A specific pattern – Mandelbrot, Fourier transform, cellular automaton no. 30 etc.. specific.
Examples are specific, Aaron, not general. And the “pattern of neuronal firings in your brain” is a fictional product of your imagination. Specifics. Evidence. e.g. https://www.google.com/search?num=10&hl=en&safe=off&site=imghp&tbm=isch&source=hp&biw=1340&bih=690&q=patterns&oq=patterns&gs_l=img.3..0l10.1679.2746.0.3037.8.7.0.1.1.0.159.516.5j2.7.0...0.0...1ac.1.hHfKyrSHq4M And here are “evolving patterns”, except that they don’t truly evolve with new elements – they aren’t truly “emergent” https://www.google.com/search?num=10&hl=en&safe=off&site=imghp&tbm=isch&source=hp&biw=1340&bih=690&q=evolving+patterns&oq=evolving+patterns&gs_l=img.3..0i24l2.11507.42532.0.42818.38.21.10.7.8.0.65.874.21.21.0...0.0...1ac.1.zE3R1TChxHw unlike natural evolution, in which new elements, like eyes, limbs, opposable thumbs, voiceboxes, etc emerge every time. From: Aaron Hosford Sent: Tuesday, November 20, 2012 12:57 AM To: AGI Subject: Re: [agi] Randomness: Mathematics as Perceptual Bias If it’s so bizarre, you’ll have no problem giving us a single example of a pattern that does change and evolve – a single logicomathematical/ computational pattern. The pattern of bits in a computer's memory, the pattern of neuronal firings or synaptic connections in my brain, the structure of a semantic network as it is modified to implement learning from experience, the pattern of state/action weights in a reinforcement learning algorithm as it seeks to maximize reward, the weights in a neural network as it learns a function, etc., etc., etc... And since patterns are formulaic/algorithmic, you might add a single formula/algorithm that evolves. Stop conflating formulas with algorithms and mathematics/logic with computation. They are not the same. Formulas and other mathematical/logical expressions may be unchanging and inflexible, frozen on a piece of paper or in a document, but that doesn't mean that algorithms and computation are as well. Programs do things. They have state. Formulas don't. There are plenty of folks who have already discussed on this list their ideas for self-modifying programs. Any such program would necessarily be a pattern which changes over time. And while you might claim or imagine that no "new elements" could be introduced via this process, I can point to the effects of interaction with the environment (or in absence of I/O, a random number generator) that could potentially allow the program to evolve to produce any possible behavior that a computer is capable of simulating or modeling. There's nothing special or magical about your "new elements". If there's a shortcoming in algorithms, it's in computers themselves, because software packages already exist (virtual machines) which simulate the full behavior of an entire computer, so algorithms necessarily are just as powerful as computers themselves. If you want to go on a crusade against algorithms, you're going to have to include computers themselves as targets, because they mirror each other's capabilities. Have fun with that one, because I'm not going to be involved. I'm quite sure what I just said will, like everything I've said to you before, be ignored or twisted or redefined into something that suits the indefensible perspective you've chosen to wed yourself to. Say whatever you want. I am done with this conversation. I am not going to pursue it further, because it's a waste of my time and it's frustrating debating with someone who interminably refuses to concede a point. If you want to argue about it, find someone else to annoy. On Mon, Nov 19, 2012 at 5:34 PM, Mike Tintner <[email protected]> wrote: Aarron/MT: Patterns cannot change or evolve in any way. That's a rather bizarre claim to make If it’s so bizarre, you’ll have no problem giving us a single example of a pattern that does change and evolve – a single logicomathematical/ computational pattern. And since patterns are formulaic/algorithmic, you might add a single formula/algorithm that evolves. P.S. GA’s do not add new elements or truly evolve – as we’ve discussed repeatedly – they only mix and remix the same old elements. (And are there any un-human-assisted GA’s, without humans intervening to select from different generations – in which case they don’t even qualify as patterns.) . From: Aaron Hosford Sent: Monday, November 19, 2012 11:11 PM To: AGI Subject: Re: [agi] Randomness: Mathematics as Perceptual Bias Patterns cannot change or evolve in any way. That's a rather bizarre claim to make. Any notion you may have of a pattern necessarily being unchanging is something restricted to your own personal definition of the term, and doesn't intrude into mine. The ability of a pattern to change over time depends on the system in which it is embedded. If the system makes room for patterns to change, then they can change. You apparently have in mind some sort of fragile, restricted system where nothing ever changes. An Evolutionary or Genetic Algorithm is a good example of a case where patterns are free to change. (Actually, Genetic Programming in particular is a much