[Wittrs] Re: Following a Rule
- From: "jrstern" <jrstern@xxxxxxxxx>
- To: wittrs@xxxxxxxxxxxxx
- Date: Fri, 18 Sep 2009 16:15:43 -0000
--- In Wittrs@xxxxxxxxxxxxxxx, Rob de Villiers <wittrs@...> wrote:
>
> > At the lowest level, the computer hardware implements
> > a causal structure, that I agree is non-rule-following.
> ....
>
> > Second, there are some things we *do* recognize as
> > rules. What are they, exactly? Interesting problem.
>
> Well, we *do* recognise them, as you say. So what is
> the problem?
Because some people, many people - most people! - have
eliminativist opinions of the matter, and the fact that
we do recognize rules, needs to be kept on the table.
> - "What are they, exactly?" Well, what
> exactly do you want? It sounds like you want an
> exact merkmal or categorical "necessary and sufficient
> contitions" type *definition*.
The good news is that, no, as a methodological consequence
of my nominalism, I do not much ask for categorical
definitions, I ask for *examples*.
>But what if our notions
> of rules and rule-following are "family-resemblance" type
> concepts (as indeed they are!) - then not only is the
> merkmal definition approach useless to getting clear
> about these matters but trying to force the concept into
> the mould of any assumed paradigmatic archetype is even
> more doomed to failure and confusion.
Oh, it may not be that bad, but as a nominalist I start
with the particulars, am suspicious of any abstract
category or natural kind, and certainly put no weight
on the use of any label.
> The only option
> is piecemeal, detailed conceptual investigation and
> clarification a-la Wittgenstein. In fact Wittgenstein
> has already answered your questions and disolved you
> problems regarding rules and rule-following and he does
> so primarily from round about PI $180 to $242... but
> contributions to this particular issue as scattered
> all over the place elswhere too.
Ahem. First, there is no part of Wittgenstein's work
more obscure in its meaning. Or use. Rob, I've been
pushing Shanker's books "Wittgenstein and the Turning Point
in the Philosophy of Mathematics", and "Wittgenstein's
Remarks on the Foundations of AI" very heavily on this
list. Have you read them? I very much like Shanker's
approach, and agree with what he explains as Wittgenstein's
position. Which brings us to the second point, which is
that, fully explained, Wittgenstein was mostly WRONG
on these matters, in which he failed to engage with
Turing's work, which takes a very different approach
to the matter. If there is a computer running on the
planet today, Wittgenstein is wrong. Now, just how
and why Wittgenstein did not engage more with Turing
is an interesting question, and for that matter Turing
did not address many of the practical aspects of the
computing that we know today. Both of these I view as
huge, lost opportunities, ... which we may simply have
to remedy as best we can today.
> > Exercising my particular-based nominalism, let's say that
> > if someone writes some rules down, and we can see that
> > some agency operates by checking the rules and acting
> > on them, then we have rule-following. The computer is
> > the archetype of rule-following.
>
> Why? "Who says? Where does this suddenly come from?
Who has to say?
> It most certainly does not *follow* from your previous
> sentences. It is just a bald assertion... which I and
> many others, including evidently Neil, completely reject.
> Is it simply axiomatic? ... for YOU?
Sure. Me. And Hilbert. And Turing. And, I assert
tendentiously and repeatedly, everyone who uses a computer
today.
> What a computer does is no more THE archetype or paradigm
> of rule-following than What PaintBox Pro does is THE
> (or even an) arhetype of painting pictures, or than a
> Formula 1 car is THE (or even a) paradigm of athletic
> prowess.
Well, here's the thing, you're talking from a viewpoint
that eschews archetypes or paradigms. I'm not. Now,
the paradigms and archetypes I'm talking about, are probably
not the paradigms and archetypes that you are eschewing.
So any reflexive rejections like this, are not going
to clarify anything.
> And you preceed this by what you describe as an exercise
> in "particular-based nominalism" ... just introduce some
> totally abstract notion of someone "writing down rules"
> and "some ageny" "checking" them.
Not sure what you mean here, but again, I lay down no
abstractions, I discuss particulars.
> You seem to find it deeply problematic (above) as to
> what rules even are let alone as to what they "exactly"
> are (above) and so what are you saying, meaning, thinking,
> "seeing" or imagining when you say: "let's say that if
> someone writes some rules down, and we can see that some
> agency operates by checking the rules and acting on them,
> then we have rule-following" ! Good grief! This is utter
> and complete confusion! How do we "see" that some "agency"
> "operates" by checking the rules?
I'm going to assume you know something of the low-level
operations of real computers, and/or the definitions of
Turing Machines. That is what I mean. If you can pass
a good undergrad class on these matters, I think you see
them and are not confused.
If you are holding onto anything from Wittgenstein about
how these matters are unclear - you might want to consider
loosening your grasp. Does Wittgenstein really forbid
students from taking and passing tests?
I suppose I could rephrase this in terms of grammars
and normative agreements, but I hope we can assume such
rephrasings as necessary. Though, the reconciliation of
the Wittgenstein view and the Turing view - and my attempts
at extending the Turing view - will have some terminological
problems, from time to time.
> What if the sentence ("rule") that your imagined person
> wrote down was "No smoking in my car"? or "the bishop
> moves diagonally"? Is that even a rule? or is it an order?
> A command? Or even just a statement of fact? What makes the
> sentences an expression of a rule? That some agent "checks"
> them and then does something, "acts on them"? What must the
> agent check? The spelling? The syntax? Is checking spelling
> and the syntax the same as checking the *rule*? What makes
> whatever the "agent" then does "rule-following" how do you
> "see" that it is rule following? with your eyes? Is what
> the agent then does to have *correctly* followed the rule?
> How do you see this correctness?
All good questions.
That are answered every day by anyone using a computer,
and moreso by those developing computer programs.
> Lord alone knows who this abstract "agency" is - Big
> Brother?
No abstractions, or at least they follow from particulars.
> or how and why he/she/it would "check" the rules
> written down! or "act upon them"? It is up to the author
> of the rules to check them? Check them against what?
> For what? Check them in accordance with what other rules?
> What makes the sentence(s) that were written down
> expressions of rules at all rather orders or
> instructions (which are not necessarily rules at all)
> Where did these come from? Where did he get the rules
> that he wrote down from? Just plucked them out of thin
> air? Presumably they are his rules? Who were the rules
> for? Himself, his children or membbers of a club? What
> were the rules for? behaviour at the dinner table?
> Playing a new game he was inventing? What? Why must
> some abstract "agency" act upon the rules? ..... No
> particulars at all, no specifics at all, all just
> complete empty abstraction! - the diametric oposite
> of what I might have called "particular-based nominalism"!
> Disaster is bound to ensue.
Try the decaf.
These are indeed all the questions, and they all
have answers, most of which are blatently obvious.
These are exactly what computers are all about!
> > Now, what about people? Do people ever follow rules?
>
> So you are in deep doubt that people ever follow rules
It is more of a rhetorical question, Rob.
> and, what is more, you are (above) deeply unresolved
> in your own mind about what "exactly" rules even are!
Again.
> AND YET you utterly convinced that computers not only
> do *follow* rules, but are THE paradigmatic *archetype*
> of rule-following!
Very good.
>In which case it is clear that you
> are simply *defining* rule-following as whatever it is
> that computers ("paradigmatically") "do"! - and "rules"
> as whatever that type of thing is that they allegedly
> (you say) "follow" or react to! Rather question begging,
> to say the very least.
I am glad to see you exercised, it means you are engaged,
but you have yet to see that there is a nominalist
procedure here that does not begin with your traditional
platonist assumptions and habits, and asserting the later
at high volume does not address the issues.
> Is it not bizarrely strange then that the very
> word "rule", notions of rules and rule-following pre-
> dated computers by some millennia!
If I claim that my local eatery produces the
archetype of the hamburger, you might find this
mildly humorous, but would you find it hard to understand,
and could you a priori find it false?
To some degree, yes, I really will defend the proposition
that we all waited around for several thousand years of
western intellectual history, until the modern computer
was conceived, developed, and built. That is, the
modern computer has a lot of insight to offer on many
ancient problems never really satisfactorily resolved.
Man has used fire for who knows how many thousands
of years, but it wasn't until the twentieth century
that we understood it terribly well, and there are
still many dynamics of it that we simply shrug and
use, even today.
So, actually, my *really* tendentious claim is not
even that computers exemplify what it means to
follow rules, MY claim is that we STILL don't really
have a satisfactory explanation of how this all
works - even though we BUILD the things to spec!
> Did humanity have
> to wait for computers appear in the world to
> find out the meaning of what it had been doing and
> talking about for centuries?
Yes, exactly so.
>I have a feeling that
> something is seriously awry here.
As well you should!
But it is a load of your philosophical assumptions
quivering in the sudden dawn, and it will all work
out well in the end.
>All you have done
> is *decided* in your own mind to ignore everything
> else and simply *define* "rule-following" as whatever
> it is that computers now do and now you are wracking
> your brains trying to work out whether people ever do
> what computers do or whether we are justified in
> describing anything they do as rule-following
> according to your new and highly presumptive
> definition. The philosopher having tied himself
> in all sorts of knots about the concepts of rules and
> rule-following looks at the latest toy to grab his
> fancy and tries to understand everything in terms
> of that!
Yes yes, Freud tried to understand the mind in
terms of steam pressures. But I repeat again,
this particular story is several thousand years
long, and if I'm proposing it's not quite at its
destination yet, I have neither invented a new
problem nor asserted anything but the obvious,
as what is not "obvious" after it is pointed out.
> > They don't seem to do so as a causal necessity,
> though one could argue the case.
>
> How? Examples?
>
> Would the fact that one can, for example, do
> arithmetic, check one's change at the shop, by
> pure rote habit be a case for arguing that we
> sometimes simply follow rules "mechanically"
> ... if not "causally"? Do rules that we have,
> as it were, internalised, to the point of making
> them habits or "second-nature" (telling expression
> that) thereby become "causal mechanisms"?
First, I am mostly concerned with understanding in
additional detail, and with the proper framework, how
it is computers do what they do all day long. Only
after making some progress here, would I really want to
address how we should look at what people do.
The relevant parts of Wittgenstein here are where
he discusses proofs and surveyability. If you have
a putative "rule", when do you accept it as a rule,
must you prove it, and prove it anew each time?
Wittgenstein demands the proof in detail so that
we have surveyed it, and then suggests we can use
it afterwards because we have surveyed and proved it -
but he does NOT answer HOW WE CAN KNOW that we have
surveyed and proved it. I refer you to Lewis Carroll's
"Achilles and the Tortoise" for details.
Which, again, is why a different framework from
Wittgenstein's is needed. For there are aspects of
the process that are going to look, to our unaccustomed
eyes, like stipulation and mechanicism. Of course this
has been in the works since Euclid, and Wittgenstein
made his contributions in another direction, but that
is just what makes this Turing/Wittgenstein relationship
so interesting.
> > If they follow a rule by choice, well, is that even
> > rule-following as such?"
>
> Evidently, yes. The premise of you question is that
> they follow a rule. That, in addition, they do so "by
> choice" makes no diffrence: they have followed a rule.
> That is an *instance* of rule-following "as such", no
> matter whatever that additional epithet "as such" is
> supposed to signify. Not very much I would have thought,
> unless, though an avowed nominalist, you are in fact a
> closet Platonist. I would have thought that the last
> thing a self-respecting nominalist would countenance
> is that there is any significant difference between
> "rule-following" and "rule-following as such"!
> (Platonist habits of thought die very hard, the
> tendency is endemic and takes all sorts of
> disguised forms - despite one's best intentions!)
Which is exactly the point I hoped to make.
> ON THE OTHER HAND if they had had NO choice in the
> matter would whatever they then did (or whatever
> happened) have even been "rule-FOLLOWING" AT ALL,
> in the first place?! Could it then be intelligibly
> so described? If I pull your hat off when we go
> into a church have *YOU* thereby *FOLLOWED* the
> rule: "No hats to be worn in church"? Sure, I know
> you are now *compliant* with the rule, that is
> why I snatched it off your head, but did you, or
> are you now, *following* the rule. Maybe you don't
> even know *why* I snatched you hat of your head,
> think it was just a joke and are happy to go along
> with it. Are you still *following* the rule in that
> case?
