RESEARCH ON THE PHILOSOPHY OF LOGIC REALIZATION
D. C. Engelbart
Contract AF 33(616)-6303
Copy No. 35
The objective of this project is to provide organization and stimulation in the search for new and better ways to obtain digital manipulation of information.
Dr. Douglas Engelbart, Mr. John Haynes, Mr. Jack Goldberg
During the second quarter of the project, work has been mainly directed toward inspection of existing logical schemes to begin the process of fitting each to the descriptive and functional mode of analysis outlined in Quarterly Progress Report 1. The process of studying these schemes is expected to occupy much of the remainder of the year; we therefore feel that a very extensive report on the progress to date is not worthwhile at this time. There are, however, a number of interesting highlights which can profitably be mentioned.
For each of the schemes that is to be scrutinized closely, we expect to need a complete engineering description as well as a yetunknown amount of background references on the basic phenomena involved. We also expect to have to do a fair amount of literature searching for new and different schemes to add to our list. For these reasons, we feel that a carefully organized bibliographic file will be necessary to the project. Quite a bit of effort this quarter has gone into initial literature searching) requisitioning of documents, and setting up a filing and retrieval system. We are also setting up a special note file for storing and retrieving the special facts, considerations, and ideas which accumulate during all phases of project work.
Actual scrutiny of some schemes has revealed the following problems and considerations:
(1) Whatever parameter (or group of parameters) is used to carry and store the information being manipulated by the facilities of a given scheme must experience certain transmission characteristics that preserve the binary signal states in the presence of the inevitable noise and loss. This functional requirement is implicit in the general functional model described in Quarterly Progress Report 1, but it seems likely that the importance of the bistable-gain characteristic will make it worth separate study in the analysis of any given scheme. It will be important to study the significance of losses, delay, and power gain, as well as the significance of nonlinearities such as saturation, clipping, and unilateral propagation.
(2) It seemed originally that the storage and logic functions required of any scheme would bear separate scrutiny, since they are the two principal functional ingredients of a workable scheme. We now feel sometimes that storage is going to emerge as the most critical characteristic,
but at other times we feel that if we have both a general logic facility and the proper gain characteristics, we can synthesize the storage function.
(3) The interconnection compatibility of the various separate elements of a scheme's system can affect the elementary logical facility required. Statements as to the allowable connectibility between such elements can be seen necessary to the scheme's description.
(4) Multi-phase clock systems, when required by the refractory characteristics of a given scheme, will not always make a workable system when combined with a basic logical facility which might otherwise be sufficient. For instance, a two-phase clock system with a core-diode, Laddic, inhibit-core, or Biax type of scheme, could not function with only the NOR facility for its basic combinational logic.
(5) We can tabulate large numbers of different schemes if we include those whose descriptions differ only by the addition or deletion of some small feature, for instance in the system part of the description. This is likely to present something of a problem, since schemes involving tubes and diodes, transistors, transistors and diodes, and cores and diodes, for instance, have been developed in many varieties.
(6) It is also something of a problem to know how detailed to make the description and analysis of the scheme that involves networks of fairly sophisticated elements. Some of these could absorb a great deal of our effort in a really thorough analysis, but it is still too early to be sure that such effort will be to the best interests of the basic project objective.
(7) Any list of known schemes includes some which are inherently far more simple than others. It seems that the problems mentioned in Items (5) and (6) seem to be more pronounced as the complexity of the schemes increases. It seems possible that a measure of scheme complexity can be developed from the number of materials, parameters, and relationships involved in a scheme, together perhaps with the number of levels of organization which seem to separate the logical networks from the raw phenomena upon which scheme function depends.
Two significant generalizations seem pertinent here. First, the trend of scheme development in the future seems almost sure to be toward the simpler schemes, which are more compatible with the goals and possibilities of those research activities variously labeled "microminiaturization," "molecular electronics," and "integrated electronics." Second, considerations such as those mentioned in Items (5) and (6) give the more complex schemes a somewhat doubtful significance to this project. To illustrate, we can point out that a scheme which is perfectly acceptable in the spirit of this project could be made up utilizing sturdy trees, heavy cables, big blocks (pulleys), and unmanned bulldozers. This would be a very complex scheme, and its significance to the project would probably be slight. It seems safe to hypothesize that
there is no end to the number of schemes that can be devised if we are free to increase complexity endlessly, i.e., we could always dream up a new scheme, more Rube-Goldbergish than the last.
PROGRAM FOR THE NEXT QUARTER
We plan to continue working upon the bibliographical file as a steady, part-time aspect of the project. The rate of effort for the first half of the contract year has been lower than was anticipated--approximately seventy percent of the project funds remain unspent. We plan to increase the rate of effort accordingly.
In line with the above program discussion, we expect to apply ourselves more to the analysis of the simpler schemes (those which can be said to be "closer to Nature"). With a more restricted field of study, and with less complicated structures, we hope to be able to utilize our descriptive and functional models in greater detail, and to obtain a good test of their worth to the project.