The Art of Simulation, an Old ( & Odd) Kind of Science

[Friam mailing list]

Simulators,

Stephen Guerin writes:
 > we need to need to think about a good first lecture for the monthly
lecture at SFI.

An endless & valuable discussion question that FRIAM has already touched on
is, how do you "scientifically" know that you are capturing enough in your
Model?
What's the Scientific theory, background to this attempt?
Seems like kind of a muddle to me, the Scientific theory of Scientific
Simulation,
but I'm an atheist about Science, maybe that's the problem.
Unfortunately, it seems hard to guarantee to a business
just what kind of modeling they are going to get when they
employ a "New Science" approach. Quacks-looks like a Duck, is it a Duck?
Seems like an "Old Kind of Science" to me, an Art.
Maybe even applying the Scientific method, doing Science, is an Art.
After all the media hoopla, I feel that many businesses have
De-validated their genuine Science, and genuine Scientists, by claiming for
years a
"New Kind" of Scientific modeling, using
"Computer Science", "in Silico", "Would be Worlds".
Wouldn't it be helpful to at least admit the problem?
And isn't that a business idea?, (the validation of models). {Lanny's
ValidoModel Inc.?}
Shouldn't SFI, in it's educational-leadership role, take some responsibility
for this, or at least the other reasons for the dismal business climate in
Santa Fe?
Politically correct now, also, as the public recognizes that they have been
force-fed an inefficient bunch of new technology hype for a while now.
Especially in this town, with it's empty Ivory Towers.

-------------Warning, "#vent long standing rant if notTooBusyForRanting:"
-----------

An example of emergence is that you cannot predict the outcome of any law.
A law is an attempt out of a model to create change,
and you simply can't tell the consequences of any law,
as the model can always fail on you, especially if incomplete.
This applies to any action you take, or any system you study.
Aren't all models by definition, incomplete? Answers are possibilities,
percentages.
It's impossible to know for certain what variable, hidden variable,
random factor, you are leaving out,
or what artifacts you are putting in accidentally on your own.

How DO you build a model a business process?
It's been said that  "good enough" systems are better than "excellent",
that one should "Pursue no optima" as theoretical optima are useless if
they depend
on a steady state that is never achieved.
I well remember a major glitch in a commodity modeling
project that I heard a lot about was that the truck Drivers were lying about
their delivery schedules and volume. How do you account for that?
Also bothers middle management when you point it out in addition
to demonstrating how shoddy their data is in other ways.
Bad for a good working relationship.

Models of new airplanes are still made,
and it's "throw it up in the air and see if it flies", in this day & age!
Recently in a science journal I saw a geology article where
terran mantle plumes were modeled with different colored & density glycerin in
a desktop FishTank, NOT a desktop SuperComputer.
Faster, Cheaper, Better.

Business is not even air, it changes more. The Stock Market, for instance,
always changes.
Yes, the answer is to do a number of runs of your model,
apply a % of random events and then apply statistics, doesn't that get
foggy quick?
Why else would it be possible to bet on football?

In another venue, I paint "a Dancer", other people see "a bed of Flowers".
When that happens, I call the piece "Flower Dance" & hope that sells!

                     -Simulation is an Art, not a Science-

-------------End of, "#vent long standing rant if notTooBusyForRanting:"
-----------

---------------------------------------------------
 From the April 1980 Washington Monthly: "Beam Me Out Of This Death Trap,
Scotty"
By Gregg Easterbrook

"The science of ballistics is much more
precise and predictable than the art of flying....These are the wild, uncharted
rivers of space. Unknown; unknowable; beyond programming.
To find out if your ship can cope with them, you have to take it up there."
---------------------------------------------------
 From SFI: "Scientific Models: Claiming and Validating"
By Cosmo Shalizi

"Some of the speakers professed to be mere slaves of their data, others
confessed to prolonged trial and error. Nobody said anything very original
about how models get built, or how they should be, and it was probably
unreasonable to expect otherwise-good musicians are rarely good musicologists.

...It is not at all clear what validation would even begin to look like...

