Rolf Nelson wrote:
> In the (3) I gave, you're indexed so that the thermal fluctuation
> doesn't dissolve until November 1, so your actions still have
> consequences.

Still not a problem: the space-time region that I can affect in (3) is too 
small (i.e., its measure is too small, complexity too large) for me to care 
much about the consequences of my actions on it.

> This is one of a larger class of problems related to volition, and the
> coupling of my qualia to an external reality, that I don't currently
> have an answer for. I want to live on in the current Universe, I don't
> to die and have a duplicate of myself created in a different Universe.
> I want to eat a real ice cream cone, I don't want you to stimulate my
> neurons to make me imagine I'm eating an ice cream cone. I would argue
> that a world where I can interact with real people is, in some sense,
> better than a world where I interact with imaginary people who I
> believe are real.

To me, these examples show that we do not care just about qualia, but also 
about attributes and features of the multiverse that can not be classified 
as qualia, and therefore we should rule out decision theories that cannot 
incorporate preferences over non-qualia.

> You have a general model, which can encompass classical decision
> theory, but can also encompass other models as well. It's not
> immediately clear to me what benefit, if any, we get from such a
> general model.

Fair question. I'll summarize:

1. We are forced into considering such a general model because we don't have 
a more specific one that doesn't lead to counterintuitive implications.

2. It shows us what probabilities really are. For someone whose preferences 
over the multiverse can be expressed as a linear combination of preferences 
over regions of the multiverse, a probability function can be interpreted as 
a representation of how much he cares about each region. I would argue that 
most of us in fact have preferences of this form, at least approximately, 
which explains why probability theory has been useful for us.

3. It gives us a useful framework for considering anthropic reasoning 
problems such as the Doomsday Argument and the Simulation Argument. We can 
now recast these questions into "Do we prefer a multiverse where people in 
our situation act as if doom is near?" and "Do we prefer a multiverse where 
people in our situation act as if they are in simulations?" I argue that its 
easier for us to consider these questions in this form.

4. For someone on a practical mission to write an AI that makes sensible 
decisions, perhaps the model can serve as a starting point and as 
illustration of how far away we still are from that goal.

You received this message because you are subscribed to the Google Groups 
"Everything List" group.
To post to this group, send email to [EMAIL PROTECTED]
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at

Reply via email to