I think you are fundamentally correct that we need better tools and in particular those tools need to do a lot of what Lisp does a little of. So while my feedback on the specifics will be mostly critical, I think you are trying to do the right thing in the first place, which is the biggest step, and thanks for sharing your design with us.
>From a brief skim of your design: The website uses buttons and some kind of frames or something to get at the documentation. This should be replaced with straight HTML pages. There's a lot of idiosyncrasy, by which I mean, cases where free design decisions have been made contrary to modern convention. For example, case insensitivity where the world has - whether rightly or wrongly - settled on case sensitivity. I'm not saying your choice is worse and I'm not saying your choice is better, I'm just observing that if you want other people to use your system, it needs to mostly follow familiar conventions and save the unfamiliarity for where it's really needed. If it's just for your own use this consideration doesn't apply of course. You mention an IDE in a browser. I think the IDE should run on the desktop and provide an text editor and optional command line compiler interface also. What use cases do you have in mind? I get the impression the focus is on building CRM websites, is this correct? How do you deal with resource management? For example if a code fragment tries to slurp a petabyte of RAM, is this safely sandboxed? How do you deal with search and inference control? Is the system open source? On Fri, Oct 19, 2012 at 4:54 PM, David Clark <[email protected]> wrote: > I have been working on a project for the last 10 years (off and on) that > runs all the time and can be programmed while running from the inside by a > program and from the outside by a programmer. All the information a > programmer has at his/her disposal is available to other programs at run > time, including the documentation. Like LISP, programs can be used to > create source code at compile time and arbitrary code can be created and run > at execution time. With the right security, all information about the > system is available in human and computer readable form while the system > runs. Compiling is done automatically when the source code is saved. An > IDE is provided that works out of a browser. A user interface can be > created using a built in Content Management System (think Wikipedia with > database, report generators etc). High level messages are sent to Server > Objects through interfaces that can also send any number of messages to > others or just forward the message to other Objects either locally or > anywhere on the Internet. > > What if: > * A number of AI researchers could create their AI that works well > in some domain and it could be exposed through a simple interface over the > internet. (Any level of security needed is included.) > * Other researchers could include that functionality through those > interfaces and build on the work of others. > * Each researcher would have physical and direct control over the > system they created and others could pick and choose what systems they want > to use in creating their own. > * The design of any of these modules could be anything where only > the code that exposes the interface on the Internet would have to be > similar. (Source code would be provided.) > * The whole project could start as small as barely useful and grow > and get smarter over time. > * At all times some functionality would be available. > * The result could be a single smart AGI or even a bunch of brother > AGI's with different abilities and strengths. They could still all use some > common functionality, available to all. > > If anyone is interested in this project, please go to www.rccconsulting.com > and check it out. > > I am not trying to create a fight about computer languages. I have a > section on why I designed most parts of the system. Many of the obvious > questions are answered there including "Why create another computer > language?". > > I think the biggest problem in implementing AGI is that we don't have the > right tools. I have no proof of this comment and I know some will disagree. > Many others might agree but disagree with what I have designed. Fair > enough. If I can get a few good suggestions to improve my design over the > next 6 months then I am willing to answer posts on this list or privately. > I am contracted to provide a system based on this code by August 2013 and I > currently have in excess of 45,000 lines of C working. > > My email address is [email protected] > > David Clark > > > > ------------------------------------------- > AGI > Archives: https://www.listbox.com/member/archive/303/=now > RSS Feed: https://www.listbox.com/member/archive/rss/303/1658954-f53d1a3f > 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/303/21088071-c97d2393 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-2484a968 Powered by Listbox: http://www.listbox.com
