harry 1 Galleon, then there are two possible consistent timelines for this universe. One where McGonagall gets 100 Galleons, and one where she doesn’t. How does the universe “choose” which one becomes reality? I don’t know but let’s say that the two possibilities have equal chance of being true, or get equal amount of “reality juice”.
Given the above it seems clear that McGonagall would prefer to have pre-committed to “give 1 Galleon even if not handed 100 Galleons an hour ago” since that would make the “not get 100 Galleons” timeline inconsistent. I think that’s also UDT’s output (although I haven’t written down the math to make sure).
ETA: I didn’t follow the previous discussion closely, so this might not apply at all to it. Hopefully, in that case the above is of interest in its own right. :)
Seems straightforward to me. McGonagall knows that she does not recall being handed 100 Galleons an hour ago, so the three states of the world with high probability are: 1) She is not in a universe where she will hand Harry 1 Galleon, 2) She is in a universe where she hands Harry 1 Galleon and Harry breaks the agreement, or 3) She is in a universe where she hands Harry 1 Galleon and Harry keeps the agreement in a way that leaves her unable to recall this happening. By not handing Harry a Galleon, she will ensure that she is in universe 1. By handing Harry a Galleon, she will find herself in universe 2 or 3. She should therefore give Harry a Galleon if she judges it less than 99 times more likely that Harry will break the agreement than fulfil it in a way consistent with her experience.
As Harry has access to a time machine, he doesn’t need to decide to give her 100 Galleons before he gets the 1 Galleon, so the situation is quite different to one based on predicting her actions, as Omega does in the Counterfactual Mugging. Rather it has most of the properties of the forward-time version of the gambit: “If you give me 1 Galleon now, I’ll hand you 100 Galleons in one hour”, except that McGonagall has a big piece of evidence that the promise will be broken, namely that she doesn’t remember it being kept.
ETA: I didn’t follow the previous discussion closely, so this might not apply at all to it. Hopefully, in that case the above is of interest in its own right. :)
Vladmir and I agree with the applicability of UDT and have suggested time-travel-with-consistency is a good way to consider Newcomblike problems and the the decision theories that can handle them.
Suppose McGonagall does not recall being handed 100 Galleons an hour ago. What should she do?
Precommit, give him the Galleon, then reach in her bag to get the 100 Galleons. (She must have been Obliviated; otherwise, she would remember.)
You can actually get around a lot of the problems with time travel by taking advantage of the difference between observation and reality. For instance, if you see one of your friends die, you can go back in time, save him, then plant a fake double so you still have the same observations.
Though Godel was interested in time travel loops, that’s not the type of consistency that his Second Incompleteness Theorem discussed (it’s limited to formal axiomatized systems that can describe arithmetic).
It seems we can’t rely on setups, which include decision making inside time loop, because it is not clear who or what makes decisions (agent? consistency law? agent’s preferences?).
It seems I can’t rely on setups, which include decision making inside time loop
If you say ‘we’ you invite contradiction as your stance implies magical thinking about decision making which not many people would want to admit to. Decisions are not a special case. If gravity works you can build a gravity powered mechanical device that makes a decision.
Either you treat decisions exactly as you usually would or nothing works at all.
Can I be in the one where the ‘created information’ is a a freak quantum conversion of graphite into diamond and not the one where the created information is freak quantum effects making my neurons fire in a way that they normally wouldn’t?
You can be there, no problem. The problem is to setup everyting to get there. I’ve just considered the question of setting up, and found that I can’t say even what causes existence of time loop. Can you say what should I do to be sure that no matter what after 1.01 hour I will timeturn 1 hour back? Being the person that will do that is important, but it will not help if in 20 minutes someone will steal my timeturner.
Returning back to 6(grand)parent, I can say, that imposing any condition on a stable state of time loop (but initial conditions before time loop) seems unphysical. All in all, thought experiments with time loops are interesting, but their validity is highly questionable.
Eliezer can do that, because he is consistency law of Harry’s universe.
You can be there, no problem. The problem is to setup everyting to get there. I’ve just considered the question of setting up, and found that I can’t say even what causes existence of time loop. Can you say what should I do to be sure that no matter what after 1.01 hour I will timeturn 1 hour back? Being the person that will do that is important, but it will not help if in 20 minutes someone will steal my timeturner.
Go think about it some more from the perspective of an integrated physics that happens to include time travel of this kind and not as a special case. That seems to be confusing you.
Eliezer can do that, because he is consistency law of Harry’s universe.
Eliezer can have Harry become a Hari Krishna if he really wants. Or he could artificially select for extremely improbable quantum events either inside or outside of time loops. Neither of those would make for an especially good story in this instance.
