If you look at the actual numbers, it’s fair to say people from all walks choose either option. A majority of people in this field or that field may choose one option, but a sizable fraction of all of those fields also choose the other option. It’s not like a total landslide from, say, decision theorists. It’s a majority, sure, but you can’t just wholesale say “see? 60 percent? That means no experts agree with one boxing.”
Reading a bit, it seems they conflate phase 1 and phase 2 sometimes. Like from the paper:
The standard formulation of EDT prescribes one-boxing (Gibbard and Harper
1978).2 Evidential decision theorists reason: “It would be good news to learn that
I had left the $1,000 behind; for then, with 99% probability, the predictor will have
filled the opaque box with $1,000,000.” Reasoning thus, EDT agents will be quite
predictable, the predictor will fill the box, and EDT agents will reliably walk away
$1,000,000 richer.
[…] (presentation of two-boxing)
If the story ended there, then we might conclude that EDT is the correct theory of how to maximize one’s utility
The problem here is that that you should differentiate between the choice in front of the boxes and the strategy before the game. Otherwise, you infer from “picking one box” that the player is predictible or susceptible to one-box or will get the million but that’s not true. What makes the player able to get the million has nothing to do with the choice in front of the boxes.
Yes. This is why Phase 1 and Phase 2 are so key. The whole misunderstanding stems from this simple conflation. It drives me insane. I am planning on drawing up a visual diagram of the timeline today. And some other diagrams. If that doesn’t work, I will give up.
I may have misrepresented it but I was just trying to be concise. Yeah, some philosophers even specialized will say OB. The problem (for the poster) is the difference in the ratios.
I personally don’t care if philosophers prefer OB or not. I wouldn’t use that as an argument for TB (and I am not saying the poster does).
That isn’t a conflation of phase one and phase two. In the problem as it is, there’s just Newcombs Problem. Certain people like to introduce phase ones or phase twos for whatever reason, but to everyone else there’s just Newcombs Problem. They aren’t confused by phases because general discussions of the problem don’t have phases
The thing is that, when you differentiate, you realize even CDT (TB) will say to be a OB in phase 1 and a TB in phase 2.
And if you believe phase 1 and phase 2 are strongly linked such that you need to have the “same strategy” then again TB will readily admit that from the beginning of phase 1 to the end, you should OB.
Yes, there are phases. But you are right that in Version 1, you are not consciously playing Phase 1. Furthermore, Phase 1 is not mentioned in the Newcomb Game.
But it is nonetheless there. It is a trivial inference. And when you start looking at it, you understand everything.
Phase 1 is defined as all the time before the predictor fills the boxes. Phase 2 is defined as the time after.
Would you disagree that these periods exist?
If reasoning about phases helps you think about the problem, fine, but interpreting other people’s writings to be about phases when they don’t mention phases is… just incorrect.
CDT loses you money in the Newcomb Game V1, if you look across all phases. (CDT wins in Phase 2 though, always)
I wouldn’t even say ‘CDT loses’ more like ‘believing in CDT’ pre-game loses but it wouldn’t be CDT itself losing as there is no decision pre-game in V1. If there is a decision then CDT would say you should not “believe in CDT” pre-game.
A CDT theorist would say you should be that, if possible. But I don’t think most CDT theorists think it is possible.
By the premise of the scenario, you are not likely to outsmart the predictor. So, don’t even try. Swtiching from OB to TB is trying to outsmart the predictor.
Instead, make yourself truly an OB in Phase 1. And if you are truly an OB in Phase 1, you will be an OB in Phase 2.
Now, personally, I cannot just make myself into an OB through thought alone. Because with no penalty imposed, TBing will always be best in Phase 2. And that is the problem. The logically best choice in Phase 2 is the globally worst choice, but you cannot play the whole game without playing Phase 2.
Therefore, you must give yourself a rational reason to be an OB throughout the whole game. I then refer you to the contract with the violation penalty fee F > 1000, forcing you to be an OB through mathematics.
I agree, I was just writing imprecisely.
CDT in application does not lose in Phase 1… because CDT is not being applied. There are no decisions in Phase 1 of Version 1.
But CDT loses in the sense that its adherents lose. And to point out this fact is pretty important, because it puts on the same page of the OBs with the why ain’cha rich argument. We understand that it is being CDT theorists in Phase 1, Version 1 of the game that loses us the game. Being on the same page is good for discussion.
But this does not make CDT irrational. That would be the tail wagging the dog.
