| Topic: |
Religions > Atheism |
| User: |
"Uncle Clover" |
| Date: |
14 Nov 2006 09:58:04 AM |
| Object: |
OT: The simulated human |
Sooner or later if it hasn't happened already, computers are going to be
capable of simulating cellular activity right on down to the single-gene level
of DNA, RNA & all the other various molecules involved. When that happens, the
structure and behaviors of the organism encoded for should naturally emerge from
simply letting the cell function in a simulated environment, at least so long as
the expressed parameters of the organism and its environment do not exceed the
capabilities of the system upon which it is being run. This would doubtless
limit us to smaller, less complex organisms at first.
Even so, this sounds to me like a truly monumental computing task. Does
anyone have any good idea as to the particular system specs that would be
required, and how far we are from such capabilities? :-?
Even farther on down the road, we should be able to do this with more
complex organisms such as humans. A further curiosity of mine is this: Suppose
we encode and enact the formulation of a human in such a simulated environment.
Just one human alone would be quite an accomplishment, so to satisfy its social
needs, there could be other humans simulated into the environment but on a
less-complex level than the fully-simulated one (i.e., the computer wouldn't
need to simulate them on a cellular level, just perhaps down to the resolution
of the major organs and organ groups, their personality being purely AI). If we
took such a human and left it grow up in such an environment and provided it
with an education that's considered good at least by today's standards, how
might it deduce the circumstances of its own existence? Would you favor an
open-ended approach where the fact of its simulated nature is made known to it
as it grows so that it never really has to go through any major shock? Or do
you think it would be a better idea to keep the nature of its existence hidden
from it?
For my own self, I would think it has the right to know. The problem is
that if you tell it that from "birth", that may interfere with the development
of its psychology in some way. If one is looking for the most natural
psychological development possible, I think that might be impossible if the
person knows it's not in and it may never get to -be- in "the real world". And
yet there seems no one single "good time" to tell it during the course of a
lifetime, no single "magical moment" where it would go from being "not ready" to
"ready". Perhaps if its simulated nature were presented to it in a toned-down
way. Like the way we learn over time that there are other countries and other
places with people who are different from us and cultures that would be strange
to us, and that we may never get to see for ourselves. It could be presented
with its place in a simulation being described as just another location -
instead of being from a small town in Pennsylvania, it would be from Simville,
Iowa (or whatever state the simulation is being run in). It could also be
taught that its condition as a simulated being is just another condition some
people are born with and some are not. That's a natural enough way for many of
us to grow up - realizing there's something "different" about us that makes us
not quite like everybody else. It wouldn't need to know right away that it's
the only one (assuming it is).
I know that if I were a simulated being, I could understand my
simulators not wanting to tell me right away so that my psychology could form
more naturally, but I feel I would have the right to know it -eventually-.
After my psychology has firmed up a good bit, I don't think that sort of
knowledge could possibly do any harm. But that's just me, I guess...
Anyway, lots and lots of issues to consider with this. I know it's a
long way off yet, but I've a feeling the capacity for such sophisticated
simulation is perhaps much closer than it seems.
Just curious as always...
--
L8r,
Uncle Clover
************************************************
In my experience, one's degree of wisdom tends
to bear an exponentially inverse relationship
to one's outpouring of words.
Clearly, I've a _long_ way to go... ;-)
************************************************
The true mark of a civilized society is that its
citizens know how to hate each other peacefully.
************************************************
.
|
|
| User: "Matt Silberstein" |
|
| Title: Re: OT: The simulated human |
14 Nov 2006 10:07:34 AM |
|
|
On Tue, 14 Nov 2006 10:58:04 -0500, in alt.atheism , Uncle Clover
<UncleClover@SpamMeNot.com> in
<0nojl25pd9jcp1dei8db5c85apjj2m4o0j@4ax.com> wrote:
Sooner or later if it hasn't happened already, computers are going to be
capable of simulating cellular activity right on down to the single-gene level
of DNA, RNA & all the other various molecules involved.
I doubt that. It may be possible to model a single cell, but not at
that level of detail and not in real-time.
When that happens, the
structure and behaviors of the organism encoded for should naturally emerge from
simply letting the cell function in a simulated environment, at least so long as
the expressed parameters of the organism and its environment do not exceed the
capabilities of the system upon which it is being run. This would doubtless
limit us to smaller, less complex organisms at first.
You seem to have gone from simulating cells to simulating full
organisms, a big jump.
Even so, this sounds to me like a truly monumental computing task. Does
anyone have any good idea as to the particular system specs that would be
required, and how far we are from such capabilities? :-?
To do it in real-time is likely impossible, to simulate a single cell
on the chemical level would take far more processing power than exists
today. Consider this: the problem of protein folding, of figuring out
how a protein folds given the DNA/RNA construction of the protein is
remarkably difficult. Doing it for lots and lots of proteins and
looking at the interactions of all the chemicals in the cell at once
is orders of magnitude more difficult.
Even farther on down the road, we should be able to do this with more
complex organisms such as humans.
Wow is that a jump.
