Principia Cybernetica Web (C) Principia Cybernetica Web (C)Parent Node(s):
I have myself developed some theories [1] about this latter subject which seem more detailed than what is offered in the manifesto. The main idea is that present-day human intelligence is based on the recombination of static concepts or symbols ("formation of associations of mental representations"), but that the concepts or symbols themselves are not constructed in a controlled way. The next level of evolution would be characterized by the ability not only to control the combination of concepts, but to control the formation of concepts and representations. Though this may sound simple formulated in this way, the implications are enormous, and the most important limitations of rational thought as it is used now, can be shown to be transcendable in a system which would be capable to consciously form new distinctions (concepts). How new distinctions can be created in a more systematic, controlled way is something which I have tried to explain in several of my papers [3,5,6,7,9].
About ultimate values: I do not see the necessity to introduce the "will to immortality" as a principle distinct from the principle of natural selection. Natural selection automatically values systems which are more stable or invariant ("immortal"), according to the tautological principle that stable systems tend to survive (are selected) while unstable ones tends to disappear (are eliminated). The main problem here is to define "stability", since stability is definitely not the same as frozen permanence. The analogy with the genes is a good one: genes survive since the beginning of life on earth (are in this sense immortal), yet they are changing all the time, without however losing their identity (i.e. the self-environment distinction is continuously maintained). If the maintenance of a stable identity, by means of a "growing", "developing" or "adapting" structure, is taken as an ultimate goal, all other values may be derived as means towards this fundamental end [11].
I am less comfortable with the "style" of the manifesto. I do not like to use words such as "God" or "Soul" or "Cosmic Creativity" (mostly words starting with capitals), in a scientific context. Though I can agree with the way these concepts are defined, I prefer not to use words with a lot of "mystical" connotations, because they are easily misinterpreted, since their meaning in our culture is determined much more by subjective, emotional connotations, than by intersubjective, more or less scientific definitions. I do agree with the use of such words if they are defined in a more formal way, which does not too much contradict their subjective connotations, but then they should be introduced only as derived concepts, not as primitive ones. What I like about the manifesto is that a first attempt is made to define them in such a way, but the text is too short to do this in an exhaustive way, which would eliminate all possible misinterpretations. In these circumstances it seems better not to use words like this at all. [ sorry, that I go on so much about minor stylistic features, but in the environment where I have got my scientific education (the Free University of Brussels, Europe), people tend to be very sensitive about possible confusions between science and religion, and I know that the use of words with religious connotations may be sufficient to create a bad impression among researchers, who in another respect would be very much interested to collaborate with the project].
Like mentioned in Cliff's text, a difficulty with semantic networks is the multiplicity of link and node types needed for reflecting the semantic richness of the domain. Moreover, there is some arbitrariness involved in deciding which types of links and nodes would be present in the network. A first step in starting the project would consist in collecting a list of possible types, and then reorganizing this list in order to make it as simple, transparant and complete as possible. I believe that node types should not only reflect "formats" such as : word, sentence, paragraph, ..., but also semantic categories, such as : value, object, predicate, proposition, rule, ... [3]
My own approach towards this problem of multiplicity of types would be to try to reduce all types to one primitive type, so that all special types of nodes and links could be recognized as higher-order systems of primitive elements. In my system, the elements are distinctions or connections (in the limit the same thing), and the principle for organizing them into higher-order concepts is called "closure" [4,5,7]. In practice it will not be possible to immediately translate all knowledge into primitive elements, and therefore we should expect to start working in an intermediate stage, where there are a small number of types, which could in a later stage be reduced to just one type.
About available software: of the list mentioned in Cliff's paper, I have only experience with Guide and with HyperCard. I like HyperCard very much, because it is very easy to extend the system with those specific functions you need (I have started building my own system in HyperCard [10]). Moreover, HyperCard does not cost anything (you get it for free if you buy a Macintosh). A drawback of HyperCard is that the rigid "card" format for nodes is rather limited for storing complex texts (especially with mathematical notation), but it is in practice possible to connect a lot of information in many different media to one card.
The idea of developing a custom software based on UNIX will be succesful depending on how much programming resources are available. I have no experience with UNIX or TEX myself, but I have recently read about a UNIX-based hypermedia system called HyperNeWS in an unpublished paper. I quote :
"GoodNeWS is a NeWS [Sun Micro Systems Network Windowing System] - based window interface that has been developed at the Turing Institute in Glasgow [...] written in C and PostScript. GoodNeWS provides a complete windowing environment which includes terminal emulation, graphics drawing tools (colour and black and white), use of captured images and a LaTeX previewer. HyperNeWS is a GoodNeWS tool which is similar in some ways to Apple's HyperCard. [...] HyperNeWs runs under the Unix multi-tasking environment which allows one to have several interactive HyperNeWs stacks at once ..." See further [13].
The use of electronic mail as a medium for exchanging information about the project among the collaborators is in principle a good idea. In practice, however, I am afraid that it will pose a number of technical and psychological constraints, because practically existing email systems are often still unreliable and not user-friendly. Moreover, many people who could contribute to the project still have no email facility. (I am thinking especially about people in the humanities who are not used to work with computers apart from word processing, or about people in technologically less developed countries, e.g. in Eastern Europe). I do not have a ready-made solution for those difficulties. In a first stage, of course, the project could be limited to those people who have the required facilities, but this may lead to a primarily American, "hard science" bias among the collaborators (which is perhaps not that bad, as long as one is aware of the bias). One could also think about integrating different media (print, telephone, fax, telex, email) into one network of message transmissions which would at some central point be collected and translated into the unifying medium of hypertext. In my own experience, it is for example impossible to communicate with Eastern Europe via email, though telex is no problem. Although I haven't seen it realized yet, I think that technically it should not be difficult to translate telex messages into electronic text files, or the other way around. With fax, on the other hand, the problem is that what is transmitted is not text but bitmapped images, and I fear that text reading scanners are still too primitive to translate bitmapped letters into computer code in an efficient manner. This applies to a less degree to ordinary typed or printed texts if the quality of the printing is good enough. People who use computers but not email could also simply send in floppy disks with ASCII files.
A more profound problem with such "distributed" preparation of texts is the coordination between different authors and different points of view. How can one check whether two different contributions are not inconsistent, redundant, or simply lack a clear connection? Possible solutions to this problem:
I am afraid that such practical problems may lead to a failure of the project, because of the overall complexity of the task. An advantage of the project as it is formulated now, however, is that it is very modular: it is easy to start with simple things and to gradually add more and more, thus building up complexity in a smooth, controlled way. Even if the project would break down after a short period, the provisional results would still be sufficiently interesting to make the whole enterprise worthwhile.
Another nice thing about the project as it is formulated now is the possibility to work with different dimensions, and then to make "projections" along one of those dimensions, thus providing a report with a more specific aim than the general aim of the network: for example, in the form of an encyclopedia, a dictionary, a bibliography, a history. I might suggest another such dimension which although not so fundamental, may be very useful in the present context, namely an overview with addresses and references of associations, organizations or centers in the CYBSYS-domain, which are active in supporting developments along particular lines or subdomains mentioned in the Principia network. The danger of course in adding dimensions is that the network would become too complex, but this depends mainly on the structure of the network, and the hardware capacities of the host computer system.
In my own research, I try to build a "structural language", which allows to express everything in terms of a network of distinctions and connections [12]. The dynamical, organizing principle is called "variation through recombination of modules and selective retention of closed combinations" [4,5]. Closure is a mathematically defined concept which allows to derive higher-order distinctions (and hence system boundaries and organizations) from lower-order ones [7]. It is possible to implement formal principles such as this in computer support systems, based on a Hypermedia interface, for structuring complex knowledge [10,5]. On the other hand, the same principle can be used as a basis for a true philosophy or world view, encompassing ontology, epistemology [2] and theory of values [11]. This research is still in a mostly theoretical, conceptual stage, but I think it could be applied without too much effort to a somewhat more concrete project such as the Principia Cybernetica network.