description - explanation - metaphor - model

[Eric Charles-2]

Glen,

I mean... assuming I know what you mean by "obtuse"... which I'm not sure of... an "obtuse model" could be useful for many, many things... but the more obtuse it is, the less one can science with it... so it is not useful regarding where the future chapter of that book are headed. I would readily agree that being more explicit about that could help some readers come along with us into those later chapters. 

For example, an obtuse model could be useful in creating a clever lyrical scheme, or inducing intended confusion (such as economic models offered by politicians), or in spurring a creative writing assignment (many science fiction and fantasy gimmicks are intriguing in proportion to their obtuseness), etc. But if our goal is to act as a community to systematically test a scientific theory, we need some agreement... not perfect agreement, but certainly some agreement... regarding what the theory predicts. 

Bloodhound Gang, for example, has lyrics that are often a fascinating mix of transparent bits and obtuse bits that fit the rhythm beautifully.... here is a stanza from "Bad Touch"... 

Come quicker than Fed Ex, 
never reaching apex,
just like Coca-Cola stock you are inclined, 
to make me rise an hour early just like Daylight Savings Time  

Solid music (at least, I think so), but not exactly something to do science with. 

As for the "artifact" thing... it would be useful to have a term for what you are getting at... but Nick and I have never been able to settle on one, and I'm not sure "artifact" works. I want to talk about the "actual thing we are talking about", but at times that seems to make people think I'm asserting some sort of Kantian claim about the existence of a never-knowable "Ding an sich", which is not anywhere I want to go in the conversation. Several other possible terms seem to slip towards substance dualism. And then at some point Nick goes down a rabbit hole of insisting that it is "models all the way down", which might be true in some sense (maybe), but not in the sense we are discussing (see my slapping him around a few emails back in this chain). In the end, whether I am right or wrong, I think there is value in fighting to NOT create a technical term to serve that purpose, and just insist in every separate instance on being clear about the non-model thing we are talking about, i.e., in the conversation about how trains switch tracks, in which Nick starts to talk about how the model train switches track, the actual trains switching actual tracks in the actual thing we are trying to explain. I think there are some points in the book  as it stands, where I allow/use the term "explanandum" which is an already established term that gets at the distinction reasonably well, but I try to keep even that to a minimum. 

Best,

Eric

On Thu, Jan 16, 2020 at 1:29 PM uǝlƃ ☣ <[hidden email]> wrote:

[sigh] Fine. We can change what I wrote from:

  "artifact = model absent the usage context"
to
  "artifact = model in a non-modeling context"

The toy train isn't a useful example for this distinction. But a wooden sphere as a model for, say, a baseball, *is* a useful example. In the "sphere models baseball" context, "model" is properly used. But in another context, say, roll the wooden sphere down a ramp to measure gravity, the sphere is no longer a model and a word like "artifact" would be better when pointing to the sphere.

It's very difficult for me to imagine you *not* already having thought of this yourself. So, by "listening generously", I would have expected you to understand my phrases like "absent it's contextual analogies" and such. I feel the same way about my description of how obtuse models can be useful. It's difficult for me to imagine you haven't *already* considered parallax and expressibility. And although I appreciate playing at being naive, or practicing the Socratic method, part of "listening generously" is to "steel man" others' conceptual constructs (as opposed to "straw man").

In these 2 recent episodes, you could easily have imagined and described to *me* how obtuse models *might* be found useful. And you could easily have changed "absent context" to "in a non-modeling context".

On 1/16/20 8:59 AM, [hidden email] wrote:
> I am not sure I understand what you say here.  But I like the idea of "listening generously" and I am trying to do it.  I guess my problem in understanding is that I don't think we perceive anything other than in a context.  Like the gorilla walking through the basketball game, we just don't see it.  I don't think it's possible to see Eric and not see him intending.  (or, say, sleeping).  This may, in fact, be an argument in favor of your position.  I just haven't worked it out yet.

--
☣ uǝlƃ

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove

[gepr]

Re: the use of a special term like "artifact" or "explanandum", I agree completely. "Model" is as good as any.

Re: the usefulness of obtuse models - I did give a description of how obtuse (indeed, totally opaque) models can be useful for science. It's possible you didn't receive that post. So, here is the archive: http://friam.471366.n2.nabble.com/description-explanation-metaphor-model-tp7594030p7594294.html

It's arguable how large N must be for this to work well. But with progress in big data, generative AI, proof assistants, automatic programming, HoTT, etc., I think we're getting pretty close to shutting down any critics of the method. My own work requires only N=3: 1) the referents (i.e. validation data observed from it/them), 2) the reference model (usually an equation-based phenomena-only model), and 3) a finer-grained component-based model relying on both unit and systemic V&V. None of these are totally opaque, because we rely solely on open source stacks. (Though you could say none of us understands processors, cache, memory, transistors, etc.) But the *method* we're using prescribes that we *treat* them as opaque and rely solely on observations of each [†]. So, any validation/falsification we do can be reduced to data validation.

[†] I'm a broken record curmudgeon to my colleagues who keep treating verification data as if it were validation data. Pffft. FWIW, they also keep trying to use Matlab instead of Octave or R ... Grrr.

On 1/17/20 5:28 AM, Eric Charles wrote:
> I mean... assuming I know what you mean by "obtuse"... which I'm not sure of... an "obtuse model" could be useful for many, many things... but the more obtuse it is, the less one can science with it... [...]
--
☣ uǝlƃ

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove

[Frank Wimberly-2]

Relevant?

https://dl.acm.org/doi/10.5555/2431518.2431951

On Fri, Jan 17, 2020, 9:37 AM uǝlƃ ☣ <[hidden email]> wrote:

Re: the use of a special term like "artifact" or "explanandum", I agree completely. "Model" is as good as any.

Re: the usefulness of obtuse models - I did give a description of how obtuse (indeed, totally opaque) models can be useful for science. It's possible you didn't receive that post. So, here is the archive: http://friam.471366.n2.nabble.com/description-explanation-metaphor-model-tp7594030p7594294.html

It's arguable how large N must be for this to work well. But with progress in big data, generative AI, proof assistants, automatic programming, HoTT, etc., I think we're getting pretty close to shutting down any critics of the method. My own work requires only N=3: 1) the referents (i.e. validation data observed from it/them), 2) the reference model (usually an equation-based phenomena-only model), and 3) a finer-grained component-based model relying on both unit and systemic V&V. None of these are totally opaque, because we rely solely on open source stacks. (Though you could say none of us understands processors, cache, memory, transistors, etc.) But the *method* we're using prescribes that we *treat* them as opaque and rely solely on observations of each [†]. So, any validation/falsification we do can be reduced to data validation.

[†] I'm a broken record curmudgeon to my colleagues who keep treating verification data as if it were validation data. Pffft. FWIW, they also keep trying to use Matlab instead of Octave or R ... Grrr.

On 1/17/20 5:28 AM, Eric Charles wrote:
> I mean... assuming I know what you mean by "obtuse"... which I'm not sure of... an "obtuse model" could be useful for many, many things... but the more obtuse it is, the less one can science with it... [...]
--
☣ uǝlƃ

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove

============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove