Game 2 is structurally identical to Game 3, meaning my argument is structurally identical. Not only is structurally identical, it is literally identical. Why?
Because you set up the causal relations identically between the options a) and b), and the choices of me-me and other-me.
The only difference you made is the way that we signal whether we choose a) or b).
In Game 2, “pressing the b) buttom” is enacted by me picking up a red and a blue box.
In Game 2, “pressing the b) buttom” is enacted by me picking up a blue box.
It doesn’t matter how you “press the button”, how you choose a) or b). What matters are the consequences. In my response to Game 2, I never talked about the mechanism for selecting a) or b). As such, my response to Game 2 can be used for any game where a) and b) are the same choice, but the mechanism is different.
But I very much appreciate your games, because the point to the deeper TB argument. A lot of the time, us TBs will say, “the money is already in Box 2, or already not in Box 2.”
We lean on this physical intuition to help make our point. You have removed the whole element of money being in the boxes. Instead, the boxes are irrelevant, and you are just making a choice, and your choice is correlated with a past choice.
But TBism never relied on the money physically (not) being in Box 2. The crux of the TB argument is of the form:
The most decisive choice has already been made in the past, and we cannot change it now in the present (Phase 2). All we do is change our knowledge regarding what likely happened, because our current choice is correlated with the past choice. But knowledge \ne money
This argument is made more intuitive when the $1M is literally lying in Box 2, or literally not lying in Box 2. But it was never necessary, because whether or not the participant has won that $1M has nonetheless already been determined in the past.
Anyways, just to be absolutely clear that I have responded to Game 3, I will paste in Game 3’s rules below, and then my response to Game 2, re-appropriated for Game 3. You will see that with literally no changes, my response to Game 2 still holds. Then at the end of this, I will give you a variation on Newcomb’s Game to show you what I mean above.
Game 3
-
There is an empty red box and an empty blue box
-
You are cloned
-
You are asked to choose between either a) the red box or b) the blue box
-
The clone is asked to choose between either a) the red box or b) the blue box
-
The clone is destroyed
-
If you both chose (a) then you win $1,000,000
-
If you both chose (b) then you win $1,000
-
If you chose (a) and your clone chose (b) then you win $0
-
If you chose (b) and your clone chose (a) then you win $1,001,000
So, again I assume neither of us know we are the clone.
Now, using the same argument as for Game 1, I am assuming no inherent, ethical care for other-me. I only care about what is beneficial to me-me, so any care (if at all) for other-me would merely be an extension of my care for me-me.
So, do I selfishly care about other-me in this scenario?
Well, if I care about other-me, then other-me probably cares about me too. But, one of two things is true:
-
Other-me has already made their decision, and I cannot impact it.
-
Other-me will make their decision after me, which means I am the clone and will die, meaning I don’t care about anything anyways.
So, caring about other-me possibly cannot help me-me. Therefore, in this scenario, I neither have selfless nor selfish reasons to care about other-me. That covers any and all reasons to care about other-me, which means I do not need to re-create my matrix above. (I only made it as a matter of clarity in the first section).
So, the possibility space is reduced to this: am I the clone or the non-clone?
If I am the clone, then my actions do not matter to me-me.
If I am not the clone, then it matters. So, I know my clone chose either a) or b), though I have yet to determine the likelihood of which. I can nonetheless evaluate how much I win in either scenario:
My clone picked a) [Option 1]
If I pick a), I get $1M. If I pick b), I get $1,001,000.
My clone picked b) [Option 2]
If I pick a), I get $0. If I pick b), I get $1000.
In both scenarios, me-me picking b) gives me more money than me-me picking a).
We say that me-me picking b) dominates. So, on first thought, I should pick b).
But, my clone would likely think the same! This increases the likelihood that Option 2 is correct. So, does that change anything? Well, choice b) is still better in Option 2, so no… no it does not change anything.
If I could affect the past choice of my clone, then I would want to make it pick a). But I cannot affect the past, I cannot affect the clone. My behavior is indicative of the likely behavior of the past clone. So, if I behave selfishly by picking b), then that is bad news to me-me.
But if I instead go for a), I am not actually changing the past behavior of the clone. If I go for a), I am simply picking the option that will give me less money because it gives me better news.
I do not optimize the niceness of the news my behavior gives me. I do not optimize the optimism of the knowledge I gain. I optimize the actualities around me, I optimize the present/future because I cannot optimize the past.
This is my nature. There are games that punish this nature, and there are games that do not. We do not define rationality based on some games. We define rationality based on truth. We use that truth to create a rational decision theory. Believing in that rational decision theory will usually outperform believing the irrational theory, because alignment with truth happens to usually help in the real world.
And as I’ve argued in an earlier post, these games do not punish me for applying CDT. These games punish me for believing in CDT.
They punish me for my nature/beliefs in the past, and that is why you see non-CDT adherents winning over me in these games. If you look across all real games, then the picture will be different. Truth matters. CDT follows the truth by upholding the unidirectionality of causality, thus properly differentiating between correlation and causation. Applying this will always be better. Believing in this, not always. Sometimes we are punished for our beliefs, even if they are better/rational.
Now, if I have a rational belief, and I know I am about to go into a situation where that belief will harm me, then (if it’s worth it), it will be rational to let go of the rational belief. This is rational adaptation, which I covered in the OP.
Now, rationally letting go of a rational belief is impossible, unless you change the very pre-conditions for that rational belief. This is what I do with my contract in Version 2 of the Newcomb Game. In the Iterated Version, I do not need to do anything extra, as the pre-conditions are already there. OB-ing is automatically rational in the Iterated Version of the Newcomb Game. No belief of mine needs to be rationally changed in that case.
The equivalent of Version 2 in your Game 2 would be somehow colluding with my clone before we face our dilemmas. But, a mere agreement would not be enough. Mere agreements are only rationally effective IF you will continue to interact with the other party: ie, there needs to be iteration. This is why the business world somewhat functions despite having plenty of psychos.
But in Game 2, one of us will die. There is no iteration, and so there is no rational, selfish reason for adhering to a mere agreement.
We would need to involve a third party and make a contract with them, such that they would sufficiently penalize the survivor for not adhering to the agreement. This would make it selfishly rational to honor the agreement to pick a).
But again, this requires collusive pre-planning! There is no such thing here. All there is, is a past I cannot affect.
So, I pick option b).
A Variation on V1 of the Newcomb Game
We’ve got all the same rules as V1 of the Newcomb Game, except for this:
When the predictor predicts that Bob is an OB, the predictor does not put $1M in Box 2. Instead, the predictor sends $1M into Bob’s bank account.
Bob is prevented from getting a notification of this, of course. He has no idea.
Bob then enters the room, and there is simply $1000 on the table. The predictor explains how it made its prediction about his behavior yesterday, and if the predictor predicted he was an OB, there is an extra $1M in his bank account. If not, there is not.
Of course, in this scenario, being an OB means not grabbing the $1000. Being a TB means grabbing that $1000. The only difference is your means of signalling your choice of OBism v. TBism, and the fact that the $1M has not been physically put into a box. Instead, it’s been put into a bank account.
So, Bob, do you grab the $1000 lying on the table?