A further curiosity of mine is this: Suppose
we encode and enact the formulation of a human in such a simulated environment.
Just one human alone would be quite an accomplishment, so to satisfy its social
needs, there could be other humans simulated into the environment but on a
less-complex level than the fully-simulated one (i.e., the computer wouldn't
need to simulate them on a cellular level, just perhaps down to the resolution
of the major organs and organ groups, their personality being purely AI).
Why do you need to simulate the organs to simulate intelligence?
If we
took such a human and left it grow up in such an environment and provided it
with an education that's considered good at least by today's standards, how
might it deduce the circumstances of its own existence? Would you favor an
open-ended approach where the fact of its simulated nature is made known to it
as it grows so that it never really has to go through any major shock? Or do
you think it would be a better idea to keep the nature of its existence hidden
from it?
Gad, after all that you want to ask some moral question? There are far
worse moral problems along the way. What about all of the prototypes
for this simulated human? (There will likely be thousands along the
way.) Do you keep them going or "kill" them? What if they operate in
100 to 1 time, so a year of the simulation takes 100 years of real
time and costs lots of money? Still keep it alive? After all, you are
responsible for it.
[snip]
--
Matt Silberstein
Do something today about the Darfur Genocide
http://www.beawitness.org
http://www.darfurgenocide.org
http://www.savedarfur.org
"Darfur: A Genocide We can Stop"
.
|
|
|
| User: "quibbler" |
|
| Title: Re: OT: The simulated human |
14 Nov 2006 04:55:37 PM |
|
|
In article <m0qjl2d8q30bnts8krngle8knd5ob3jje8@4ax.com>,
RemoveThisPrefixmatts2nospam@ix.netcom.com says...
On Tue, 14 Nov 2006 10:58:04 -0500, in alt.atheism , Uncle Clover
<UncleClover@SpamMeNot.com> in
<0nojl25pd9jcp1dei8db5c85apjj2m4o0j@4ax.com> wrote:
Sooner or later if it hasn't happened already, computers are going to be
capable of simulating cellular activity right on down to the single-gene level
of DNA, RNA & all the other various molecules involved.
I doubt that. It may be possible to model a single cell, but not at
that level of detail and not in real-time.
I agree that we won't be doing it too soon in real time. But it ought
to be possible eventually, particularly with advances such as
spintronics, quantum computing, continued nano-miniaturization, etc.
When that happens, the
structure and behaviors of the organism encoded for should naturally emerge from
simply letting the cell function in a simulated environment, at least so long as
the expressed parameters of the organism and its environment do not exceed the
capabilities of the system upon which it is being run. This would doubtless
limit us to smaller, less complex organisms at first.
You seem to have gone from simulating cells to simulating full
organisms, a big jump.
Given the trend of moore's law, it ought to eventually be possible.
Perhaps it won't be done with silicon, but it should only be a matter of
time. Naturally, we can only predict future technological capacity in
rough outline.
Even so, this sounds to me like a truly monumental computing task. Does
anyone have any good idea as to the particular system specs that would be
required, and how far we are from such capabilities? :-?
To do it in real-time is likely impossible,
If it can be done in nature, then in principle, it should be possible to
simulated it via some other physical process.
Why do you need to simulate the organs to simulate intelligence?
Granted, that would likely be the long way round.
If we
took such a human and left it grow up in such an environment and provided it
with an education that's considered good at least by today's standards, how
might it deduce the circumstances of its own existence? Would you favor an
open-ended approach where the fact of its simulated nature is made known to it
as it grows so that it never really has to go through any major shock? Or do
you think it would be a better idea to keep the nature of its existence hidden
from it?
Gad, after all that you want to ask some moral question?
It's true that there might be bigger fish to fry. But it might be an
interesting psychological experiment to try things various different
ways, I'd imagine. Of course, the reaction would largely depend upon
the specifics of the organism simulated. It seems more likely, that if
we were already able to simulate this organism, that we could solve for
its reactions without the mess of simulation.
--
Quibbler (quibbler247atyahoo.com)
"It is fashionable to wax apocalyptic about the
threat to humanity posed by the AIDS virus, 'mad cow'
disease, and many others, but I think a case can be
made that faith is one of the world's great evils,
comparable to the smallpox virus but harder to
eradicate." -- Richard Dawkins
.
|
|
|
| User: "Matt Silberstein" |
|
| Title: Re: OT: The simulated human |
14 Nov 2006 05:05:26 PM |
|
|
On Tue, 14 Nov 2006 15:55:37 -0700, in alt.atheism , quibbler
<quibbler247@yahoo.com> in
<MPG.1fc3c8f3b5182f1f98974d@news.readfreenews.net> wrote:
[snip]
If it can be done in nature, then in principle, it should be possible to
simulated it via some other physical process.
I suspect you are making claims based on Turing equivalence. If so,
the claim is not supported. Turing Equivalence says that what one
*Turing Machine* can do another can do *if we ignore time. Of course
no actual Turing Machine exists (there is that pesky problem of
infinite memory).
