On Tue, Mar 19, 2013 at 9:39 PM, Matt Mahoney <[email protected]>wrote:
> Heidenbugs >> > Well gentlemen, I am very nearly shocked by this patent and the discussion it is generating. I can see how it fits in Steve's universe, after all it is not his first patent, but I can hardly see anything else. Presumably the patent "team" has faith in the utility and patentability of the idea, while I lack both. In terms of patentability, I can see that one can and should insert a lot of "patentese" into an algorithm to make it patentable, for example other parsing patents list "method and apparatus", because as "we" know, no apparatus means no patent. Now, I am not a patent lawyer but it seems to me that you expect me to pay if I generate text, with a marketing message or otherwise, by keyword-matching and a certain hashing function and rule workflow, but a hell of a lot depends on me creating brilliant rules. What if I tried to patent "method and apparatus for harvesting free energy" and basically provided a battery, two cables and a blank in the middle where any and all free energy devices would plug in, if and when they were developed and sold. Would that be a brilliant way to put a patent tax on free energy so-called research and development? If someone tried to circumvent my patent by including 4 cables, two batteries and hydraulic energy storage, would I be able to claim they essentially violate my patent? The abstract of the claims differs from your posting here, partly for obvious and understandable reasons. Here you show a bigger picture, which I just don't see. The speed advantage of whatever you are proposing seems debatable, especially with rules being the slower part. Avoiding Heidenbugs would be worth a few major patents on its own but I don't think you have proof or will ever have proof you avoid Heidenbugs with this workflow, I can see keyword-rules as heidenbuggy as hell emerging. Ordering the lexicon according to word frequency has been done for ever and seems to weaken the patent - if I use a similar workflow and change some minute things in your patent I expect to go scot-free. Not to mention that "the" is *the* word is certain situations and word-frequency orderings probably are too weak for linguistic nuances. You do mention the word disambiguation, now disambiguation is the alpha and omega of NLP, if we could map words to an ontology unambiguously we would be on cloud seven, sadly I don't see one bit of disambiguation "infrastructure" in your proposal. In short, I don't see more potential in this patent than patent-trolling advertisers! PS What would you market to someone who used heidenbuggy twice! AT ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