better example, since some GA encodings can be rather stiff and restricted.) There are plenty of other algorithms out there designed for dealing with shifting, changing patterns. New elements, fundamental change, evolution, creativity – AGI – and the real world – have bugger all (or v. little) to do with patterns. And this, to repeat, is demonstrable and incontrovertible. Evolution is itself a pattern (of behavior). It can be simulated on a machine (by copying that pattern) and then used to generate creativity just as it does in the real world. What you claim is demonstrable and incontrovertible has not been demonstrated and is therefore being controverted. If you have a demonstration, I'm still waiting for it. The things you keep saying patterns can't do (or have nothing to do with) are things that don't apply to my definition of pattern. What is your definition of pattern that you make such outlandish claims about them? pat·tern [pat-ern; Brit. pat-n] Show IPA noun 1. a decorative design, as for wallpaper, china, or textile fabrics,etc. 2. decoration or ornament having such a design. 3. a natural or chance marking, configuration, or design: patterns of frost on the window. 4. a distinctive style, model, or form: a new pattern of army helmet. 5. a combination of qualities, acts, tendencies, etc., forming a consistent or characteristic arrangement: the behavior patterns of teenagers. The definition I'm using is a blend of 3, 4, and 5 above. The key words to me are "chance...configuration", "style", and "arrangement". Note there is no mention of being unable to change over time under any of them. The closest any of them comes is #5, which describes it as "consistent or characteristic", but this is not the same as saying "permanent or unchanging". On Mon, Nov 19, 2012 at 12:12 PM, Mike Tintner <[email protected]> wrote: I never said patterns are repeated. Patterns can be v. complicated but they are ultimately sickmaking because all variations on a given pattern have the same elements in the same structural relationships – and that includes cellular automata where there is random variation in the elements. Patterns thus get boring (except to mathematicians). Every patchwork in a given collection is new and different while still similar, because it has, by definition, new elements – and satisfies our need for newness and not-to-be-bored by the same old patterns. Ultimately any patchwork can be evolved by steps into any form, picture or scene *WHATSOEVER*. A patchwork dress of abstract shapes can be evolved into a sea of human faces or a nuclear explosion or a battlescene – or anything. Patterns cannot change or evolve in any way. Patchworks mirror the real world. Your local street can and will evolve into a very different form over sufficient time. All forms and scenes in the real world evolve over time. And no one street is exactly like any other right now – at a given point in time. Every street can be regarded as an “evolution” of every other street. New elements, fundamental change, evolution, creativity – AGI – and the real world – have bugger all (or v. little) to do with patterns. And this, to repeat, is demonstrable and incontrovertible. From: Aaron Hosford Sent: Tuesday, November 13, 2012 4:03 PM To: AGI Subject: Re: [agi] Randomness: Mathematics as Perceptual Bias I didn't say anything at all about repeated patterns. There are other types of patterns besides repetition. Clearly a sentence like, "My dog ate my homework," or the equivalent predicate in logic, doesn't indicate a repeating pattern. (Unless, of course, the excuse gets used repeatedly.) And yet this is a pattern. I would go so far as to say, it's a pattern made up of a "patchwork" of relationships between several objects and events. On Tue, Nov 13, 2012 at 4:11 AM, Mike Tintner <[email protected]> wrote: Patterns exist - but they are islands in the patchwork seas of the real world. The brain is primarily designed to make sense of patchwork scenes and patchwork objects - and if you look too long at a heavily patterned scene, like a specially designed patterned room, you get sick – it ain’t natural. The patchwork nature of real world scenes is obvious and incontrovertible. From: Aaron Hosford Sent: Monday, November 12, 2012 8:43 PM To: AGI Subject: Re: [agi] Randomness: Mathematics as Perceptual Bias Summaries of perceptual information. These are the elusive "patterns" you say don't exist. On Sun, Nov 11, 2012 at 12:53 PM, Mike Tintner <[email protected]> wrote: Aaron: Todor's point was simply that logic (and language in general) merely express summaries .... summaries of what? Would you care to expand? AGI | Archives | Modify Your Subscription AGI | Archives | Modify Your Subscription AGI | Archives | Modify Your Subscription AGI | Archives | Modify Your Subscription AGI | Archives | Modify Your Subscription AGI | Archives | Modify Your Subscription AGI | Archives | Modify Your Subscription AGI | Archives | Modify Your Subscription ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-c97d2393 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-2484a968 Powered by Listbox: http://www.listbox.com