[snip]
--
Matt Silberstein
Do something today about the Darfur Genocide
http://www.beawitness.org
http://www.darfurgenocide.org
http://www.savedarfur.org
"Darfur: A Genocide We can Stop"
.
|
|
|
|
|
| User: "Uncle Clover" |
|
| Title: Re: OT: The simulated human |
14 Nov 2006 10:30:39 AM |
|
|
On Tue, 14 Nov 2006 16:07:34 GMT, Matt Silberstein
<RemoveThisPrefixmatts2nospam@ix.netcom.com> wrote:
On Tue, 14 Nov 2006 10:58:04 -0500, in alt.atheism , Uncle Clover
<UncleClover@SpamMeNot.com> in
<0nojl25pd9jcp1dei8db5c85apjj2m4o0j@4ax.com> wrote:
Sooner or later if it hasn't happened already, computers are going to be
capable of simulating cellular activity right on down to the single-gene level
of DNA, RNA & all the other various molecules involved.
I doubt that. It may be possible to model a single cell, but not at
that level of detail and not in real-time.
When that happens, the
structure and behaviors of the organism encoded for should naturally emerge from
simply letting the cell function in a simulated environment, at least so long as
the expressed parameters of the organism and its environment do not exceed the
capabilities of the system upon which it is being run. This would doubtless
limit us to smaller, less complex organisms at first.
You seem to have gone from simulating cells to simulating full
organisms, a big jump.
I know. But if it's a truly good simulation of a cell, and that cell is an
embryo of some organism, it should form the organism naturally, complete with
all emergent properties and behaviors. The psychology and physical processes of
the organism wouldn't need to be simulated, because they're encoded in the genes
already. Provided you've also simulated the right kind of environment in which
for it to feed and grow.
Even so, this sounds to me like a truly monumental computing task. Does
anyone have any good idea as to the particular system specs that would be
required, and how far we are from such capabilities? :-?
To do it in real-time is likely impossible, to simulate a single cell
on the chemical level would take far more processing power than exists
today. Consider this: the problem of protein folding, of figuring out
how a protein folds given the DNA/RNA construction of the protein is
remarkably difficult. Doing it for lots and lots of proteins and
looking at the interactions of all the chemicals in the cell at once
is orders of magnitude more difficult.
Thanks for clarifying that. I figured it wasn't possible in the near-term. But
our technology has had a way of sort of "leapfrogging" itself, alternating
between relatively short outbursts of amazing advance with somewhat longer
periods of relative stasis between. At this point, I think the only limitation
is processing speed and memory size. No small impediments, I'm sure. But once
those gaps are bridged, I would think that running a successful organism
simulation would be as simple as specifying the nature of the individual
elements involved and just letting them run - kind of like a cellular automata
experiment, though perhaps 2,000 generations removed.
Even farther on down the road, we should be able to do this with more
complex organisms such as humans.
Wow is that a jump.
Yup. :-) But the only real major jump between a small organism and humans is
that of size. If a cell is simulated correctly and the environment as well, the
human should result from it without any additional programming needed. At least
that's how I would -expect- it to work. Were we to do the same thing with
real-world molecules - assemble them together into an embryonic cell and provide
this cell a womb-like environment in which to grow - a human should result from
that. And so - while it may just be a failure to consider something on my part
- I don't see why the same thing wouldn't hold true in a fully simulated
environment.
A further curiosity of mine is this: Suppose
we encode and enact the formulation of a human in such a simulated environment.
Just one human alone would be quite an accomplishment, so to satisfy its social
needs, there could be other humans simulated into the environment but on a
less-complex level than the fully-simulated one (i.e., the computer wouldn't
need to simulate them on a cellular level, just perhaps down to the resolution
of the major organs and organ groups, their personality being purely AI).
Why do you need to simulate the organs to simulate intelligence?
That's not to simulate intelligence, that's just so that if anything ever
"happens" to the other people in the simulation, the fully-simulated human
("FS-human") wouldn't be led to an incorrect impression of what humans are made
of. It wouldn't be necessary for the half-simulated humans ("HS-humans"), but
for the the benefit of the FS-human. The FS-human might want to become a doctor
or something, and even if not, it may witness death among its HS-human
acquaintences. In order for it to have the most realistic human experience
possible, it would need to be able to see blood and guts any time blood and guts
would be the expected experience in the real world.
If we
took such a human and left it grow up in such an environment and provided it
with an education that's considered good at least by today's standards, how
might it deduce the circumstances of its own existence? Would you favor an
open-ended approach where the fact of its simulated nature is made known to it
as it grows so that it never really has to go through any major shock? Or do
you think it would be a better idea to keep the nature of its existence hidden
from it?
Gad, after all that you want to ask some moral question?
Better late than never, right? ;-) I jumped to the moral questions that are of
most interest to me at the moment. In a real-world situation, however, I would
approach them as they come, whether they're the most interesting aspect of the
situation or not.
There are far
worse moral problems along the way. What about all of the prototypes
for this simulated human? (There will likely be thousands along the
way.) Do you keep them going or "kill" them? What if they operate in
100 to 1 time, so a year of the simulation takes 100 years of real
time and costs lots of money? Still keep it alive? After all, you are
responsible for it.
Those are some very good questions, thanks for asking them. That's another
reason I like discussing things like this, here.
As for prototypes, I think once we get the process of cellular development
properly simulated by producing less-complex organisms, that will be the end of
the prototyping needed. Everything else about an organism that we observe in
the real world is a function that emerges naturally as the organism grows and
develops within its given environment. We don't need to simulate any of those
properties, they should arise naturally as they do in the real world.
But perhaps there's something I'm overlooking. Is there a reason we wouldn't be
able to jump directly from any of a gajillion non-sapient organisms to humans
once the processes of cellular and organism development are perfected? Is there
some reason that the human traits of the simulated human wouldn't evolve
naturally in a suitably simulated environment just from the activity ocurring on
a cellular level? If not, then the risks for damage or malformation to the
simulated human will be much the same as they are for real-world humans.
Thanks again for your feedback. You're an individual of intelligence, and I
appreciate your taking the time to respond to these what must seem like "naive"
queries of mine.
--
L8r,
Uncle Clover
************************************************
In my experience, one's degree of wisdom tends
to bear an exponentially inverse relationship
to one's outpouring of words.
Clearly, I've a _long_ way to go... ;-)
************************************************
The true mark of a civilized society is that its
citizens know how to hate each other peacefully.
************************************************
.
|
|
|
| User: "Matt Silberstein" |
|
| Title: Re: OT: The simulated human |
14 Nov 2006 11:14:51 AM |
|
|
On Tue, 14 Nov 2006 11:30:39 -0500, in alt.atheism , Uncle Clover
<UncleClover@SpamMeNot.com> in
<7kqjl25bll116g6ku2qkmjmugsifh3t1oo@4ax.com> wrote:
On Tue, 14 Nov 2006 16:07:34 GMT, Matt Silberstein
<RemoveThisPrefixmatts2nospam@ix.netcom.com> wrote:
On Tue, 14 Nov 2006 10:58:04 -0500, in alt.atheism , Uncle Clover
<UncleClover@SpamMeNot.com> in
<0nojl25pd9jcp1dei8db5c85apjj2m4o0j@4ax.com> wrote:
Sooner or later if it hasn't happened already, computers are going to be
capable of simulating cellular activity right on down to the single-gene level
of DNA, RNA & all the other various molecules involved.
I doubt that. It may be possible to model a single cell, but not at
that level of detail and not in real-time.
When that happens, the
structure and behaviors of the organism encoded for should naturally emerge from
simply letting the cell function in a simulated environment, at least so long as
the expressed parameters of the organism and its environment do not exceed the
capabilities of the system upon which it is being run. This would doubtless
limit us to smaller, less complex organisms at first.
You seem to have gone from simulating cells to simulating full
organisms, a big jump.
I know. But if it's a truly good simulation of a cell, and that cell is an
embryo of some organism, it should form the organism naturally, complete with
all emergent properties and behaviors.
Suppose it takes half the worlds computing power to simulate a cell.
It would take a few hundred times that to simulate a small organ. Ok,
thousands times more since you have to have the interactions handled.
The psychology and physical processes of
the organism wouldn't need to be simulated, because they're encoded in the genes
already.
No, they are not. Genes are not magic containers encoding everything.
Genes are chains of DNA. Given a particular environment that DNA will
produce particular proteins. Given a particular environment those
proteins will produce certain systems. Remember, every (human) cell in
the body is encoded by the same DNA, yet the cells differ so much.
Provided you've also simulated the right kind of environment in which
for it to feed and grow.
A remarkably difficult task. Here is the thing, a secret to good
modeling is knowing which things to model and which things you don't
need to model. Building a cell by modeling quantum physics, for
example, is likely over-kill. If your goal is to understand a cell you
probably have to model some of the chemicals. If your goal is to
understand the organism you probably don't have to know that level of
detail.
Even so, this sounds to me like a truly monumental computing task. Does
anyone have any good idea as to the particular system specs that would be
required, and how far we are from such capabilities? :-?
To do it in real-time is likely impossible, to simulate a single cell
on the chemical level would take far more processing power than exists
today. Consider this: the problem of protein folding, of figuring out
how a protein folds given the DNA/RNA construction of the protein is
remarkably difficult. Doing it for lots and lots of proteins and
looking at the interactions of all the chemicals in the cell at once
is orders of magnitude more difficult.
Thanks for clarifying that. I figured it wasn't possible in the near-term. But
our technology has had a way of sort of "leapfrogging" itself, alternating
between relatively short outbursts of amazing advance with somewhat longer
periods of relative stasis between. At this point, I think the only limitation
is processing speed and memory size.
And communication speed. And, these days, heat dissipation.
No small impediments, I'm sure. But once
those gaps are bridged, I would think that running a successful organism
simulation would be as simple as specifying the nature of the individual
elements involved and just letting them run - kind of like a cellular automata
experiment, though perhaps 2,000 generations removed.
As a thought experiment hand wave that is fine, as a prediction it is
orders of magnitude beyond what we can even think of today.
