PM On Fri, Oct 17, 2014 at 12:27 AM, Piaget Modeler via AGI <[email protected]> wrote:
> When you're looking at a new language you must first ask yourself, > what problem is this language going to tackle that is not handled by > any existing language. > > Math is pretty well covered by R , C, and most other languages. > You mean arithmetic is covered - as these won't do even the simplest algebra. Neither will they track significance or dimensionality. I might be the only one here who ever held a University job as a mathematician. In my case, it worked for the Physics and Astronomy departments of the University of Washington. Most of my work was chasing the bugs out of people's numerical analysis programs. Often this started with a bunch of equations that I turned into cumbersome English, which I then rewrote into an English procedure to evaluate, which I then changed into comments, and sprinkled the (then FORTRAN) code in to do what the English said. Something like COBOL would eliminate the last step in this, and a good algebraic tool would eliminate the step before that. > > So what problem with this CAGIL (Common AGI Language) address? > The most important thing is that the code would become READABLE. People could actually post snippets and everyone could see what they do, without a difficult reverse engineering effort. Another thing needed to get AGI ready for prime time is some sort of standard module interface, so that feature extraction, image recognition, and other manipulations could be easily daisy-chained or otherwise strung together without concern for calling sequences or internal restrictions. More importantly, what aspects do we want to make black boxes > so the language handles it for us. > I think you want to chase capabilities out of the language, but allow the creation of new verbs that act something like subroutine calls. > One example, in Premise there > is a *System.Chat* namespace to facilitate communication among > Premise servers, so one server can interrogate others using simple > primitives such as TELL, ASK, and ADVISE. > I would think you would want to make server interfaces as invisible as possible, so that the code needn't consider them. > > What black boxes do you want in your INTERNET AGIs? > Your question seems to presume a strictly procedural sort of implementation, rather than (for example) specifying how information flow from one matrix through others without concern for the detailed handling of specific elements, or even matrices. Wouldn't you rather diagram your process, and then run it and see what comes out the other end - or what is fed back to make a system autonomous? There are some important capabilities in SQL that might work well, most especially virtual matrices, where you define their elements by specifying how they are computed. When you refer to an element, it is computed on the fly. Since the references are often buried into optimized SQL statements, the actual order of computation is complex and often not deterministic. Arithmetic statement functions were an interesting part of FORTRAN that seems to have fallen into the dustbin, that made virtual arrays (and other advanced concepts) simple to implement with one-line functions. The argument has been made that subroutines are just as good as arithmetic statement functions, but subroutines short-circuit the optimizations that are available for in-line code, and even implementations of inline functions end up transcending readability. Until a decade ago, the primary limitation on complexity used to be RAM size, but now it is readability - and that was what COBOL excelled at, at least when it was well written. Now, if we can forge some sort of marriage between COBOL concepts and modern language capabilities, we will be able to push the frontier of complexity out another order of magnitude or so. I doubt whether an order of magnitude will be enough - but that is just my opinion that is NOT shared by many others here on this forum. Once we make it possible FOR HUMANS to write really BIG programs that work, who knows, maybe those programs might become bright enough to continue their own development. Steve > > ~PM > > ------------------------------ > Date: Fri, 17 Oct 2014 00:17:24 -0700 > Subject: Re: [agi] Event Models > From: [email protected] > To: [email protected] > > Mike, > > You are 100% correct. The BIG problem for COBOL doing things like serious > AI has been a lack of imagination in producing language extensions that are > grammatically correct English. > > Here, we have all of the imagination needed to do ANYTHING, so let's come > up with some proposed language extensions/changes so that people can write > their AI code in something that is readable as English. It is a little > cumbersome to write this way, but it DOES chase out most of the bugs before > ever running the code, and it makes it easy