Friam

Re: Fwd: Latest issue of JASSS appears to contain a nugget

Classic

List

Threaded

4 messages Options

Nick Thompson

Re: Fwd: Latest issue of JASSS appears to contain a nugget

Jim,

Actually, I wasnt entirely clear what the controversy was and was more
concerned by how a Great Man like Epstein might be used as the authority
for unbearable round of obscurantist blather about how one could have
"understanding" without being able to generate any reasonable expectations
about the state of the world. Think about it: a phalanx of
post-modernists led by Joshua Epstein. If you think I am being paranoid,
you only have to remember what silly hay was made of Kuhn's philosophy of
science or, for that matter, einstein's relativity.

Anyway: have a look at Epstein's original article and see if it doesnt
inspire similar worries in you. Note that he has written a much longer
work on "generative" science, as if all science were not generative. I get
the impression he keeps meeting idiots at cocktail parties, and thinks of
these idiots as representative of the way the sane world think of the
relation between agent based modeling and other forms of scientific
thinking. Agent based modeling is no more subject to the fallacies of
induction or the fallacy of affirming the consequent than any other form of
modeling.

Take care,

Nick

Nicholas S. Thompson
Emeritus Professor of Psychology and Ethology,
Clark University ([hidden email])

> [Original Message]
> From: Jim Gattiker <[hidden email]>
> To: <[hidden email]>
> Date: 1/31/2009 12:19:11 PM
> Subject: Fwd: [FRIAM] Latest issue of JASSS appears to contain a nugget
>
> > And lets not overlook the brewing dogfight on prediction:
> > - Contra Epstein, Good Explanations Predict (our own Nick Thompson)
> > - Not All Explanations Predict Satisfactorily, and Not All Good
Predictions Explain

>
> What's the controversy?
>
> On the first, I suppose one could drum something up over unstated
> definitions of "good". After all, if we can't argue over semantics,
> what can we argue over?
>
> I can't construe anything difficult about the second one at all.
>
>
> ---------- Forwarded message ----------
> From: Owen Densmore <[hidden email]>
> Date: Sat, Jan 31, 2009 at 10:15 AM
> Subject: Re: [FRIAM] Latest issue of JASSS appears to contain a nugget
> To: The Friday Morning Applied Complexity Coffee Group <[hidden email]>
>
>
> And lets not overlook the brewing dogfight on prediction:
> - Contra Epstein, Good Explanations Predict (our own Nick Thompson)
> - Not All Explanations Predict Satisfactorily, and Not All Good
> Predictions Explain
>
> Also..cast your vote for pdf vs html for paper formats!
>
> Great stuff.
>
> -- Owen
>
>
> On Jan 31, 2009, at 8:45 AM, Robert Holmes wrote:
>
> > Amongst the usual plethora of validation-by-kinda-looks-like papers in

this
> > month's JASSS is a paper that looks like it might be useful:
*"**Techniques

> > to Understand Computer Simulations: Markov Chain Analysis"*
> > http://jasss.soc.surrey.ac.uk/12/1/6.html. The paper includes the source
> > code for the 10 "classic" ABMs that they consider.
> >
> > Robert
> > ============================================================
> > FRIAM Applied Complexity Group listserv
> > Meets Fridays 9a-11:30 at cafe at St. John's College
> > lectures, archives, unsubscribe, maps at http://www.friam.org
>
>
> ============================================================
> FRIAM Applied Complexity Group listserv
> Meets Fridays 9a-11:30 at cafe at St. John's College
> lectures, archives, unsubscribe, maps at http://www.friam.org

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org

Owen Densmore

Re: Fwd: Latest issue of JASSS appears to contain a nugget

Administrator

Nick -- I'm wondering: would it be a good idea for you to give a talk
at the Complex on the paper? As much as we discussed it, I found
myself a bit confused about the explanation vs prediction debate.

Partly it has to do with the difference with simulation (like TranSim)
modeling and smaller scale, concept modeling (like Prisoner's Dilemma).

Partly it has to do with the explanation vs prediction difference. PD
can explain the evolution of cooperation. TS can predict traffic
patterns.