Even farther on down the road, we should be able to do this with more
complex organisms such as humans.
Wow is that a jump.
Yup. :-) But the only real major jump between a small organism and humans is
that of size. If a cell is simulated correctly and the environment as well, the
human should result from it without any additional programming needed.
The interaction between the various parts is non-trivial. Synching the
timing is not at all easy. Getting a cell to work is one thing,
getting the millions of cells to work properly together is yet
another. And then we have all of the non-cellular stuff like blood and
hormones and such.
At least
that's how I would -expect- it to work. Were we to do the same thing with
real-world molecules - assemble them together into an embryonic cell and provide
this cell a womb-like environment in which to grow - a human should result from
that. And so - while it may just be a failure to consider something on my part
- I don't see why the same thing wouldn't hold true in a fully simulated
environment.
A further curiosity of mine is this: Suppose
we encode and enact the formulation of a human in such a simulated environment.
Just one human alone would be quite an accomplishment, so to satisfy its social
needs, there could be other humans simulated into the environment but on a
less-complex level than the fully-simulated one (i.e., the computer wouldn't
need to simulate them on a cellular level, just perhaps down to the resolution
of the major organs and organ groups, their personality being purely AI).
Why do you need to simulate the organs to simulate intelligence?
That's not to simulate intelligence, that's just so that if anything ever
"happens" to the other people in the simulation, the fully-simulated human
("FS-human") wouldn't be led to an incorrect impression of what humans are made
of. It wouldn't be necessary for the half-simulated humans ("HS-humans"), but
for the the benefit of the FS-human. The FS-human might want to become a doctor
or something, and even if not, it may witness death among its HS-human
acquaintences. In order for it to have the most realistic human experience
possible, it would need to be able to see blood and guts any time blood and guts
would be the expected experience in the real world.
Oh, wow, does that make the problem even more difficult. If this is
not in real-time you now have to simulate an external world as well.
If we
took such a human and left it grow up in such an environment and provided it
with an education that's considered good at least by today's standards, how
might it deduce the circumstances of its own existence? Would you favor an
open-ended approach where the fact of its simulated nature is made known to it
as it grows so that it never really has to go through any major shock? Or do
you think it would be a better idea to keep the nature of its existence hidden
from it?
Gad, after all that you want to ask some moral question?
Better late than never, right? ;-) I jumped to the moral questions that are of
most interest to me at the moment. In a real-world situation, however, I would
approach them as they come, whether they're the most interesting aspect of the
situation or not.
There are far
worse moral problems along the way. What about all of the prototypes
for this simulated human? (There will likely be thousands along the
way.) Do you keep them going or "kill" them? What if they operate in
100 to 1 time, so a year of the simulation takes 100 years of real
time and costs lots of money? Still keep it alive? After all, you are
responsible for it.
Those are some very good questions, thanks for asking them. That's another
reason I like discussing things like this, here.
As for prototypes, I think once we get the process of cellular development
properly simulated by producing less-complex organisms, that will be the end of
the prototyping needed.
Then you have never worked in systems development. Think of the
abortion questions, now make them worse. Each system start-up is a
conception. No matter where you stand on the abortion question you
have to realize that there are difficult moral questions involved. Now
you are making up cells and trying to simulate humans. Even the real
humans have problem with development, your simulation system is going
to go through lots of work to get it right. And not all of the
problems are catastrophic (i.e. leading to death), lots will just
cause deformities of some sort.
Everything else about an organism that we observe in
the real world is a function that emerges naturally as the organism grows and
develops within its given environment. We don't need to simulate any of those
properties, they should arise naturally as they do in the real world.
And so you have all of those problems.
But perhaps there's something I'm overlooking. Is there a reason we wouldn't be
able to jump directly from any of a gajillion non-sapient organisms to humans
once the processes of cellular and organism development are perfected?
Each of those steps takes work. It is not a magical jump.
Is there
some reason that the human traits of the simulated human wouldn't evolve
naturally in a suitably simulated environment just from the activity ocurring on
a cellular level?
Billion of years of work? But I think you mean develop, not evolve.
And for that there are all of the problems in systems development plus
all of the problems in human gestation development.
If not, then the risks for damage or malformation to the
simulated human will be much the same as they are for real-world humans.
Yep. Now think of your first 100 tries that are messed up in some way.
Each of which taking a major amount of computing power. Think of
modern terms and assume it costs $1 billion a year to run one of the
simulations. Do you keep the deformed ones alive? What if 100 years of
real-time is 1 year of simulation time.
Thanks again for your feedback. You're an individual of intelligence, and I
appreciate your taking the time to respond to these what must seem like "naive"
queries of mine.
Oh, no, I know where this is going. First the compliments, then some
remarks about how the light and my eyes. Then chocolate and flowers.
Pretty soon I am carrying your baby and crying while you go out
drinking with the boys. Sorry, but I've been around enough to be
careful.