A trolley is running out of control down a track. In its path are 5 people who have been tied to the track by a mad physicist. Fortunately, if trolley ran down different track, it would activate timeturner that would send it 10 seconds back in time and then trolly would flip a switch, which will lead the trolley down a different track to timeturner. And also a bystander runs to flip a switch.
Possible stable states.
Bystander stops and wait for trolley to appear from nowhere. Trolley kills 5.
Bystander stops and wait for trolley to appear from nowhere. Trolley appears from nowhere and flips the switch, then original trolley runs to timeturner and disappears.
Trolley appears from nowhere kills bystander, which almost managed to reach the switch, and flips the switch, then original trolley runs to timeturner and disappears.
Bystander appears from nowhere and flips the switch, then original bystander runs to timeturner and disappears. Trolley run past deactivated timeturner.
Bystander appears from nowhere shouts “watch your step!” and fails to flip switch in time, original bystander looks on timetraveller, missteps and falls, then manages to activate timeturner and disappears. Trolley kills 5.
Bystander appears from nowhere shouts “careful!” and flips switch, original bystander looks on timetraveller, missteps, but manages to not fall, then activates timeturner and disappears. Trolley run past deactivated timeturner.
Bystander appears from nowhere looks on trolley, shouts “damn! too late” and fails to flip switch in time, original bystander hearing that tries to run faster, missteps and falls, then activates timeturner and disappears. Trolley kills 5.
Bystander appears from nowhere, shouts “Shout, what I shouting now, it is very important!” and flips switch, original bystander activates timeturner and disappears. Trolley runs past deactivated timeturner.
Even in those rigid conditions stable states can be numerous without any improbable events. But yes, maybe you are right, because amount of information-from-nowhere must be limited by communication channel bandwidth (spoken or written speach, nonverbal actions of future-agent, agent’s memory throughput), thus rendering stable states in principle analyzable.
In a time loop do things still fall down? Is there gravity? And if I implement a decision making mechanism using a gravity powered system of gears and pulleys where is the magic thing that makes reason no longer apply?
Information from future will affect any kind of decision making. This information is a consequence of consistency law, it even doesn’t need to have cause.
Sounds like UDT might be applicable here. Here’s a time-traveling version of Counterfactual Mugging:
Harry appears to McGonagall and tells her, “If you give me 1 Galleon now, I’ll go back in time and hand you 100 Galleons an hour ago.” Suppose McGonagall does not recall being handed 100 Galleons an hour ago. What should she do?
Here’s my analysis. Suppose McGonagall decides not to give Harry 1 Galleon, then there are two possible consistent timelines for this universe. One where McGonagall gets 100 Galleons, and one where she doesn’t. How does the universe “choose” which one becomes reality? I don’t know but let’s say that the two possibilities have equal chance of being true, or get equal amount of “reality juice”.
Given the above it seems clear that McGonagall would prefer to have pre-committed to “give 1 Galleon even if not handed 100 Galleons an hour ago” since that would make the “not get 100 Galleons” timeline inconsistent. I think that’s also UDT’s output (although I haven’t written down the math to make sure).
ETA: I didn’t follow the previous discussion closely, so this might not apply at all to it. Hopefully, in that case the above is of interest in its own right. :)
Seems straightforward to me. McGonagall knows that she does not recall being handed 100 Galleons an hour ago, so the three states of the world with high probability are: 1) She is not in a universe where she will hand Harry 1 Galleon, 2) She is in a universe where she hands Harry 1 Galleon and Harry breaks the agreement, or 3) She is in a universe where she hands Harry 1 Galleon and Harry keeps the agreement in a way that leaves her unable to recall this happening. By not handing Harry a Galleon, she will ensure that she is in universe 1. By handing Harry a Galleon, she will find herself in universe 2 or 3. She should therefore give Harry a Galleon if she judges it less than 99 times more likely that Harry will break the agreement than fulfil it in a way consistent with her experience.
As Harry has access to a time machine, he doesn’t need to decide to give her 100 Galleons before he gets the 1 Galleon, so the situation is quite different to one based on predicting her actions, as Omega does in the Counterfactual Mugging. Rather it has most of the properties of the forward-time version of the gambit: “If you give me 1 Galleon now, I’ll hand you 100 Galleons in one hour”, except that McGonagall has a big piece of evidence that the promise will be broken, namely that she doesn’t remember it being kept.
Vladmir and I agree with the applicability of UDT and have suggested time-travel-with-consistency is a good way to consider Newcomblike problems and the the decision theories that can handle them.
