YESTERDAY, TODAY AND TOMORROW
YESTERDAY
The Electric Abacus is just a "connection machine". A cynic might say that all it does is wire zillions of connections to each and every arithmetic, logic and control operation and therefore 1,248 x 128 would have its own circuit to the answer, as would 13 + 289 and so on so the Electric Abacus is trivial. But given that it could have worked as stated, both the Japanese and American clerks would have lost in 1945.

SHORTCUTS

But the geniuses of the ancient world in Babylon, China, Greece and India and perhaps elsewhere came up with shortcuts, ie rules such that those who memorized the rules and practiced their application could do arithmetic problems with great speed and power.

Likewise, Aristotelean logic (which later became Boolean or combinatorial logic) provides a set of shortcut rules for logic problems of any length because there are a few basic patterns of logic in the more fundamental truth tables which can be applied to the longer problems. There are also short cuts to the basic If-Then control procedure which should enable it to be programmable.

The connection machine substitutes the inexorable results of wired connections and switches for the inexorable results of arithmetic and logic. At least that is how we think of arithmetic and logic. Whatever the valid conclusions are, they MUST prevail, or so our minds tell us. Those defying such reasoning are considered at best as in error and at worst, deluded or mentally "challenged".

The enormous size required to build an Electric Abacus does not refute its conceptual capability. As for size, consider the size of the human brain with its 100 gigabytes in neuron capacity and at least two "wires" for each neuron capable of binary operations. So there is a biological form of Electric Abacus. And as for hardware shortcuts, consider that Steve Kilby received the Nobel Prize for his work on the Integrated Circuit, mainly because he miniaturized the system by dispensing with the wires.

PROGRAMMABLE?


Here is an example of a genuine program in action in a modern computer. The program allows the user to type in numbers in series, each one after a display prompt by the computer, and the machine calculates and displays a running total. Can an Electric Abacus do this? It would seem so. It can do all of the steps, one at a time as we have discussed earlier. Why not have over-ride switches which connect all of the individual parts to that series of instructions and set them in motion after the parts have been programmed?

Thus the first number is entered on the first abacus according to the rules for coding numbers. That abacus is electrically isolated from the others after that because it has no switched on connection. The display abacus can be connected to it later by throwing a switch and at that time it will display the number in lights. Also at that time the number entered into the first abacus can have a switch thrown to connect it to a second abacus which will do the arithmetic calculation according to the arithmetic rules determined by the wired connections between them. Next the result of that connection is sent to the display abacus when the timing calls for it and the display will give the added result and call for the next number to be entered.

If that can be done, it is a genuine program. The steps in it are pre-programmed and then executed by throwing by-pass switches which make the If-Then connections to the next step.

We still have a connection machine and in its primitive state it is nothing more than light bulbs wired by series and parallel circuits to on/off switches. But those rules for short cuts provide a huge saving in the number of components required in the system and would have made it feasible even for the ancients to have used first generation computers as powerful as those used at Harvard in the 1940's but maybe 100 times bigger. Hammering out tiny wires on stones would have been a major industry in Babylon.  What might they have done with 1,000 years of improvement thereafter?
TODAY
Do a search on "Robosapien". This $100 walking-talking, humanoid toy-bot will be in ToysRus stores soon. Developed by a NASA scientist, it could be modified for wired or wireless connection to a pc.

TOMORROW
We KNOW how to teach natural language to a Robosapien. The NLP problem is solved at a conceptual level but it would take enormous shared manpower to do the teaching as might be available online. It would also take enormous computing power to enable the NLP to run with fluent (rapid) conversation.  Cubits anyone?
According to a recent Popular Science article, 80 qubits would pack 151 trillion gigabytes of processing power into 80 entangled bits. Whether by quantum computers or otherwise, the miniaturization of computing has has implications both positive and negative:
http://www.newscientist.com/news/news.jsp?id=ns9999.6488
http://www.dustnetworks.com

The burning question for tomorrow has to do with the control of the machine. Whose meta-logic will
determine how it is used? Do we trust the "Iconographers" to be the high priests? Who will be technolords and who will be technopeasants and do on?
What might Robosapien as a new NLP-AI enabled Artificial Life Form look like?  K-bot by sculptor-roboticist David Hansen has natural-like skin and 24 facial muscles, with 28 human-like expressions all provided by about $400 in parts:
http://www.Human-Robot.org
              References/Citations available on request
                                          FINI
Hosted by www.Geocities.ws

1