--
Matt Silberstein
Do something today about the Darfur Genocide
http://www.beawitness.org
http://www.darfurgenocide.org
http://www.savedarfur.org
"Darfur: A Genocide We can Stop"
.
|
|
|
| User: "Uncle Clover" |
|
| Title: Re: OT: The simulated human |
14 Nov 2006 03:23:02 PM |
|
|
On Tue, 14 Nov 2006 17:14:51 GMT, Matt Silberstein
<RemoveThisPrefixmatts2nospam@ix.netcom.com> wrote:
On Tue, 14 Nov 2006 11:30:39 -0500, in alt.atheism , Uncle Clover
<UncleClover@SpamMeNot.com> in
<7kqjl25bll116g6ku2qkmjmugsifh3t1oo@4ax.com> wrote:
<snip>
I know. But if it's a truly good simulation of a cell, and that cell is an
embryo of some organism, it should form the organism naturally, complete with
all emergent properties and behaviors.
Suppose it takes half the worlds computing power to simulate a cell.
It would take a few hundred times that to simulate a small organ. Ok,
thousands times more since you have to have the interactions handled.
How many times faster than ENEAC are today's computers? Unless we're
approaching some sort of critical maximum in computational speed potential, I
don't see the above as any problem at all. It would be if we were trying to do
this -now-, but in the future? There's every reason to believe we'll get there
- at least so long as we aren't obliviated first.
The psychology and physical processes of
the organism wouldn't need to be simulated, because they're encoded in the genes
already.
No, they are not. Genes are not magic containers encoding everything.
I understand that. Bad wording on my part, I suppose.
Genes are chains of DNA. Given a particular environment that DNA will
produce particular proteins. Given a particular environment those
proteins will produce certain systems. Remember, every (human) cell in
the body is encoded by the same DNA, yet the cells differ so much.
That doesn't negate the fact that all a single human embryonic cell needs to
grow and develop as a human - with all the typical physiological and
psychological properties - is the right environment. The genes are every bit as
responsible for this as anything else. Genes are the only thing which
differentiate us from inorganic matter. They may not -directly- produce such
cellular phenomenon, but without genes, such cellular phenomenon is impossible.
Designing a single embryonic cell is all that is needed to ensure that the
simulated human will -have- a personality and phsyciological condition. These
things don't need to be programmed, they will form naturally just by the
processes taking place within the cells, which I say for the simple fact that
that's exactly what they do in the real world. So there is no functional
difference between simulating a less complex organism and a human.
Provided you've also simulated the right kind of environment in which
for it to feed and grow.
A remarkably difficult task. Here is the thing, a secret to good
modeling is knowing which things to model and which things you don't
need to model. Building a cell by modeling quantum physics, for
example, is likely over-kill. If your goal is to understand a cell you
probably have to model some of the chemicals. If your goal is to
understand the organism you probably don't have to know that level of
detail.
Understood. My own purpose, were this to be doable today, wouldn't be to model
anything, it would be to reproduce through intellectual rather than biological
means. Thus, it would have to be as real as it possibly could be.
<snip>
At this point, I think the only limitation is processing speed and
memory size.
And communication speed. And, these days, heat dissipation.
I tend to view communication speed as included in processing speed, but even
separately, it's speed just the same. As for the heat dissipation, there are
any number of a thousand ways to deal with that. There's no reason computers
can't be equipped with hydraulic cooling systems, for instance.
No small impediments, I'm sure. But once
those gaps are bridged, I would think that running a successful organism
simulation would be as simple as specifying the nature of the individual
elements involved and just letting them run - kind of like a cellular automata
experiment, though perhaps 2,000 generations removed.
As a thought experiment hand wave that is fine, as a prediction it is
orders of magnitude beyond what we can even think of today.
Obviously not, or I wouldn't have thought of it. ;-) Seriously, though, I know
what you mean, but there is no harm in speculating on what we'll likely be able
to do at some future point, assuming civilization isn't destroyed in the
interim. There's no reason not to at least kick the subject around a little,
nor is there any reason to demean the conversation by calling it a "handwave".
The scenario of which I speak is a scenario we can see coming. As long as there
are people ready, willing and able to devote time to it, I favor working on the
problems we haven't encountered long before they arrive.
<snip>
Yup. :-) But the only real major jump between a small organism and humans is
that of size. If a cell is simulated correctly and the environment as well, the
human should result from it without any additional programming needed.
The interaction between the various parts is non-trivial. Synching the
timing is not at all easy. Getting a cell to work is one thing,
getting the millions of cells to work properly together is yet
another. And then we have all of the non-cellular stuff like blood and
hormones and such.
Sure the interactions are non-trivial, I didn't suggest otherwise. The point I
was making is that we don't need to -program- them. Yes, it will take
incomprehensible amounts of computing power to "allow" such processes to occur,
but they can be made to occur just by creating the first cell. You don't need
to tell a stomach how to interact with the intestine, you need only to have left
the two organs emerge. You don't need to tell a spleen how to produce blood
cells or plasma, you don't need to do any of that stuff. The cells will do it
on their own because that's just how cells in a multicellular organism have
evolved to behave.
