[FRIAM] Fontana 1996: Barrier to Objects (was Re: anthropological observations)

Sat Apr 18 13:27:55 EDT 2020

Since then, linear logic has made its way into the practice of software engineering:  Rust’s linear types, and Mercury’s uniqueness modes<https://www.mercurylang.org/information/doc-latest/mercury_ref/Unique-modes.html> and Idris<http://docs.idris-lang.org/en/latest/reference/uniqueness-types.html> uniqueness types.    It goes back quite a ways, at least to Linear Lisp<http://home.pipeline.com/~hbaker1/LinearLisp.html>.   It makes sense for variables that are actually reactants.

From: Friam <friam-bounces at redfish.com> on behalf of Stephen Guerin <stephen.guerin at simtable.com>
Reply-To: The Friday Morning Applied Complexity Coffee Group <friam at redfish.com>
Date: Saturday, April 18, 2020 at 9:44 AM
To: The Friday Morning Applied Complexity Coffee Group <friam at redfish.com>, "Mohammed at Optomatica. Com" <mohammed at optomatica.com>
Subject: [FRIAM] Fontana 1996: Barrier to Objects (was Re: anthropological observations)

Eric,

Even if your quote wasn't from Barrier to Objects, it was fun for me to pull it back up. I worked through that one pretty closely in 2000 and I'm reading it in a new light this morning. I remember how beautiful the tutorial appendix (https://scholar.harvard.edu/files/walterfontana/files/object.appendices.pdf) was as well and do remember this quote had an impact on me which impelled me to focus decentralized architectures on processes over functions.
The world of functions and the world of processes emphasize the halting problem
differently. While termination is a desideratum for functions or algorithms, the
opposite is typically true for processes. There one looks for conditions under
which a community of processes is guaranteed never to dead-lock, as there are
many situations where ongoing communication or interactivity is required. Examples
include operating systems, whether in air traffic control systems, computer
systems, mobile telephone networks, or...living and cognizing systems. The focus
on the absence of dead-lock shifts the attention from computation to organization.
This clearly locates concurrency very close to our project.

There was a distinct lunch I remember with Bill Macready and Mohammed El-Beltagy (cc'd) in 2000 where Bill was describing the one instance of getting muli-level emergence in a computational system via a pi-calculus implementation but said he never wrote it up. 20 years later, I still consider this a grail. Mohammed, did you ever see an implementation from Bill?

The tutorial also was another encouragement to read Milner but never fully digested. Also thought of diving into OCaml but never could focus enough attention. Mohammed, you messed around with it more that I as I recall.  I've always been too focused on deployability. That said, a lot of our thinking around our work on "Acequia", a decentralized network as a distributed graph passing Javascript functions through WebRTC ports with network graph rewriting topologies takes a lot of inspiration from this description in Walter's Barrier to Objects appendix and Milner's book.:
[cid:image001.png at 01D6156B.FCF16210]
.

-Stephen
_______________________________________________________________________
Stephen.Guerin at Simtable.com<mailto:stephen.guerin at simtable.com>
CEO, Simtable  http://www.simtable.com<http://www.simtable.com/>
1600 Lena St #D1, Santa Fe, NM 87505
office: (505)995-0206 mobile: (505)577-5828
twitter: @simtable
zoom.com/j/5055775828<http://zoom.com/j/5055775828>

On Sat, Apr 18, 2020 at 12:59 AM David Eric Smith <desmith at santafe.edu<mailto:desmith at santafe.edu>> wrote:
I don’t know, Steve,

I looked at that one, and at all the early ones I could find, and in quick skimming I didn’t find what I thought was a quote in the epigraph position.  I am beginning to wonder if it was a draft of something that never got published in the manuscript version I saw.  A pity if so.

I also looked for poetic quotes on “Ever focused on objects”, but the only google hits I got were a bunch of studies on autistic kids.  Poetic in a different sense, but now what I was looking for.

Also, it turns out I put in a dud link to Rota’s phenomenology lectures; apologies.  A link that at least goes to a first page is here:
https://www.pdcnet.org/nyppp/content/nyppp_2008_0008_0225_0319

You are up either very early, or very late.

Best,
E

On Apr 18, 2020, at 3:55 PM, Stephen Guerin <stephen.guerin at simtable.com<mailto:stephen.guerin at simtable.com>> wrote:

Eric,

Was it Barrier to Objects?
https://scholar.harvard.edu/walterfontana/publications/barrier-objects-dynamical-systems-bounded-organizations

That was the constructivist lambda calculus paper. Bill Mckelvey extended to pi calculus

On Sat, Apr 18, 2020, 12:36 AM David Eric Smith <desmith at santafe.edu<mailto:desmith at santafe.edu>> wrote:
Very good Nick.

You see, unfortunately it appears that the reason I was put on Earth was to be the evangelist of distributional thinking.

In one of Walter Fontana’s early papers, which I probably saw in 1998, he opened with a quote “Ever focused on objects, we something something something…(some expression of loss)”.  This was in his Lambda-calculus papers about the concreteness of realized patterns that are not objects, and their fundamental role for biology.  For as much as I like it, you would think I had remembered either the text or the source.  Cannot find it now.

But, to your point:

I think where the discussion happens is not about knowledge, or even regularity, but rather how wide and how flexible a scope you are willing to cast for what counts as an “object of knowledge”.  Or even “objects in knowledge”.

Yes, the values taken at events can be very good things to have found out about.  They inhabit the past, and our sense of knowing them is heavily wrapped up in both the senses of “the past” and of “knowledge".  It is a very small set of cases that are so constrained that the future may as well be in the past.  Nonetheless, the longing for it seems to be an eternal wellspring for delusions.  The Popol Vu has something about, for the adepts, ’The future and the past are laid out before them [like symmetric spatial dimensions]’ or something to that effect.  Lakoff probably can cite no end of metaphors by which people have mapped between the two, conceiving of time as having the same symmetric availability as space.  I expect it is a human cognitive and cultural universal.

But what happens when the future really is different from the past?  Do we insist that every “real” object of knowledge about the future must have a model in only the most singular of things archived from the past?  I would say no.  There are lots of cases in which the outcome delivered by an event not yet performed is not available for knowing.  How you plan to sample, though, and features of the distribution from which you will sample, may be very good things to know.  Back in Ancient Greece, we could have argued interminably about whether a distribution is less privileged as an “object” of knowledge than the particular value yielded by a sample from the distribution.

But a lot has happened since Ancient Greece, and today we have many many reasons to see them as deserving peers, and even to be cautious that we may not be able to tell them apart.

Entropy in thermodynamics is a distributional concept, yet it does very very much of the work in the world that we used to ascribe to Newtonian objects.

In high-energy physics, post Gell-Mann/Wilson (so 1954 Gell-Mann and Lowe, Wilson 1974), we have learned that everything we used to think _were_ objects, turned out to be distributions.  In hindsight this was of crucial conceptual importance.  If objects had been primary, and distributions had been mere step-children when we could not pin things down, and that had been _all_ there was to our science, we would have suffered an infinite regress.  Until we had a Theory of Everything, or a bottoming out of the well of smallness, we could never know if the science was predictively closed.  But now with some understanding of phase transitions, we know that the world could as well be distributions all the way down forever (or it might not be; it might have a bottom), and the foundation of _any_ of the predictive science we currently use would not be any worse in one case than in another.  They are not currently “exactly” closed, but we can put bounds on how closed they must be.  Everything is Probably Approximately Correct (Leslie Valiant), and that was all we had ever had.  It was more valuable to learn that there are ceilings and floors in the scope of influence of variations within distributions, than whether there is any smallest level of objects, or even any need for a concept of “object” distinct from what we can do with distributions.  A short incantation that I use to ward off the vampires who mis-use the word “reductionism” is that “Only with a theory of emergence did reductionist science become well-founded."

Biology has been conceptually impaired by too literalist a view of objects, whether organisms for Darwin, genes for Williams and Dawkins, or whatever other “unit of selection” you want to use as a shibboleth.  People fret over whether “viruses are alive”, having already committed that “alive” must a predicate defined over objects, and they worry whether there “really are” any individuals, since material is always coming and going and there are more bacterial cells in my gut than human cells in the rest of me.  Habits of understanding that determinism can dwell in the distribution opens a treasure chest of methods but also styles of thought, with which all these “not-even-wrong” frets simply dissipate the same way we no longer agonize over Zeno paradoxes.

I have no gripe with object-oriented thinking, or event-outcome-oriented thinking; we can do much with those, and they account for a lot of our animal habit and our “folk physics”.  But to put it up as a gold standard is very limiting.  We know lots of things that cannot be done within that frame, but that can be done, and some things where we thought it was the right frame and we were wrong.

There was a source I thought of putting on the list early in this thread, here:
https://philpapers.org/rec/ROTLOB
I have seen a copy of this, but I don’t know where to get a legally distributable copy and this is either paywalled or not even electronic.  Some of you may have it already.

It was when Dave gave the assertion that the rural people are actually the careful balanced thinkers, and Frank put up an article as “another perspective, or perhaps David will see it as confirming evidence”.

I know probably most of you have read Heidegger, and Husserl, and Fink, and lots of others.  I had not.  So I found Rota’s notes, structured by Heidegger but drawing in many of their good parts on Husserl, helpful to sort-of recognize what the phenomenologists are on about.

In one of the late lectures, Rota explains the phenomenologists strong emphasis that no mere events in our existence have any particular meaning.  There can be the sequence of lines in a proof, or a recipe for doing an experiment.  By themselves, they are just artifacts, inscribed in a library somewhere.  Even read through, or performed, they may just be motions in nature.  They become “a proof” or “evidence”, when they are experienced as evidence-for a truth or a bit of knowledge.  This concept of “perceiving-as” or “evidence-for”, the phenomenologists claim, is simply different in kind from any of the procedures in which it occurs.  If I understand them, they assert that the moment of experiencing something “as” in fact defines an experiential notion of the temporal present that is different entirely in kind from the notations of either the past or the future.

It is a bit of a digression from this post, but I will remark, that this position makes the world look pretty hopeless to me, since anyone can experience anything “as” evidence for anything.  And there is a part of reality-building in those moments for them, that nobody else outside them has any grip on.

But in a more positive note, and on the point of this post, I feel like it is a Husserlian/Heideggerian shift in the occupancy of the temporal present, to find it as normal to experience distributionally-defined patterns as objects, as event-outcome-defined patterns.  They just do different things.

Anyway, sorry.  Big long TLDR to state the obvious.

Eric

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://redfish.com/pipermail/friam_redfish.com/attachments/20200418/06b22a11/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.png
Type: image/png
Size: 143741 bytes
Desc: image001.png
URL: <http://redfish.com/pipermail/friam_redfish.com/attachments/20200418/06b22a11/attachment-0001.png>