Hi Team,



Neil Nelson wrote:

> I worked on the dynamic IP lookup code yesterday and hope to

> have much of that working today.



Since you took over this area, I'll just pass you the problem of

transmiting to computers behind firewalls. My guess is that your

(Neil's) naming procedures will solve the problem or maybe what is

discussed below.



> I was talking to Harold about the kinds of interesting things

> an Internet connected supercomputer might do, and one of

> the ideas that seems useful to me and likely useful to many

> others including potential customers that is a very central AI

> area would be to build an NLU including a knowledgebase in

> the manner of CYC.



If you (Neil) remember the link I sent you recently, www.activetools.si

has already devoloped methods for direct H2H (hardware to hardware)

computer merging. "Supercomputer" would probably not be a good name for

what *we* are building, since the main positive point of our system is

that it's working with cooperating individuals, not one big collective.

Our system is a parallel processor + you know, each machine is good for

something else, one big collective does not cover all the abbilities as

well as a parallel processor can.



> The components would be:

> 

> (1) An English parser of very good precision, and one that

> did not have a restrictive license, or rather one that had

> our own license.



Spread the program troughout multiple computers, then brake down the

text to smaller units, crosspost and recombine the results.



> (2) Creation of a knowledgebase by reading web pages such

> as web encyclopedias and basic fact sites and then using the

> prior parser to extract that data into a knowledgebase.



Haha, yeah... =] Or monitoring chatrooms. =]



> (3) Providing an ability to access the knowledge base by,

> e.g., manually typed inquiry and via automatic program as

> might occur from AIMetaSearch.   This kind of knowledge-

> base and English text processing capability would be very

> useful for AIMetaSearch.  And it may be a way of utilizing

> the Internet cluster resource to obtain customer revenues.



Anyone/anything can use the ICI's services as long as it can get

connected,



> And then there is the close relation and potential for voice

> recognition and that kind of interface to the knowledge base.



And a phone-based search engine, eh? =] This might take a lot of work if

we dig into it. =]



> Harold mentioned that his computers are connected together

> using some NT directory method where the other computers

> may be communicated with by file transfer in a directory

> tree.  That is at the moment for Harold, in order to utilize

> the other computers on his network, we cannot write to an

> IP address only but must write to an IP address and then

> retransmit to a directory file from which the final cpu gets

> its intructions.  And then a reply must be sent back in the

> reverse manner.



Well actualy, this is simpler than you think. Since the Console uses a

file interface to the ICSA, if we didn't need an internet connection, we

would simply leave ICSA out of the chain, we would write our output to

the remote Console's input files. Of course we would need the other

ICSA's services here, maybe in a seperate application or maybe this

could be partly avoided by making all Consoles in a local net use the

same input file.



The NT directory method, as far as I know is the method of network

drives. It is very very common in big companies and/or any one-OS

computer comunities with a firewall in front.



> What sems to be required in order to utilize the computing

> power of all the computers in the NT cluster is an extra agent

> to start programs on the non-Internet connected but NT

> connected computer via a retransmit step.  The remainder

> may likely be then that communications for all processes in

> an NT connected cluster would be, as currently, though files

> in some common NT directory.  And then there needs to be

> some additional communication routing (addressing)

> information to get the communication to that particular NT

> computer that may just be a particular kind of path/file name.

> Some cluster computers would be identified as an IP plus

> some path/file name.



Since the procedure is to be so common, you might think of intigrating

the procedure in the very ICSA itself. The job would be exactly the

same, just with the Internet transfer aspect left out; a file or file's

meaningfull content would be transfered from one path to another

according to the given name of the target computer.



It just happened to pop to mind, that my Console could realy utilize a

command to ICSA to spread the transmition over all computers in the

cluster.



> What other interesting things can we do with an Internet

> cluster?



Very many things. 



Since tasks can be spread over multiple computers for processing,

Shaun's genetic-algorithm-building-neural-net thingy could be kept

running on such a system full time and the

this-is-input-this-is-output-what-is-program requests could be simply

sent to a particular share of computers in the system indicating

priority.



I could continue devolopment on my system optimizer (that was also said

to be once replacing procedures in a system with newer, more efficient

ones, like DirectX), that would likely eventualy get merged with the

genetic-neuron thing above, making a world-wide

self-programing/utilizing/optimizing computer comunity, which would

someday be able to simply leave us, human programers, out of the loop.



Since everything is available to anyone in the ideal ICI network, new

technologies would very quickly spread over all functions in it, which

means that the usability will boost by a f(x)=x^2, where x is the number

of diffirent functions. Older methods would harmlessly be replaced by

new ones, named the same way. Diffirent functions doing the same thing

in diffirent ways, would be functioning side by side, increasing the

stability of the system (and preventing situations like with the 486 and

the Pentium, where a 100MHz 486 is still better in intiger calculations

than any Pentium, with the notion that 486s are no longer used for some

weird reason), ETC.



C'ya!



--



Don't feel bad about asking/telling me anything, I will always gladly

reply.



Digging for info? Try AI Meta Search:

Http://WWW.AIMetaSearch.Com



MesonAI -- If nobody else wants to do it, why shouldn't we?(TM)

Http://WWW.MesonAI.Com