<snip>
That's not to simulate intelligence, that's just so that if anything ever
"happens" to the other people in the simulation, the fully-simulated human
("FS-human") wouldn't be led to an incorrect impression of what humans are made
of. It wouldn't be necessary for the half-simulated humans ("HS-humans"), but
for the the benefit of the FS-human. The FS-human might want to become a doctor
or something, and even if not, it may witness death among its HS-human
acquaintences. In order for it to have the most realistic human experience
possible, it would need to be able to see blood and guts any time blood and guts
would be the expected experience in the real world.
Oh, wow, does that make the problem even more difficult. If this is
not in real-time you now have to simulate an external world as well.
That's definitely part of the plan. I know it's beyond anything we can do
today, but even without the actual computers to run the simulation, we very much
can begin formulating the necessary algorithms in the here and now. While some
things are still mysteries to us, there's plenty we do know that we could begin
working on right away.
<snip>
As for prototypes, I think once we get the process of cellular development
properly simulated by producing less-complex organisms, that will be the end of
the prototyping needed.
Then you have never worked in systems development. Think of the
abortion questions, now make them worse. Each system start-up is a
conception. No matter where you stand on the abortion question you
have to realize that there are difficult moral questions involved. Now
you are making up cells and trying to simulate humans. Even the real
humans have problem with development, your simulation system is going
to go through lots of work to get it right. And not all of the
problems are catastrophic (i.e. leading to death), lots will just
cause deformities of some sort.
What you say is true, but I don't see how it contradicts what I've written.
These are questions we deal with in the real world, why on earth -shouldn't- we
deal with them in a simulated one? :-?
Everything else about an organism that we observe in
the real world is a function that emerges naturally as the organism grows and
develops within its given environment. We don't need to simulate any of those
properties, they should arise naturally as they do in the real world.
And so you have all of those problems.
Yes, as we do in real life. It doesn't matter how the simulated humans are
made, the risks would be every bit the same as they are with real humans, and
there's really nothing to even talk about in that regard. So some of the
organisms will have health and behavioral problems - so will some of our
real-world kids, that's not a justifiable reason to avoid having them.
But perhaps there's something I'm overlooking. Is there a reason we wouldn't be
able to jump directly from any of a gajillion non-sapient organisms to humans
once the processes of cellular and organism development are perfected?
Each of those steps takes work. It is not a magical jump.
My words seem to be leading you to think I think of this all in "magical" terms,
as you keep bringing up the word. I'm well aware of the issues at play - most
of them, at least. I'm not about to address each and every single possible
contingency just to discuss the scenario in general terms, though I definitely
enjoying discussions about the related issues of interest to others. One
doesn't paint every vein in a picture of a leaf on a tree without first having
painted the basic shape and color of the leaf first. As you correctly pointed
out, the possibility of anything like this is far off. We can work on painting
the veins onto it later, there's just too many preliminary discussions to be
had.
Is there
some reason that the human traits of the simulated human wouldn't evolve
naturally in a suitably simulated environment just from the activity ocurring on
a cellular level?
Billion of years of work? But I think you mean develop, not evolve.
And for that there are all of the problems in systems development plus
all of the problems in human gestation development.
The system development problem is one which I'm confident will be solved with
time. They are not issues we can seriously address at this stage of the game,
there's just too much we don't yet know about the specifics of what would be
required.
As for the other problems you mention, I don't see why they would justify not
doing such a simulation. They're nothing that would be created as a result of
the simulation, they're things that actually happen to people. I'm not saying
such issues won't be difficult to deal with - they will be, because they already
are. All of the same conditions and decisions we must make in the real world
cannot possibly be excluded from a simulated one, not if we expect it to bear
any relationship to the real world. As we solve some of our problems in the
real world, we can also solve them in the simulated one.
If not, then the risks for damage or malformation to the
simulated human will be much the same as they are for real-world humans.
Yep. Now think of your first 100 tries that are messed up in some way.
Each of which taking a major amount of computing power. Think of
modern terms and assume it costs $1 billion a year to run one of the
simulations. Do you keep the deformed ones alive? What if 100 years of
real-time is 1 year of simulation time.
We deal with them in precisely the same way we would deal with real-world
humans. Of course it could be very expensive. It might not be all that
expensive, too. To produce the computing power of a modern high-end computer,
you'd have needed a whole continent full of vacuume tubes and wires. Yes,
trying to build that much communication speed in that day and age would not be
an easy or wise thing to do. Likewise with this simulation scenario. For the
time being, we have no earthly idea how to deal with those issues because we
don't know the details. That doesn't mean there aren't issues we -can- hammer
out.
Thanks again for your feedback. You're an individual of intelligence, and I
appreciate your taking the time to respond to these what must seem like "naive"
queries of mine.
Oh, no, I know where this is going. First the compliments, then some
remarks about how the light and my eyes. Then chocolate and flowers.
Pretty soon I am carrying your baby and crying while you go out
drinking with the boys. Sorry, but I've been around enough to be
careful.