Sounds like a good topic for the philosophical
debate team this week.
> All this makes me wonder whether you are capable of
> distinguishing between a cyclist turning right at
> a T-junction because the rules of the road there
> require him to do so and a tram turning right at
> the same place because the rails it runs on constrain
> it to do so! It would seem that by your lights the
> latter would be a paradigmatic archetype of rule-
> following but that you would have all sorts of deep
> philosophical doubts about the former being rule-
> following at all! Bizarre. Totally bizarre.
My immediate concern is with understanding
computers, not transportation rules. I have said
many times, but perhaps not in this thread, that
I am not expressing any opinion on whether the planets
"follow rules" in traveling their orbits. Per
nominalism, my particular at this point is the
computer on my desktop, not Jupiter, nor a bicyclist
nor the tram. All in good time, I suppose, but none
of these other things looks at first glance much
like my computer, so we may be some time in
getting to them.
(In some circles these are very well-worn discussions,
whether the planets "follow rules" and the like, and
to re-survey the spectrum of them would be a mistake,
for better or worse I am exploring in depth one
particular opinion on the matter)
> > Some of this may have been behind LW's skepticism
> > about rules."
>
> What? What skepticism did Wittgenstein have about
> rules? I have never come across any hint that *LW*
> was skeptical about rules in anything he wrote. If
> anything quite the contrary. Certainly not the major
> published works ... I don't have much access to the
> "nachlass" ....
The sections you mentioned, PI $180 to $242. If that
is not skepticism about rules, just what do you think
it is?
> > Other parts remain problematic."
>
> Indeed. But Wittgenstein resolved/disolved the
> problems, or at least pointed the way to so doing.
Whether or not he did, is at issue here, because
it seems to leave no room for a satisfactory explanation
of why we use computers as we do today. My goal is
to preserve a large part of what Wittgenstein says
on these matters, rather than reject the whole.
> > Going back to computers, I went to recognize two
> > levels, the hardware level that is causal, and the
> > program level that is the paradigm case for rule-following.
>
> > So, when you (Neil) say that the computer "just"
> > is a machine that doesn't know symbols, at the low
> > level, I agree with you. However, at the high level,
> > I see symbolic programs being executed, and must see this
> > as rule-based systems being done by machines.
>
> > How do you like that?
>
> Nope. Not much. Why *must* you "see" it thus?
Because my friend Fred has trouble getting the new
payroll program written and operating correctly. Because
Fodor has trouble believing in his own computational theories
of mind, and in particular makes statements about computation
that see to conflict with Fred's practices. I'm trying
to address these and other matters, and have settled on
a particular (!) theory of how this might best be done.
It's not a must, it's a choice and an effort.
>Under what
> what compulsion? Wittgenstein remarked to the effect
> that all too often when a philosopher uses the word
> "must", thinking that he is under some unshakeable
> logical constraint, he is in fact under the sway
> of a *picture*, a deep psychological presupposition,
> a presupposed mental image or picture, or subliminal
> image perhaps embedded in our language itself.
> "A picture held us captive. And we could not get
> outside it, for it lay in our language and language
> seemed to repeat it to us inexorably."
> It would seem that one picture that holds you
> captive is precisely that of the computer as the
> paradigm and archetype of rule-*following*.
This is all quite interesting, in a meta-way.
Until these recent discussions on this group, I did
not realize there were people walking around who took
Wittgenstein's philosophy quite so seriously, nor really
that Wittgenstein had been quite so extreme in his claims.
There are a (very) few places in the Shanker books where he
engages in such exercises, but I had simply squinted at
those and read on. But on this forum I have seen
people who seem to believe that Wittgenstein forbids
the use of Webster's dictionary, and others who seem
to feel that any expressed generalization is necessarily
wrong, an invalid exercise no matter what the facts.
Now, I do see the basis for this in Wittgenstein's
texts, and I suppose I could engage it in detail,
but I don't have anything *new* to add to the debate,
which is some flavor of how far one wants to drive
eliminativist theories, and whether an extreme empiricism
is a workable theory in the modern world. Chomksy's
take on behaviorism answers the empiricist theory, and
the Churchland's eliminativism in favor of neural networks
has been the locus for much work on the eliminativist
views relating to philosophy of mind and/or computation.