Precommit, give him the Galleon, then reach in her bag to get the 100 Galleons. (She must have been Obliviated; otherwise, she would remember.)
You can actually get around a lot of the problems with time travel by taking advantage of the difference between observation and reality. For instance, if you see one of your friends die, you can go back in time, save him, then plant a fake double so you still have the same observations.
You can never know consistency, can never rely on it, otherwise you are inconsistent (2nd incompleteness theorem).
Though Godel was interested in time travel loops, that’s not the type of consistency that his Second Incompleteness Theorem discussed (it’s limited to formal axiomatized systems that can describe arithmetic).
I suspect Vladmir is considering the system in question here in a way that meets that description.
It seems we can’t rely on setups, which include decision making inside time loop, because it is not clear who or what makes decisions (agent? consistency law? agent’s preferences?).
All of the above. Decision making is just physics.
As I wrote earlier I am not yet convinced, that this is a kind of physics we can effectively reason of.
Consider correcting your claim to:
If you say ‘we’ you invite contradiction as your stance implies magical thinking about decision making which not many people would want to admit to. Decisions are not a special case. If gravity works you can build a gravity powered mechanical device that makes a decision.
Either you treat decisions exactly as you usually would or nothing works at all.
No. It’s not a decision making that is special. It is closed time loops. They possibly can “magically” create information from nothing.
Can I be in the one where the ‘created information’ is a a freak quantum conversion of graphite into diamond and not the one where the created information is freak quantum effects making my neurons fire in a way that they normally wouldn’t?
You can be there, no problem. The problem is to setup everyting to get there. I’ve just considered the question of setting up, and found that I can’t say even what causes existence of time loop. Can you say what should I do to be sure that no matter what after 1.01 hour I will timeturn 1 hour back? Being the person that will do that is important, but it will not help if in 20 minutes someone will steal my timeturner.
Returning back to 6(grand)parent, I can say, that imposing any condition on a stable state of time loop (but initial conditions before time loop) seems unphysical. All in all, thought experiments with time loops are interesting, but their validity is highly questionable.
Eliezer can do that, because he is consistency law of Harry’s universe.
Go think about it some more from the perspective of an integrated physics that happens to include time travel of this kind and not as a special case. That seems to be confusing you.
Eliezer can have Harry become a Hari Krishna if he really wants. Or he could artificially select for extremely improbable quantum events either inside or outside of time loops. Neither of those would make for an especially good story in this instance.
Ok. Almost pure physics.
A trolley is running out of control down a track. In its path are 5 people who have been tied to the track by a mad physicist. Fortunately, if trolley ran down different track, it would activate timeturner that would send it 10 seconds back in time and then trolly would flip a switch, which will lead the trolley down a different track to timeturner. And also a bystander runs to flip a switch.
Possible stable states.
Bystander stops and wait for trolley to appear from nowhere. Trolley kills 5.
Bystander stops and wait for trolley to appear from nowhere. Trolley appears from nowhere and flips the switch, then original trolley runs to timeturner and disappears.
Trolley appears from nowhere kills bystander, which almost managed to reach the switch, and flips the switch, then original trolley runs to timeturner and disappears.
Bystander appears from nowhere and flips the switch, then original bystander runs to timeturner and disappears. Trolley run past deactivated timeturner.
Bystander appears from nowhere shouts “watch your step!” and fails to flip switch in time, original bystander looks on timetraveller, missteps and falls, then manages to activate timeturner and disappears. Trolley kills 5.
Bystander appears from nowhere shouts “careful!” and flips switch, original bystander looks on timetraveller, missteps, but manages to not fall, then activates timeturner and disappears. Trolley run past deactivated timeturner.
Bystander appears from nowhere looks on trolley, shouts “damn! too late” and fails to flip switch in time, original bystander hearing that tries to run faster, missteps and falls, then activates timeturner and disappears. Trolley kills 5.
Bystander appears from nowhere, shouts “Shout, what I shouting now, it is very important!” and flips switch, original bystander activates timeturner and disappears. Trolley runs past deactivated timeturner.
Even in those rigid conditions stable states can be numerous without any improbable events. But yes, maybe you are right, because amount of information-from-nowhere must be limited by communication channel bandwidth (spoken or written speach, nonverbal actions of future-agent, agent’s memory throughput), thus rendering stable states in principle analyzable.
In a time loop do things still fall down? Is there gravity? And if I implement a decision making mechanism using a gravity powered system of gears and pulleys where is the magic thing that makes reason no longer apply?
Information from future will affect any kind of decision making. This information is a consequence of consistency law, it even doesn’t need to have cause.