LoL! Ah, well, I guess I'll try someone else. ;-)
--
L8r,
Uncle Clover
************************************************
In my experience, one's degree of wisdom tends
to bear an exponentially inverse relationship
to one's outpouring of words.
Clearly, I've a _long_ way to go... ;-)
************************************************
The true mark of a civilized society is that its
citizens know how to hate each other peacefully.
************************************************
.
|
|
|
|
|
|
|
| User: "Conspiracy of Doves" |
|
| Title: Re: OT: The simulated human |
14 Nov 2006 10:17:45 AM |
|
|
Uncle Clover wrote:
Sooner or later if it hasn't happened already, computers are going to be
capable of simulating cellular activity right on down to the single-gene level
of DNA, RNA & all the other various molecules involved. When that happens, the
structure and behaviors of the organism encoded for should naturally emerge from
simply letting the cell function in a simulated environment, at least so long as
the expressed parameters of the organism and its environment do not exceed the
capabilities of the system upon which it is being run. This would doubtless
limit us to smaller, less complex organisms at first.
Even so, this sounds to me like a truly monumental computing task. Does
anyone have any good idea as to the particular system specs that would be
required, and how far we are from such capabilities? :-?
Even farther on down the road, we should be able to do this with more
complex organisms such as humans. A further curiosity of mine is this: Suppose
we encode and enact the formulation of a human in such a simulated environment.
Just one human alone would be quite an accomplishment, so to satisfy its social
needs, there could be other humans simulated into the environment but on a
less-complex level than the fully-simulated one (i.e., the computer wouldn't
need to simulate them on a cellular level, just perhaps down to the resolution
of the major organs and organ groups, their personality being purely AI). If we
took such a human and left it grow up in such an environment and provided it
with an education that's considered good at least by today's standards, how
might it deduce the circumstances of its own existence? Would you favor an
open-ended approach where the fact of its simulated nature is made known to it
as it grows so that it never really has to go through any major shock? Or do
you think it would be a better idea to keep the nature of its existence hidden
from it?
For my own self, I would think it has the right to know. The problem is
that if you tell it that from "birth", that may interfere with the development
of its psychology in some way. If one is looking for the most natural
psychological development possible, I think that might be impossible if the
person knows it's not in and it may never get to -be- in "the real world". And
yet there seems no one single "good time" to tell it during the course of a
lifetime, no single "magical moment" where it would go from being "not ready" to
"ready". Perhaps if its simulated nature were presented to it in a toned-down
way. Like the way we learn over time that there are other countries and other
places with people who are different from us and cultures that would be strange
to us, and that we may never get to see for ourselves. It could be presented
with its place in a simulation being described as just another location -
instead of being from a small town in Pennsylvania, it would be from Simville,
Iowa (or whatever state the simulation is being run in). It could also be
taught that its condition as a simulated being is just another condition some
people are born with and some are not. That's a natural enough way for many of
us to grow up - realizing there's something "different" about us that makes us
not quite like everybody else. It wouldn't need to know right away that it's
the only one (assuming it is).
I know that if I were a simulated being, I could understand my
simulators not wanting to tell me right away so that my psychology could form
more naturally, but I feel I would have the right to know it -eventually-.
After my psychology has firmed up a good bit, I don't think that sort of
knowledge could possibly do any harm. But that's just me, I guess...
Anyway, lots and lots of issues to consider with this. I know it's a
long way off yet, but I've a feeling the capacity for such sophisticated
simulation is perhaps much closer than it seems.
Just curious as always...
Ever read Greg Egan's "Permutation City"?
http://tinyurl.com/ydm2x2
.
|
|
|
| User: "Uncle Clover" |
|
| Title: Re: OT: The simulated human |
14 Nov 2006 10:34:10 AM |
|
|
On 14 Nov 2006 08:17:45 -0800, "Conspiracy of Doves" <mark_dp73@yahoo.com>
wrote:
<snip>
Ever read Greg Egan's "Permutation City"?
http://tinyurl.com/ydm2x2
No, but that's a very fascinating premise. Thanks for the pointer, I'm
definitely going to be checking it out soon! :-)
--
L8r,
Uncle Clover
************************************************
In my experience, one's degree of wisdom tends
to bear an exponentially inverse relationship
to one's outpouring of words.
Clearly, I've a _long_ way to go... ;-)
************************************************
The true mark of a civilized society is that its
citizens know how to hate each other peacefully.
************************************************
.
|
|
|
| User: "Conspiracy of Doves" |
|
| Title: Re: OT: The simulated human |
14 Nov 2006 11:46:55 AM |
|
|
Uncle Clover wrote:
On 14 Nov 2006 08:17:45 -0800, "Conspiracy of Doves" <mark_dp73@yahoo.com>
wrote:
<snip>
Ever read Greg Egan's "Permutation City"?
http://tinyurl.com/ydm2x2
No, but that's a very fascinating premise. Thanks for the pointer, I'm
definitely going to be checking it out soon! :-)
Be sure to check out some of his other books like Distress and
Diaspora. Quarantine is pretty good too.
I'd give Teranesia a miss, though.
.
|
|
|
|
|
|
|
Related Articles |
|
|