So, yes, I understand your warnings about being held
captive by assumptions, but freeing myself from the
illusion of the world, is not my current activity, but
something else - having freed oneself from illusions,
can one build a better mousetrap?
I hope this helps you understand the kind of project
that I (and Fodor, and the like) are on, and just
where Wittgenstein can help this project, and where
he, at best, has nothing to offer.
I'll review the rest of your post, but I think
the lines should already be clear.
> Computer programs do not *follow*
> rules, only people writing them do.
Again, that is the question, I cannot accept
your assertion here.
> And when the
> hardware "runs" the program what rules is it
> "followng"? None.
I like that, but others will not.
Even a UTM has some foundational hardware stipulated,
but does not even that run by rules - because one UTM
can emulate another? It is tendentious to assert that
there are NOT rules at the lowest level.
> It is hardwired binary logic
> and machine code ticking over. But it seems you
> want to say that the combined h/w - s/w is
> *following* the rules "it" was "given". No so?
Close enough.
> ANY program, or even set of programs, can in
> principle, be realised in hardware - i.e.
> essentially arrays of two-state devices,
> switches, logic-gates, relays &c. In the early
> days of computers, programing them *consisted* in
> manually setting hundreds upon hundreds of on-off
> switchces - and that is in fact what programming
> a computer still is - until we have "computers"
> that are NOT Turing machines. And so I suspect
> your argument here does not work. You will have to
> find some other way of making your case, assuming
> that it can intelligibly be made.
Right, there is in general a tradeoff between hardware
and software. However, that is not explanatory of what
the difference is, and why we find it convenient to do
things quite the way we do, with limited hardware and
lots of software, which is after all just the image
of the TM in the first place - a very limited state
machine, with a very long tape. It may just be that
we have not yet fully understood even this simple
model, and that is close to my point.
> Appart from that, sure - pocket calculators also
> provide a rule-based system "being done" by machines.
> So too does an automated railway points switching
> system. That does not mean that these are paradigms
> of rule-*following* rather than ingenious machines
> that simulate the results of (correct) rule-
> following.
I suggest that, in a completed theory of all this
stuff, I would find the above statements very confused.
> So, come clean, forget about trying to draw some
> line between hardware and software because I suspect
> your real problem now is, and always has been:
> then why are not people just ingenious machines that
> simulate the results of (correct) rule following!
> ... If that now is a meaningful question. Where
> do we get the notion of correct rule following from?
Fodor has spent a lot of time worrying about the
tension between reductionism and our holding onto
the "special sciences" like biology and psychology,
as opposed to the Vienna/positivist notion of the unity
of science. I think we have something similar going
on here, between the latter Wittgenstein's eliminativist
tendencies, and the notion that particulars (even rules!)
have meaning in use that is at odds with this
eliminativist bent.
I see the issues as Wittgenstein versus Wittgenstein,
and (as I've said in earlier messages) that I believe
Turing was quite strongly inspired by Wittgenstein,
the early Turing by the early Wittgenstein, and the
latter Turing by the latter Wittgenstein, so even where
I take Turing's part in things, much of the credit goes
back to Wittgenstein in one part or another.
Are people just ingenious machines? Of course.
"simulate the results of (correct) rule following!"?
Sorry, that doesn't really parse.
On the one hand, the question at hand is whether people
can do what they do by rule-following, not whether they
can simulate something else, that's just another question,
and the answer to that second question is certainly yes,
that is if they can follow rules, they could probably
emulate someone else's rules, but we've known that
since 1936.
Josh
WEB VIEW: http://tinyurl.com/ku7ga4
TODAY: http://alturl.com/whcf
3 DAYS: http://alturl.com/d9vz
1 WEEK: http://alturl.com/yeza
GOOGLE: http://groups.google.com/group/Wittrs
YAHOO: http://groups.yahoo.com/group/Wittrs/
FREELIST: //www.freelists.org/archive/wittrs/09-2009
Other related posts:
