Billy Tetrud [ARCHIVE] on Nostr: 📅 Original date posted:2022-05-01 📝 Original message:> if you are perfectly ...
📅 Original date posted:2022-05-01
📝 Original message:> if you are perfectly rational, you can certainly imagine a "what if"
where your goal is different from your current goal and figure out what you
would do ***if*** that were your goal instead.
I see what you're saying, and I'm a lot more on board with that. I still
think "rational" can't mean "perfect" - like "perfectly rational" is not
the same as "you magically get to the optimal answer". But I think my line
of thinking on this is a lot more pedantic than my previous contention. But
I will agree that for a given specific objective goal (that ignores other
goals), there is an objective set of answers that any logical person should
eventually be able to agree on. Of course, if there's any subjectivity in
the goal, then discussing the goal amongst two different people will really
mean that each of them are discussing slightly different goals, which
breaks the premise. So really for alignment to happen, the goal in question
needs to be really specific in order to remove any significant
subjectivity.
> better-than-human rationality
I like to think of rationality in the following way. Any economic actor is
a being that has goals they want to maximize for, and tools at their
disposal to analyze and affect their world. A rational actor is one that
attempts to use their tools to the best of their ability to maximize their
goals. Perhaps goals is a misleading word to use here, since it implies
something that can be achieved, whereas I really mean a set of weighted
metrics that can hypothetically always be improved upon. But in any case, a
human starts with goals built into their genetics, which in turn build
themselves into the structure of their body. The tools a human has is also
their body and their brain. The brain is not a perfect tool, and neither is
the rest of the body. However, humans use what they have to make decisions
and act on their world. The goals a human has evolve as they have
experiences in the world (which end up physically changing their brain). In
this sense, every human, and every possible actor really, must be a
rational actor. They're all doing the best they can, even if the tools at
their disposal are very suboptimal for maximizing their underlying goals.
What more can you ask of a rational actor than to use the tools they have
to achieve their goals?
So I don't think anyone is more or less "rational" than anyone else. They
just have different goals and different levels of ability to maximize those
goals. In my definition above, the goals are completely arbitrary. They
don't have to be anything in particular. A person could have the goal of
maximizing the number of paper clips in the world, at all other costs. This
would almost certainly be "bad" for that person, and "bad" for the world,
but if that's really what their goals are, then that "badness" is a
subjectivity that you and I would be placing on that goal because our goals
are completely different from it. To the being with that goal, it is a
totally perfect goal.
The idea that someone can be "more rational" than someone else kind of
boils everything down to one dimension. In reality, everyone has their
different skills and proficiencies. In a futures market, you might be
better at predicting the price of salmon, but you might be quite bad at
predicting human population changes over time. Does this mean you're "more
rational" about salmon but "less rational" about how human populations
change? I would say a better way to describe this is proficiency, rather
than rationality.
</digression>
> a future market
A futures market for predictions is an interesting idea. I haven't really
heard about such a thing really being done other than in little
experiments. Are you suggesting we use one to help make decisions about
bitcoin? One issue is that the questions a futures market answers have to,
like my conclusion in the above paragraph, be completely objective. So a
futures market can't answer the question "what's the best way to design
covenants?" tho it could answer the question "will CTV be activated by
2024?". But as a consequence, I don't think a future's market could help
much in the formulation of appropriate goals for bitcoin. That would need
to be hashed out by making a lot of different compromises amongst
everyone's various subjective opinion's about what is best.
And I think that's really what I'm suggesting here, is that we bitcoiners
discuss what the objective goals of bitcoin should be, or at least what
bounds on those goals there should be. And once we have these objective
goals, we can be aligned on how to appropriately solve them. It wouldn't
avoid the nashing of teeth needed to hash out the subjective parts of our
opinions in getting to those goals, but it could avoid much nashing of
teeth in the other half of the conversation: how to achieve the goals we
have reached consensus on.
Eg, should a goal of bitcoin be that 50% of the world's population should
require spending no more than 1% of their income to be able to run a full
node? Were we to decide on something akin to that, it would at least be a
question that has an objective truth value to it, even if we couldn't
feasibly confirm to 100% certainty whether we have achieved it, we could
probably confirm with some acceptable level of certainty below 100%.
On Sat, Apr 30, 2022 at 1:14 AM ZmnSCPxj <ZmnSCPxj at protonmail.com> wrote:
> Good morning Billy,
>
> > @Zman
> > > if two people are perfectly rational and start from the same
> information, they *will* agree
> > I take issue with this. I view the word "rational" to mean basically
> logical. Someone is rational if they advocate for things that are best for
> them. Two humans are not the same people. They have different circumstances
> and as a result different goals. Two actors with different goals will
> inevitably have things they rationally and logically disagree about. There
> is no universal rationality. Even an AI from outside space and time is
> incredibly likely to experience at least some value drift from its peers.
>
> Note that "the goal of this thing" is part of the information where both
> "start from" here.
>
> Even if you and I have different goals, if we both think about "given this
> goal, and these facts, is X the best solution available?" we will both
> agree, though our goals might not be the same as each other, or the same as
> "this goal" is in the sentence.
> What is material is simply that the laws of logic are universal and if you
> include the goal itself as part of the question, you will reach the same
> conclusion --- but refuse to act on it (and even oppose it) because the
> goal is not your own goal.
>
> E.g. "What is the best way to kill a person without getting caught?" will
> probably have us both come to the same broad conclusion, but I doubt either
> of us has a goal or sub-goal to kill a person.
> That is: if you are perfectly rational, you can certainly imagine a "what
> if" where your goal is different from your current goal and figure out what
> you would do ***if*** that were your goal instead.
>
> Is that better now?
>
> > > 3. Can we actually have the goals of all humans discussing this topic
> all laid out, *accurately*?
> > I think this would be a very useful exercise to do on a regular basis.
> This conversation is a good example, but conversations like this are rare.
> I tried to discuss some goals we might want bitcoin to have in a paper I
> wrote about throughput bottlenecks. Coming to a consensus around goals, or
> at very least identifying various competing groupings of goals would be
> quite useful to streamline conversations and to more effectively share
> ideas.
>
>
> Using a future market has the attractive property that, since money is
> often an instrumental sub-goal to achieve many of your REAL goals, you can
> get reasonably good information on the goals of people without them having
> to actually reveal their actual goals.
> Also, irrationality on the market tends to be punished over time, and a
> human who achieves better-than-human rationality can gain quite a lot of
> funds on the market, thus automatically re-weighing their thoughts higher.
>
> However, persistent irrationalities embedded in the design of the human
> mind will still be difficult to break (it is like a program attempting to
> escape a virtual machine).
> And an uninformed market is still going to behave pretty much randomly.
>
> Regards,
> ZmnSCPxj
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linuxfoundation.org/pipermail/bitcoin-dev/attachments/20220501/c23d0f10/attachment-0001.html>;
📝 Original message:> if you are perfectly rational, you can certainly imagine a "what if"
where your goal is different from your current goal and figure out what you
would do ***if*** that were your goal instead.
I see what you're saying, and I'm a lot more on board with that. I still
think "rational" can't mean "perfect" - like "perfectly rational" is not
the same as "you magically get to the optimal answer". But I think my line
of thinking on this is a lot more pedantic than my previous contention. But
I will agree that for a given specific objective goal (that ignores other
goals), there is an objective set of answers that any logical person should
eventually be able to agree on. Of course, if there's any subjectivity in
the goal, then discussing the goal amongst two different people will really
mean that each of them are discussing slightly different goals, which
breaks the premise. So really for alignment to happen, the goal in question
needs to be really specific in order to remove any significant
subjectivity.
> better-than-human rationality
I like to think of rationality in the following way. Any economic actor is
a being that has goals they want to maximize for, and tools at their
disposal to analyze and affect their world. A rational actor is one that
attempts to use their tools to the best of their ability to maximize their
goals. Perhaps goals is a misleading word to use here, since it implies
something that can be achieved, whereas I really mean a set of weighted
metrics that can hypothetically always be improved upon. But in any case, a
human starts with goals built into their genetics, which in turn build
themselves into the structure of their body. The tools a human has is also
their body and their brain. The brain is not a perfect tool, and neither is
the rest of the body. However, humans use what they have to make decisions
and act on their world. The goals a human has evolve as they have
experiences in the world (which end up physically changing their brain). In
this sense, every human, and every possible actor really, must be a
rational actor. They're all doing the best they can, even if the tools at
their disposal are very suboptimal for maximizing their underlying goals.
What more can you ask of a rational actor than to use the tools they have
to achieve their goals?
So I don't think anyone is more or less "rational" than anyone else. They
just have different goals and different levels of ability to maximize those
goals. In my definition above, the goals are completely arbitrary. They
don't have to be anything in particular. A person could have the goal of
maximizing the number of paper clips in the world, at all other costs. This
would almost certainly be "bad" for that person, and "bad" for the world,
but if that's really what their goals are, then that "badness" is a
subjectivity that you and I would be placing on that goal because our goals
are completely different from it. To the being with that goal, it is a
totally perfect goal.
The idea that someone can be "more rational" than someone else kind of
boils everything down to one dimension. In reality, everyone has their
different skills and proficiencies. In a futures market, you might be
better at predicting the price of salmon, but you might be quite bad at
predicting human population changes over time. Does this mean you're "more
rational" about salmon but "less rational" about how human populations
change? I would say a better way to describe this is proficiency, rather
than rationality.
</digression>
> a future market
A futures market for predictions is an interesting idea. I haven't really
heard about such a thing really being done other than in little
experiments. Are you suggesting we use one to help make decisions about
bitcoin? One issue is that the questions a futures market answers have to,
like my conclusion in the above paragraph, be completely objective. So a
futures market can't answer the question "what's the best way to design
covenants?" tho it could answer the question "will CTV be activated by
2024?". But as a consequence, I don't think a future's market could help
much in the formulation of appropriate goals for bitcoin. That would need
to be hashed out by making a lot of different compromises amongst
everyone's various subjective opinion's about what is best.
And I think that's really what I'm suggesting here, is that we bitcoiners
discuss what the objective goals of bitcoin should be, or at least what
bounds on those goals there should be. And once we have these objective
goals, we can be aligned on how to appropriately solve them. It wouldn't
avoid the nashing of teeth needed to hash out the subjective parts of our
opinions in getting to those goals, but it could avoid much nashing of
teeth in the other half of the conversation: how to achieve the goals we
have reached consensus on.
Eg, should a goal of bitcoin be that 50% of the world's population should
require spending no more than 1% of their income to be able to run a full
node? Were we to decide on something akin to that, it would at least be a
question that has an objective truth value to it, even if we couldn't
feasibly confirm to 100% certainty whether we have achieved it, we could
probably confirm with some acceptable level of certainty below 100%.
On Sat, Apr 30, 2022 at 1:14 AM ZmnSCPxj <ZmnSCPxj at protonmail.com> wrote:
> Good morning Billy,
>
> > @Zman
> > > if two people are perfectly rational and start from the same
> information, they *will* agree
> > I take issue with this. I view the word "rational" to mean basically
> logical. Someone is rational if they advocate for things that are best for
> them. Two humans are not the same people. They have different circumstances
> and as a result different goals. Two actors with different goals will
> inevitably have things they rationally and logically disagree about. There
> is no universal rationality. Even an AI from outside space and time is
> incredibly likely to experience at least some value drift from its peers.
>
> Note that "the goal of this thing" is part of the information where both
> "start from" here.
>
> Even if you and I have different goals, if we both think about "given this
> goal, and these facts, is X the best solution available?" we will both
> agree, though our goals might not be the same as each other, or the same as
> "this goal" is in the sentence.
> What is material is simply that the laws of logic are universal and if you
> include the goal itself as part of the question, you will reach the same
> conclusion --- but refuse to act on it (and even oppose it) because the
> goal is not your own goal.
>
> E.g. "What is the best way to kill a person without getting caught?" will
> probably have us both come to the same broad conclusion, but I doubt either
> of us has a goal or sub-goal to kill a person.
> That is: if you are perfectly rational, you can certainly imagine a "what
> if" where your goal is different from your current goal and figure out what
> you would do ***if*** that were your goal instead.
>
> Is that better now?
>
> > > 3. Can we actually have the goals of all humans discussing this topic
> all laid out, *accurately*?
> > I think this would be a very useful exercise to do on a regular basis.
> This conversation is a good example, but conversations like this are rare.
> I tried to discuss some goals we might want bitcoin to have in a paper I
> wrote about throughput bottlenecks. Coming to a consensus around goals, or
> at very least identifying various competing groupings of goals would be
> quite useful to streamline conversations and to more effectively share
> ideas.
>
>
> Using a future market has the attractive property that, since money is
> often an instrumental sub-goal to achieve many of your REAL goals, you can
> get reasonably good information on the goals of people without them having
> to actually reveal their actual goals.
> Also, irrationality on the market tends to be punished over time, and a
> human who achieves better-than-human rationality can gain quite a lot of
> funds on the market, thus automatically re-weighing their thoughts higher.
>
> However, persistent irrationalities embedded in the design of the human
> mind will still be difficult to break (it is like a program attempting to
> escape a virtual machine).
> And an uninformed market is still going to behave pretty much randomly.
>
> Regards,
> ZmnSCPxj
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linuxfoundation.org/pipermail/bitcoin-dev/attachments/20220501/c23d0f10/attachment-0001.html>;