That is, the Newcomb Game wagging the entire definition of rationality; and all other games and situations and truth itself can be damned!
1 Like
Being a CDT theorist makes me probably lose Phase 1 of Version 1 of the Newcomb Game.
Being a CDT theorist makes me definitely win Phase 2 of the Newcomb Game.
Being a CDT theorist means I uphold a truthful understanding of causality, and of correlation v. causation.
Being a CDT theorist helps me win in so many games.
Being a CDT theorist helps me win over EDT / FDT / LDT whenever not conflating correlation and causation is necessary to win.
Being a CDT theorist is worth it, despite the fact that I lose in some versions of the Newcomb Game, a hypothetical, contrived game literally designed to challenge our notions of rationality.
Can you say the same, OBs? With your theory of decision-making: if you contort and bend it to allow you to probably win those 1 million in Phase 1… are you winning in the rest of life with such a departure from truth?
@Suny , @Michael , @FlannelJesus , @Pseudonym , @noAxioms , @Pierre-Normand , @Patterner
I am not a OB but you might be going too far. We could say that for many or even most situations, CDT (TB) and EDT (OB) don’t differ at all. So if we take most situations + Newcomb problems then OB will win more in general.
My decision-making is simple:
- Most people who do A walk away with $1,000,000
- Most people who do B walk away with $1,000
- Therefore, I will do A
This is the only fact that matters, and is true whatever game is being played, whether it be Newcomb’s boxes or the lottery.
Absolutely not. Conflating correlation with causation has disatrous consequences. It loses in general.
EDT is built on conflating correlation with causation.
You are conflating correlation and causality.
This is the only fact that matters
Definitely not. Just look at the IQ test example. People who take IQ tests have more psychological troubles, you would conclude that therefore one should avoid taking IQ tests to not have psychological troubles.
Yeah but they will say they don’t and when you’ll give more obvious cases where their theory would lead them to conflate it, they won’t and will deny it. So technically even if they don’t follow a consistent way of making decision, you could say they win more.
Unless you can show OB do conflate correlation and causality in other (non-Newcomb) scenarios too.
One does not simply accept with no repercussions. Your choice now to be an OB is a choice to neglect correlation vs causation. All so that on the off-chance you play the Newcomb Game, you win a million bucks.
You are hindering your understanding of the game in order to win it. And that is a winning strategy for that game!
But it loses in most real-life games. It just so happens… the truth matters. The difference between causation and correlation matters. That is my basis for calling the winners of this contrived game irrational.
They chose to sacrifice reason itself in order to win the Newcomb Game.
You are looking at the bottom-line, but forgetting the truth. And your real bottom-line will be hurt by it. Do you think that you can affect the past? If so, you will be harmed by this belief (unless you have proven retro-causality to be true or something. If so, I commend you!)
No, I don’t. This is game theory, not retro-causality.
These three games are equivalent, and so whichever is the most rational choice in Game 3 is also the most rational choice in Game 1:
Game 1
- $1,000 is placed in a blue box
- You are cloned and the clone is asked to choose between either a) the red box or b) the red and the blue box
- The clone is destroyed
- If the clone chose (a) then $1,000,000 is placed in a red box
- You are asked to choose between either a) the red box or b) the red and the blue box
- You win the contents of the box(es) you chose
Game 2
- 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 red and the blue box
- The clone is asked to choose between either a) the red box or b) the red and 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
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
The most rational choice in Game 3 is (a) and so the most rational choice in Game 1 is (a).
Of course they do! By definition!
In Phase 1, they are an OB without their knowledge. They are not being an OB as a rational response to their impending dilemma. They are just being an OB.
And just being an OB is irrational, because it necessitates being someone who conflates correlation with causation.
Being an “OB” in Phase 1 of Version 2 however, is somewhat different. You can make yourself into an OB by other means than being an EDT adherent, because you have time to prepare. You can make a contract.
If you look at other versions than Version 1, then OB-ism is not equivalent to EDT with all its faults. Pure OB-ism is though. And in Version 1, the only form of OB-ism is pure OB-ism, which finds it formalization as EDT.
And EDT loses. There’s a reason why most decision theorists adhere to CDT, and not EDT. This is key. Really.
Pure OBs argue that they make more money, so their notion of rationality is correct. That would be a valid definition of rationality… if only it were correct! But it isn’t! Because you cannot be a pure OB and win in general, because you have to generally conflate correlation and causation to be a pure OB. It has to be your nature to be someone who conflates the two. And that is not, generally, a beneficial nature. It is not a rational nature.