From Aaron Mon Nov 20 23:49:15 GMT 1995 Newsgroups: talk.philosophy.misc,talk.philosophy.humanism,sci.philosophy.tech,sci.philosophy.meta,comp.ai.philosophy,alt.philosophy.zen,alt.philosophy.taoism,alt.philosophy.objectivism,alt.philosophy.no-dogs,alt.philosophy.jarf,alt.philosophy.debate,alt.christnet.philosophy References: <199511082348.SAA79587@atlanta.american.edu> <480d0b$8u2@paladin.american.edu> <1995Nov10.223444.16241@media.mit.edu> Subject: Originality (Was Re: This whole Descartes thing) A few minutes random browsing when I should have been doing some other work revealed this gem, which has provoked me to comment. mfc3@cornell.edu (Michael Canino) writes: > Date: Thu, 16 Nov 1995 14:40:43 -0400 > Organization: Cornell University > ..... > > In <1995Nov10.223444.16241@media.mit.edu> minsky@media.mit.edu (Marvin > Minsky) writes: > > > > |Myself, I find it more interesting to figure things out than to look > > |for answers in ancient books written by people who couldn't figure > > |things out. > > .... > > Of course when you consider how few original thoughts there have been in > the millenia of human history, the odds really stack up against anyone > saying anything that hasn't been said or thought by a person somewhere, > sometime. Depends on the kind of originality you look for. There are different ways of being original. Here's a subset: 1. Create a new well-formed sentence, never previously uttered. We probably all do this hundreds, if not, thousands, of times during a normal life-time. Your sentence and Minsky's could both be examples. (I once heard a child say: "Today might be much more hotter than it usually bees" I doubt that it has ever been said (spontaneously) before or since, apart from my reports of the utterance. 2. Discover a surprising and useful new consequence of what was previously known. I suppose good mathematicians achieve this quite often. It's also the task of a good programmer. 3.a. Ask a new question whose answer turns out to have all sorts of important consequences. (E.g. why do apples disconnected from trees tend to move, rather than remain where they are?) 3.b. Formulate a deep conjecture about the answer to a question of type 3. A deep conjecture is one that tends to generate major new investigations in pursuit of ways of testing the conjecture. (3.a. and 3.b. are two sides of the same coin.) (This concept of "depth" requires further clarification, which I'll not offer now.) 4. Create an extension to an existing language, which enables new things to be said, new questions, to be asked, and new derivations, to be formulated, which were previously inexpressible. A simple version of this merely extends the vocabulary by adding some word or phrase not definable in terms of pre-existing concepts. A deep extension typically adds a major family of new concepts extending pre-existing ontologies in powerful ways, e.g. by providing new forms of explanation for previously known phenomena. An example is the extension of common sense (pre-scientfic) language with concepts referring to the hidden (unobservable) constituents of matter. Another example is the extension of common sense knowledge of reproduction and family resemblances, with concepts within a theory about collections of genes within individuals and populations and mechanisms for extending and filtering a gene-pool. (All this is inconsistent with Fodor's idea of a language of thought. For that rules out certain kinds of novelty.) 5. Create an extension not merely to the vocabulary and concepts and theories, but also to the forms of expression and manipulation of those forms, e.g. in reasoning. This typically involves creating new forms of syntax. E.g. the invention of differential and integral calculus made possible new forms of expression and derivation. Tensor calculus was another example. Contour lines on maps were another example. New forms of logic (e.g. parametrised modal logics?) are another. New programming languages can also do this: e.g. higher order functional languages, object-oriented languages graphical programming languages. Languages for specifying asynchronous concurrent processes. (Apart from the last, all of these are subsumed by the expressive power of Turing machines, though some specialised notations extend heuristic power beyond Turing machines for specific classes of problems). 6. Creation of totally new forms of expression and derivation, which are not simply extensions of previous ones. It is arguable, for example, that the forms of parallel distributed representation used in neural nets are not just an extension of logic and familiar forms of computation, but a parallel development. Will quantum computation be another? 7. Creation of a new medium within which to explore varieties of forms of representation supporting all the other forms of originality. I suspect that this point is the most important feature of computing machinery: computers provide us with new ways of inventing, implementing, and exploring new virtual machines, providing media for new forms of representation. Previously we were restricted to 2-D surfaces, 3-D structures, patterns of change in 2-D or 3-D, and perhaps acoustic patterns also. Richer and more varied topologies can be embedded in virtual computing machines (e.g. actual manipulable lists which contain themselves as elements could not exist previously -- now they are easy to make). The types of novelty in points 1 through 7 are arranged in roughly increasing power of the new development. E.g. providing a new medium can sometimes provide support for a raft of new languages, extending what we can think, and and what we can do, more than any theory expressed in a pre-existing language can. (I first tried to spell all this out in chapter 2 of The Computer Revolution in Philosophy. Others have made similar points. My main claim was that the most important discoveries in science are not discoveries of new laws or theories, but the discovery of new ranges of possibilities, about which various laws or theories could be formulated. This deepens our knowledge of the "form" of the world, as opposed to its "contents" or its "constraints" -- the laws.) But there's much more to be said, as usual. Aaron