The issue of whether computers "grasp" the game of chess is rather complicated. They do "grasp" the rules, if that is what is meant, but there is a lot more to the game. Until fairly recently, chess computers worked mostly by "brute force" tactical calculation of moves and had next to no grasp of strategy. However, in the recent years, human grandmasters became part of the programming teams, and this resulted in modern programs developing / being tuned into a better understanding of strategy. O.K. ________________________________ From: Richard Henninge <RichardHenninge@xxxxxxxxxxx> To: lit-ideas@xxxxxxxxxxxxx Sent: Saturday, April 14, 2012 6:13 AM Subject: [lit-ideas] Re: Philosophical Investigations online - amplification re PI Pace my correctors, Messieurs Ritchie and McEvoy, respectively, re weaponry, the petard has already exploded before the hoisting begins and, re grammar, when concluded, one has been hoisted, not hoist, on one's own petard. So many things go wrong in Donal's handling of my post that I am limiting myself to addressing his first six paragraphs. Donal hoists himself on his own petard because he challenges us to produce an example of a rule whose sense is said or stated in the rule and not just shown. The whole enterprise is skewed from the start as Robert Paul has eloquently and accurately noted: > The issue is stark and goes to fundamentals. My [i.e. Donal's] claim is not merely that > there are _some_ "rules" [as W means the term] whose sense is not 'said' > in a statement of that "rule", but that (for W) *there is no "rule" > whose sense is _said_ in a statement of that "rule"*. Nowhere, in Wittgenstein's examination of rule following does anything remotely like this appear. He nowhere speaks of the 'sense' of a rule, or, by implication how its 'sense' differs from what it 'shows.' His treatment of rules; rule following; being guided by, grasping, continuing, understanding a rule is, as I said earlier, interesting, complicated, essentially linked to some of what he says about agreement, games—and so on. And, to repeat: Wittgenstein himself has no clear 'answer' or 'solution' to the questions posed and the problems that arise between §201 and §243 (and appear in various formulations elsewhere). The 'say/show' distinction, is not part of his discussion of rules and the problems attending them. See too e.g. §§82-85. And so, though I know that Donal is barking up the wrong tree with his notion of the "sense" (which word he endows with a hypnotic, ephemeral, mystic significance, so to speak, all his own) of a rule, which he puts forward as if it were state's evidence that will prove the validity of his big-deal say/show-distinction theory of early and late Wittgenstein, I, fool, rush in with what strikes me as being the obvious refutation of his no-sense-of-the-rule-in-the-statement-of-the-rule theory: that a computer produces rule-based behavior (plays a game--chess, for example, continues a sequence, or, as Donal suggested, takes a number, adds 2 to it, and then adds 2 to the number so generated) solely on the basis of the statement of the rule--the part of its programming that results in its rule-based behavior. Donal can blather as much as he wants about the inadequacy of computer translations, but it was he who chose a very simple rule as the test of his theory, a rule concerning mathematics, and it is simply bluffing on his part to say that "What computers do by way of processing 'rules' does not involve their grasping the 'sense' of the 'rules' involved." What does Donal think this mysterious "sense of the rule" is? Probably Donal thinks that Wittgenstein has some special "sense" of "sense." He is also grasping at straws when he puts so much emphasis on the word "grasping," as if to exclude what a computer does when it "grasps" the game of chess. This is all just "definitional," but that is the point of the exercise when understanding Wittgenstein. If you can deny that a computer "grasps" the sense of chess even though it can play chess perfectly according to the rules, you can as easily deny that any individual grasps the sense of chess even though he or she can play chess perfectly according to the rules. And so what computers do cannot be taken to show anything about the “sense” of a “rule” in the sense in which W is interested in “rules” and their “sense”: for what computers do by way of processing does not involve computers grasping the “sense” of a “rule”. And since their processing does not involve grasping the sense of a rule, what they do cannot be taken to show that the “sense” of a “rule” may be said (on the basis that it is “said” in what they process). > >Richard’s attempted refutation is therefore entirely misconceived. > >What Richard should really try to do (if he really wants to refute W's >position) You see how far afield Donal is when he thinks I am trying to refute Wittgenstein's position and not his own, but Donal has apparently convinced himself that his position and Wittgenstein's are the same! As Robert Paul explained, Wittgenstein never used the word "sense" in the sense of the "sense of a rule." I would maintain that Wittgenstein "would" never have used it in that sense, but if forced to use it as applied to rules, I think, in keeping with Wittgenstein's thought, you would have to say that the sense of a rule is "grasped" or whatever if a person, or even a computer, acts according to it, and I am afraid Donal is not so clear on the distinction between showing and stating when it comes to rules. is state a “rule” so that its sense is _said_ in what is stated: but stating a computer programme, even one comprising “rules” [of computation], is not to state any “rules” whose “sense” is _said_ in what is stated. > I, or a computer, receive the instruction, which is also a statement of the rule--"take a number, add 2 to it, and then add 2 to the number so generated"--and the computer and I do just that: Have I not stated "a 'rule' so that its sense is _said_ in what is stated," i.e. --"take a number, add 2 to it, and then add 2 to the number so generated." Donal, when he is at his best, is a good lawyer. But this isn't law. And Donal remains hoisted. Richard Henninge University of Mainz ----- Original Message ----- >From: Donal McEvoy >To: lit-ideas@xxxxxxxxxxxxx >Sent: Friday, April 13, 2012 11:50 AM >Subject: [lit-ideas] Re: Philosophical Investigations online - amplification >re PI > > >There are longer ways but here is a shorter way with Richard Henninge’s post. > >Richard, as I recall, knows important things about the shortcomings of the >English translation of Wittgenstein’s work. Whatever these shortcomings, for >example in conveying the exact sense of W’s writing in German, they were >translated by humans not computers. And this for good reason.Their >shortcomings would arguably be much greater if the translation had been >performed by a computer following some programme: and that is because >computers have no grasp of the “sense” of language as humans do. What >computers do by way of processing “rules” does not involve their grasping the >“sense” of the “rules” involved. > >A simple example: a computer may contain an error in its programme so that it > produces the calculation ‘2 + 2 = 44’; but where a human would baulk because >they grasp that this must be mistaken, the computer, having no grasp of the >sense of what it processes, will not baulk at this mistake – it will not baulk > because it does not grasp it must be a mistake [it will only ‘baulk’ if >another part of its programme causes it to ‘baulk’ i.e. treat this as an >“error”]. A computer will no more baulk at ‘2 + 2 = 44’ than will a blackboard > it is chalked on: for, in Popper’s useful terms, both the computer and the >blackboard are confined to processing information at the level of World 1, and > lack conscious understanding of the content they process [still less, in >Popper's terms, do they grasp the World 3 content or mathematical principle >according to which '2 + 2 = 44' must be a mistake). > >And so what computers do cannot be taken to show anything about the “sense” >of a “rule” in the sense in which W is interested in “rules” and their >“sense”: for what computers do by way of processing does not involve >computers grasping the “sense” of a “rule”. And since their processing does >not involve grasping the sense of a rule, what they do cannot be taken to >show that the “sense” of a “rule” may be said (on the basis that it is “said” >in what they process). > >Richard’s attempted refutation is therefore entirely misconceived. > >What Richard should really try to do (if he really wants to refute W's >position) is state a “rule” so that its sense is _said_ in what is stated: >but stating a computer programme, even one comprising “rules” [of >computation], is not to state any “rules” whose “sense” is _said_ in what is >stated. >