Finally, it may have to do with the stochastic nature of modeling, and
if our set-theoretic reduction of Steve's thermodynamic interpretation
of models works, we may get to something solid enough for me to admit
to prediction.

-- Owen

On Jan 31, 2009, at 2:21 PM, Nicholas Thompson wrote:

> Jim,
>
> Actually, I wasnt entirely clear what the controversy was and was more
> concerned by how a Great Man like Epstein might be used as the
> authority
> for unbearable round of obscurantist blather about how one could have
> "understanding" without being able to generate any reasonable
> expectations
> about the state of the world. Think about it: a phalanx of
> post-modernists led by Joshua Epstein. If you think I am being
> paranoid,
> you only have to remember what silly hay was made of Kuhn's
> philosophy of
> science or, for that matter, einstein's relativity.
>
> Anyway: have a look at Epstein's original article and see if it
> doesnt
> inspire similar worries in you. Note that he has written a much
> longer
> work on "generative" science, as if all science were not
> generative. I get
> the impression he keeps meeting idiots at cocktail parties, and
> thinks of
> these idiots as representative of the way the sane world think of the
> relation between agent based modeling and other forms of scientific
> thinking. Agent based modeling is no more subject to the fallacies of
> induction or the fallacy of affirming the consequent than any other
> form of
> modeling.
>
> Take care,
>
>
> Nick
>
>
>
>
>
>
>
> Nicholas S. Thompson
> Emeritus Professor of Psychology and Ethology,
> Clark University ([hidden email])
>
>
>
>
>> [Original Message]
>> From: Jim Gattiker <[hidden email]>
>> To: <[hidden email]>
>> Date: 1/31/2009 12:19:11 PM
>> Subject: Fwd: [FRIAM] Latest issue of JASSS appears to contain a
>> nugget
>>
>>> And lets not overlook the brewing dogfight on prediction:
>>> - Contra Epstein, Good Explanations Predict (our own Nick Thompson)
>>> - Not All Explanations Predict Satisfactorily, and Not All Good
> Predictions Explain
>>
>> What's the controversy?
>>
>> On the first, I suppose one could drum something up over unstated
>> definitions of "good". After all, if we can't argue over semantics,
>> what can we argue over?
>>
>> I can't construe anything difficult about the second one at all.
>>
>>
>> ---------- Forwarded message ----------
>> From: Owen Densmore <[hidden email]>
>> Date: Sat, Jan 31, 2009 at 10:15 AM
>> Subject: Re: [FRIAM] Latest issue of JASSS appears to contain a
>> nugget
>> To: The Friday Morning Applied Complexity Coffee Group <[hidden email]
>> >
>>
>>
>> And lets not overlook the brewing dogfight on prediction:
>> - Contra Epstein, Good Explanations Predict (our own Nick Thompson)
>> - Not All Explanations Predict Satisfactorily, and Not All Good
>> Predictions Explain
>>
>> Also..cast your vote for pdf vs html for paper formats!
>>
>> Great stuff.
>>
>> -- Owen
>>
>>
>> On Jan 31, 2009, at 8:45 AM, Robert Holmes wrote:
>>
>>> Amongst the usual plethora of validation-by-kinda-looks-like
>>> papers in
> this
>>> month's JASSS is a paper that looks like it might be useful:
> *"**Techniques
>>> to Understand Computer Simulations: Markov Chain Analysis"*
>>> http://jasss.soc.surrey.ac.uk/12/1/6.html. The paper includes the
>>> source
>>> code for the 10 "classic" ABMs that they consider.
>>>
>>> Robert
>>> ============================================================
>>> FRIAM Applied Complexity Group listserv
>>> Meets Fridays 9a-11:30 at cafe at St. John's College
>>> lectures, archives, unsubscribe, maps at http://www.friam.org
>>
>>
>> ============================================================
>> FRIAM Applied Complexity Group listserv
>> Meets Fridays 9a-11:30 at cafe at St. John's College
>> lectures, archives, unsubscribe, maps at http://www.friam.org
>
>
>
> ============================================================
> FRIAM Applied Complexity Group listserv
> Meets Fridays 9a-11:30 at cafe at St. John's College
> lectures, archives, unsubscribe, maps at http://www.friam.org

Nick Thompson

Re: Fwd: Latest issue of JASSS appears to contain a nugget

In reply to this post by Nick Thompson

Owen,

Thank you for this kind note.

I am, as you know, overjoyed to talk about this stuff any time. I would
sit down with you and anybody else any weekday at any coffee house to
explore the issues you raise. On the other hand, for me to "Give A Talk"
would require some serious preparation, of the order of six weeks, I would
think. I don't mind. I would be honored to do it. It's probably time I
put something together. I think the systematic deployment of scientific
metaphors of all kinds is fascinating to explore. Whether there is
anything special about ABM's I am not quite sure. But they are subject to
the same disability as any scientific explanation: they are "bold
conjectures" that cannot be confirmed but only falsified ... if even that!

It will be interesting to see what Epstein says. I think his paper was not
so much wrong as VERY careless. I am not familiar with all the models you
mention in your note, but I am prepared to bet you that it will be easy to
show that just to the extent that they are explanatory, they are also
predictive.

I agree with the second paper that the issue is largely a semantic one.
Unfortunately, I was never able to find simple words to describe the
difference between prediction as embraced by the philosophers of science
familiar to me and prediction as feared by Epstein. The second commentary
confuses the issue by attributing to Derr and myself the view that
"earthquakes will occur" is an example of a scientific prediction. Of
course that is stupid. Somewhere between such a vacuous prediction and the
kinds of predictions that Epstein fears he will be called upon to provide
(Jean Dixon says "San Francisco will fall into the sea next Tuesday at
11.31 am") is an enormous range of refutable expectations concerning
earthquakes that can and should be tested through observation and even
experiment. This middle ground is what I meant be scientific predications.

Let's Tawk. You might find that a few minutes of coffee drinking might
rapidly sate your appetite on this subject.

All the best,

Nick

_
Nicholas S. Thompson
Emeritus Professor of Psychology and Ethology,
Clark University ([hidden email])

> [Original Message]
> From: Owen Densmore <[hidden email]>
> To: <[hidden email]>; The Friday Morning Applied Complexity
Coffee Group <[hidden email]>
> Date: 1/31/2009 7:53:31 PM
> Subject: Re: [FRIAM] Fwd: Latest issue of JASSS appears to contain a
nugget

>
> Nick -- I'm wondering: would it be a good idea for you to give a talk
> at the Complex on the paper? As much as we discussed it, I found
> myself a bit confused about the explanation vs prediction debate.
>
> Partly it has to do with the difference with simulation (like TranSim)
> modeling and smaller scale, concept modeling (like Prisoner's Dilemma).
>
> Partly it has to do with the explanation vs prediction difference. PD
> can explain the evolution of cooperation. TS can predict traffic
> patterns.
>
> Finally, it may have to do with the stochastic nature of modeling, and
> if our set-theoretic reduction of Steve's thermodynamic interpretation
> of models works, we may get to something solid enough for me to admit
> to prediction.
>
> -- Owen
>
>
> On Jan 31, 2009, at 2:21 PM, Nicholas Thompson wrote:
>
> > Jim,
> >
> > Actually, I wasnt entirely clear what the controversy was and was more
> > concerned by how a Great Man like Epstein might be used as the
> > authority
> > for unbearable round of obscurantist blather about how one could have
> > "understanding" without being able to generate any reasonable
> > expectations
> > about the state of the world. Think about it: a phalanx of
> > post-modernists led by Joshua Epstein. If you think I am being
> > paranoid,
> > you only have to remember what silly hay was made of Kuhn's
> > philosophy of
> > science or, for that matter, einstein's relativity.
> >
> > Anyway: have a look at Epstein's original article and see if it
> > doesnt
> > inspire similar worries in you. Note that he has written a much
> > longer
> > work on "generative" science, as if all science were not
> > generative. I get
> > the impression he keeps meeting idiots at cocktail parties, and
> > thinks of
> > these idiots as representative of the way the sane world think of the
> > relation between agent based modeling and other forms of scientific
> > thinking. Agent based modeling is no more subject to the fallacies of
> > induction or the fallacy of affirming the consequent than any other
> > form of
> > modeling.
> >
> > Take care,
> >
> >
> > Nick
> >
> >
> >
> >
> >
> >
> >
> > Nicholas S. Thompson
> > Emeritus Professor of Psychology and Ethology,
> > Clark University ([hidden email])
> >
> >
> >
> >
> >> [Original Message]
> >> From: Jim Gattiker <[hidden email]>
> >> To: <[hidden email]>
> >> Date: 1/31/2009 12:19:11 PM
> >> Subject: Fwd: [FRIAM] Latest issue of JASSS appears to contain a
> >> nugget
> >>
> >>> And lets not overlook the brewing dogfight on prediction:
> >>> - Contra Epstein, Good Explanations Predict (our own Nick Thompson)
> >>> - Not All Explanations Predict Satisfactorily, and Not All Good
> > Predictions Explain
> >>
> >> What's the controversy?
> >>
> >> On the first, I suppose one could drum something up over unstated
> >> definitions of "good". After all, if we can't argue over semantics,
> >> what can we argue over?
> >>
> >> I can't construe anything difficult about the second one at all.
> >>
> >>
> >> ---------- Forwarded message ----------
> >> From: Owen Densmore <[hidden email]>
> >> Date: Sat, Jan 31, 2009 at 10:15 AM
> >> Subject: Re: [FRIAM] Latest issue of JASSS appears to contain a
> >> nugget
> >> To: The Friday Morning Applied Complexity Coffee Group

<[hidden email]

> >> >
> >>
> >>
> >> And lets not overlook the brewing dogfight on prediction:
> >> - Contra Epstein, Good Explanations Predict (our own Nick Thompson)
> >> - Not All Explanations Predict Satisfactorily, and Not All Good
> >> Predictions Explain
> >>
> >> Also..cast your vote for pdf vs html for paper formats!
> >>
> >> Great stuff.
> >>
> >> -- Owen
> >>
> >>
> >> On Jan 31, 2009, at 8:45 AM, Robert Holmes wrote:
> >>
> >>> Amongst the usual plethora of validation-by-kinda-looks-like
> >>> papers in
> > this
> >>> month's JASSS is a paper that looks like it might be useful:
> > *"**Techniques
> >>> to Understand Computer Simulations: Markov Chain Analysis"*
> >>> http://jasss.soc.surrey.ac.uk/12/1/6.html. The paper includes the
> >>> source
> >>> code for the 10 "classic" ABMs that they consider.
> >>>
> >>> Robert
> >>> ============================================================
> >>> FRIAM Applied Complexity Group listserv
> >>> Meets Fridays 9a-11:30 at cafe at St. John's College
> >>> lectures, archives, unsubscribe, maps at http://www.friam.org
> >>
> >>
> >> ============================================================
> >> FRIAM Applied Complexity Group listserv
> >> Meets Fridays 9a-11:30 at cafe at St. John's College
> >> lectures, archives, unsubscribe, maps at http://www.friam.org
> >
> >
> >
> > ============================================================
> > FRIAM Applied Complexity Group listserv
> > Meets Fridays 9a-11:30 at cafe at St. John's College
> > lectures, archives, unsubscribe, maps at http://www.friam.org

Ted Carmichael

Re: Fwd: Latest issue of JASSS appears to contain a nugget

Hi, Nick.

I haven't read the papers yet, so maybe I am misunderstanding you when you say:

but I am prepared to bet you that it will be easy to
show that just to the extent that they are explanatory, they are also
predictive.

You go on to mention earthquakes, which was the first thing I thought of when I read the above.

I guess the part I don't get is, aren't a lot of phenomena like this? I mean, you could construct a model that perfectly explains earthquakes, but won't be able to predict them. I'm assuming here that the model in question "explains" and "predicts" the real world ... but often in the real world we do not have all the measurements we need to predict.

So in this sense, even a perfect explanatory model would not be able to predict some things, simply because it lacks the right inputs.

What point am I missing here?

Thanks,

Ted

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org