...in particular cases (it) would seem to lie in studying the details of
that particular
system, squashing the hope of finding general,
details-independent  mechanisms."
---------------------------------------------------
From: "The Simply Complex"
By John Casti
"there is no single, best way to process information"
---------------------------------------------------

Regards,

Lanny Bear

[Friam mailing list]

  >>Maybe even applying the Scientific method, doing Science, is an Art.

Ah, but if science is Art, what is Art?
I'll know it when I see it?  Do we need to develop, after all, aesthetic
criteria for complexity wranglers?  I would propose elegance as
such a criteria, but that might be too easy an answer.  Alternatively,
said application is a "craft", which is certainly more appealing
from a programmer's perspective, but then we're back to aesthetics.

  >>How DO you build a model a business process?

A possibly adjunct deep question might be why?  (Aside from the income
stream.  Not advocating cynicicm here, merely keeping the question alive.)

  >>It's been said that  "good enough" systems are better than "excellent",
  >>that one should "Pursue no optima" as theoretical optima are useless if
  >>they depend on a steady state that is never achieved.

Even more interesting, what do they show if the *model* steady state is
never acheived?  When said model changes, how do any
experiments/optimizations
compare?

  >>inefficient bunch of new technology hype

I will set my hype optimizers on the problem... ;-)

  >>Shouldn't SFI, in it's educational-leadership role, take
  >>some responsibility
  >>for this, or at least the other reasons for the dismal
  >>business climate in Santa Fe?

It would be terrific if they would help define some roles and take
responsibility for some subset, situated against the FRIAM/Safari
axis of <whatever it is we're an axis/quaternion of>.

Carl

-----Original Message-----
From: [hidden email] [mailto:[hidden email]]On Behalf
Of Lanny H. Bear
Sent: Wednesday, February 12, 2003 10:19 AM
To: [hidden email]
Subject: [FRIAM] The Art of Simulation, an Old ( & Odd) Kind of Science



Simulators,

Stephen Guerin writes:
 > we need to need to think about a good first lecture for the monthly
lecture at SFI.

An endless & valuable discussion question that FRIAM has already touched on
is, how do you "scientifically" know that you are capturing enough in your
Model?
What's the Scientific theory, background to this attempt?
Seems like kind of a muddle to me, the Scientific theory of Scientific
Simulation,
but I'm an atheist about Science, maybe that's the problem.
Unfortunately, it seems hard to guarantee to a business
just what kind of modeling they are going to get when they
employ a "New Science" approach. Quacks-looks like a Duck, is it a Duck?
Seems like an "Old Kind of Science" to me, an Art.
Maybe even applying the Scientific method, doing Science, is an Art.
After all the media hoopla, I feel that many businesses have
De-validated their genuine Science, and genuine Scientists, by claiming for
years a
"New Kind" of Scientific modeling, using
"Computer Science", "in Silico", "Would be Worlds".
Wouldn't it be helpful to at least admit the problem?
And isn't that a business idea?, (the validation of models). {Lanny's
ValidoModel Inc.?}
Shouldn't SFI, in it's educational-leadership role, take some responsibility
for this, or at least the other reasons for the dismal business climate in
Santa Fe?
Politically correct now, also, as the public recognizes that they have been
force-fed an inefficient bunch of new technology hype for a while now.
Especially in this town, with it's empty Ivory Towers.

-------------Warning, "#vent long standing rant if notTooBusyForRanting:"
-----------

An example of emergence is that you cannot predict the outcome of any law.
A law is an attempt out of a model to create change,
and you simply can't tell the consequences of any law,
as the model can always fail on you, especially if incomplete.
This applies to any action you take, or any system you study.
Aren't all models by definition, incomplete? Answers are possibilities,
percentages.
It's impossible to know for certain what variable, hidden variable,
random factor, you are leaving out,
or what artifacts you are putting in accidentally on your own.

How DO you build a model a business process?
It's been said that  "good enough" systems are better than "excellent",
that one should "Pursue no optima" as theoretical optima are useless if
they depend
on a steady state that is never achieved.
I well remember a major glitch in a commodity modeling
project that I heard a lot about was that the truck Drivers were lying about
their delivery schedules and volume. How do you account for that?
Also bothers middle management when you point it out in addition
to demonstrating how shoddy their data is in other ways.
Bad for a good working relationship.

Models of new airplanes are still made,
and it's "throw it up in the air and see if it flies", in this day & age!
Recently in a science journal I saw a geology article where
terran mantle plumes were modeled with different colored & density glycerin
in
a desktop FishTank, NOT a desktop SuperComputer.
Faster, Cheaper, Better.

Business is not even air, it changes more. The Stock Market, for instance,
always changes.
Yes, the answer is to do a number of runs of your model,
apply a % of random events and then apply statistics, doesn't that get
foggy quick?
Why else would it be possible to bet on football?

In another venue, I paint "a Dancer", other people see "a bed of Flowers".
When that happens, I call the piece "Flower Dance" & hope that sells!

                     -Simulation is an Art, not a Science-

-------------End of, "#vent long standing rant if notTooBusyForRanting:"
-----------

---------------------------------------------------
 From the April 1980 Washington Monthly: "Beam Me Out Of This Death Trap,
Scotty"
By Gregg Easterbrook

"The science of ballistics is much more
precise and predictable than the art of flying....These are the wild,
uncharted
rivers of space. Unknown; unknowable; beyond programming.
To find out if your ship can cope with them, you have to take it up there."
---------------------------------------------------
 From SFI: "Scientific Models: Claiming and Validating"
By Cosmo Shalizi

"Some of the speakers professed to be mere slaves of their data, others
confessed to prolonged trial and error. Nobody said anything very original
about how models get built, or how they should be, and it was probably
unreasonable to expect otherwise-good musicians are rarely good
musicologists.

...It is not at all clear what validation would even begin to look like...

...in particular cases (it) would seem to lie in studying the details of
that particular
system, squashing the hope of finding general,
details-independent  mechanisms."
---------------------------------------------------
From: "The Simply Complex"
By John Casti
"there is no single, best way to process information"
---------------------------------------------------

Regards,

Lanny Bear


=========================================================
FRIAM Complexity Coffee listserv
Meets Fridays 9AM @ Museum Hill Cafe
Archives, unsubscribe, etc.:
http://www.redfish.com/mailman/listinfo/friam_redfish.com

[Friam mailing list]

Artists & Crafters,

LB: >>Maybe even applying the Scientific method, doing Science, is an Art.

CT:
 >
I would propose elegance as such a criteria (for doing Science),
but that might be too easy an answer.  Alternatively, said application is a=
=20
"craft"...
 >

-----------------------
Yes, I certainly agree!
-----------------------

 >>How DO you build a model a business process?

 >
A possibly adjunct deep question might be why?  (Aside from the income
stream.  Not advocating cynicism here, merely keeping the question alive.)
 >

-----------------------
Yes, absolutely, have to, decide, what modeling, if any, will further=20
understanding.
Of course, there's always something that can be optimized!
-----------------------

 >When said model changes, how do any experiments/optimizations compare?

-----------------------
From: "Industrial%20Science%20Overview.ppt" by George Danner
    http://www.industrial-science.com/

"We believe that the process of developing a model is
just as important as the numerical result."

"Models are much more than simply ways to compute numbers=85
they should be about communicating ideas.=94
We believe in not only solving problems but helping our clients
create their own analysis capability, long term.
Much of our =93deliverable=94 is in the form of a model that can be
used and updated by our clients long after the initial project is complete."
-----------------------
Thanks to REF for this pointer:
Donella Meadows' ... summarizes some very deep concepts of complex systems=
=20
theory.
http://www.wholeearthmag.com/ArticleBin/447.html
"Dancing with systems: What to do when systems resist change."

"Pay attention to what is important, not just what is quantifiable.

Our culture, obsessed with numbers, has given us the idea that
what we can measure is more important than what we can't measure.

...If something is ugly, say so.

...Celebrate complexity.

Let's face it, the universe is messy. It is nonlinear, turbulent, and
chaotic. It is dynamic. It spends its time in transient behavior on its way
to somewhere else, not in mathematically neat equilibria. It
self-organizes and evolves. It creates diversity, not uniformity. That's
what makes the world interesting, that's what makes it beautiful, and
that's what makes it work. "
-----------------------

Regards,

Lanny