to manage HUGE programs. > > Perhaps something like the Matrix extensions that were in the original > Dartmouth BASIC but were later dropped by almost everyone. There, a > statement like: > > MAT A = B + C > > added the elements of the B matrix to the elements of the C matrix, and > stored the results in the A matrix. An extended COBOL to do the same might > be > > Add matrix B to matrix C giving matrix A. > > This worked for most statements, e.g. > > MAT INPUT A > > input the elements of matrix A > > Of course matrices were brought to great refinement in APL, whose > semantics would probably be much better than BASIC's simplistic semantics. > > Much more would be needed. > > > > On Thu, Oct 16, 2014 at 10:19 PM, Mike Archbold via AGI <[email protected]> > wrote: > > I spent ten years writing COBOL code. Everybody makes fun of COBOL > but if you are doing bread and butter stuff, it's fine. I mean, you > don't need no fancy graphics or fancy anything, no big data, no neural > nets, no Tononi, no chaotic dynamics,.... for most business > transactions. COBOL is simple, like rock and roll. Three chord > stuff. > > On 10/15/14, Ben Goertzel via AGI <[email protected]> wrote: > > I can't remember where I was the 97% figure > > > > However, Linux dominates mobile, and has a clear majority of webservers > as > > well > > > > > http://en.wikipedia.org/wiki/Usage_share_of_operating_systems#Servers_on_the_Internet > > > > -- Ben > > > > On Thu, Oct 16, 2014 at 11:43 AM, Matt Mahoney via AGI <[email protected]> > > wrote: > > > >> On Wed, Oct 15, 2014 at 9:26 PM, Ben Goertzel <[email protected]> wrote: > >> > Would you be surprised if a ragtag group of OSS developers made a > >> > better > >> OS > >> > than Microsoft, IBM or Apple -- one used to power 97% or so of the > >> Internet > >> > ? > >> > >> Actually the market share for Linux is 1.64%. > >> > >> > http://www.netmarketshare.com/operating-system-market-share.aspx?qprid=10&qpcustomd=0 > >> > >> And the total development effort is quite substantial, similar to > >> Windows. The difference is in how the work is funded. In the open > >> source model, information has negative value. You sell yourself by > >> giving away your work. > >> > >> I think a distributed AGI can be funded this way too. But only if > >> there is some compelling reason to use it before it is fully > >> developed. > >> > >> > I don't think an AGI has to pretend to be human ... > >> > >> Yes, we agree on that. Unless we are talking about applications like > >> uploading or identity theft. But once we solve the problem of modeling > >> human minds, it is a simple matter to program them to carry out their > >> predictions of your actions in real time. > >> > >> -- > >> -- Matt Mahoney, [email protected] > >> > >> > >> ------------------------------------------- > >> AGI > >> Archives: https://www.listbox.com/member/archive/303/=now > >> RSS Feed: > https://www.listbox.com/member/archive/rss/303/212726-deec6279 > >> Modify Your Subscription: > >> https://www.listbox.com/member/?&; > >> Powered by Listbox: http://www.listbox.com > >> > > > > > > > > -- > > Ben Goertzel, PhD > > http://goertzel.org > > > > "The reasonable man adapts himself to the world: the unreasonable one > > persists in trying to adapt the world to himself. Therefore all progress > > depends on the unreasonable man." -- George Bernard Shaw > > > > > > > > ------------------------------------------- > > AGI > > Archives: https://www.listbox.com/member/archive/303/=now > > RSS Feed: > https://www.listbox.com/member/archive/rss/303/11943661-d9279dae > > Modify Your Subscription: > > https://www.listbox.com/member/?&; > > Powered by Listbox: http://www.listbox.com > > > > > ------------------------------------------- > AGI > Archives: https://www.listbox.com/member/archive/303/=now > RSS Feed: https://www.listbox.com/member/archive/rss/303/10443978-6f4c28ac > Modify Your Subscription: https://www.listbox.com/member/?&; > Powered by Listbox: http://www.listbox.com > > > > > -- > Full employment can be had with the stoke of a pen. Simply institute a six > hour workday. That will easily create enough new jobs to bring back full > employment. > > *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> > <https://www.listbox.com/member/archive/rss/303/19999924-4a978ccc> | > Modify <https://www.listbox.com/member/?&;> Your Subscription > <http://www.listbox.com> > *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> > <https://www.listbox.com/member/archive/rss/303/10443978-6f4c28ac> | > Modify > <https://www.listbox.com/member/?&;> > Your Subscription <http://www.listbox.com> > -- Full employment can be had with the stoke of a pen. Simply institute a six hour workday. That will easily create enough new jobs to bring back full employment. ------------------------------------------- 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
