From Paul Prueitt:12/12/2003 6:27 PM
The Knowledge Operating System
Relationship between the natural sciences and computer technology
Michael and colleagues,
I will say something first about autopoiesis and the causes that perhaps hold the knowledge sciences back, for being established in what would otherwise be a rational process - whereby areas of scholarly interest not reflected by the university composition would end up creating in the university a new academic department (or division).
Autopoiesis is a term invented by Maturana and Varela in the 1988 book, "The Tree of Knowledge". Along with this term came another, "structural coupling" which talks about how process chains occurring in the environment of a living system and process chains occurring within the system interact "across" a gap.
So each of us can in some way be described as having an internal reality and external reality. A large part of what this proposal writing is about is to make common adjustments in our internal perceptions. This adjustment is needed so that an external resource can be allowed to fulfill the needs now indicated by a wide efficiency gap between what we as innovators feel about what ought to be, and what we see as being in fact the social reality.
The proposal can go to either George Washington University or American University. Even given two days, this process may not be successful, depending on the faculty’s awareness of the plan and willingness to help.
But just suppose that the proposal does not get to NSF on the 16th of December (Tuesday). The exercise of bringing together a community and learning about a common purpose is sufficient.
So the first step is to allow those who can spend 4 – 8 hours this weekend working on a NSF proposal to identify themselves.
Then someone, myself if no one else will step forward, needs to offer a template and manage the various roles needed to fill in the details.
I will say that most of us feel that such exercises, as writing proposals to NSF, is hopeless.
The “funding” process has abused us all, because there are not enough federal dollars to fund everyone and because the selection process is geared in the way that it is geared (not being judgmental – just trying to establish a model).
A model of systemic abuse is based on my PhD thesis and on theoretical immunology, as applied to understanding why college freshman seem “infected” with an inability to even cognitively address algebra and arithmetic.
I have some writing about this “learned disability in mathematics learning”. I have also developed a learning strategy that is based on the notion that a curriculum can be descriptively enumerated into a set of topics (again – not precisely and not perfectly – but functionally and pragmatically). These topics can be sorted to create a type of model for a learner of the learner’s present state with respect to the curriculum. The categories are natural and are
(1) topics that I understand,
(2) topics that I do not understand and
(3) topics that I do not know that I do not understand.
Such a learning strategy might apply to the retrieval problem faced by our society’s love affair with the Internet. In fact such a strategy might be built into a Knowledge Operating System (KOS), as I have discussed with others, most notably Don Mitchell.
In the late 1990s I developed an operational system that was for me a precursor to the KOS, although at the time I had not found the language that later occurred in Don and my conversations starting in 2001. The materials are partially published in the description of a many-to-many communication system. The material is scattered about in my web site. In any case, the NSF proposal needs NOT to expose this work, as we certainly would not be funded in this case.
However, if the group were to move in this direction, the NSF program manager community might see it as a useful next step.
The conjecture is that the current foundations for computer science is not exactly correct, that something is missing from how it is formulated by academic computer science and by the Information Technology business processes.
This conjecture is grounded in various ways; none of these ways are without controversy. Sir Roger Penrose has made his attempt at revealing the core problem with formal models of natural systems. Robert Rosen’s career exemplified for some of us the distinction that should be made between a formal system and natural system.
For me, the conjecture is grounded in what appears to be the need for a stratification of formal models into level in which coherence is not sought between levels of organization. What is a “level” of organization returns me to the notion of autopoiesis, where there is a need to separate the model of the system’s processes, and the processes that are outside the “system” in the system’s environment.
For me, the “non-stratified” formal system must fix a set of axioms and rules in order to be a formal system. A computer program, and even a complete programming language, is similar in this regard to a single formal system. The relational database, for example, fixes the data schema in order to allow programming to depend on regularity.
XML steps us away from this dependency on a fix pre-established set of axioms and rules. If someone wishes to claim, and many do, that fixed axioms and rules is ok; then I will argue that they are incorrect when we ask more from programming languages then what we have today. And we, as a society, are asking for this.
Topic Maps takes us one step further by asking that the gap between the computer structure and the human perception of the real natural world by reified (made human like).
A completely thought through Python 3 language will also need to have a set of fixed axioms, or what counts as axioms, but perhaps the community that we have gathered together will make an attempt to open up the axiomatic structure so that the language can evolve within a simple set of principles.
I know that the first reactions by many will be “this is NOT the way to get NSF funding”. But frankly that is not the point. The fact may be that no proposal of this nature would even be considered by NSF, simply because the foundational concepts are destructive of some assumptions made by anyone who would be an NSF program manager. So we have to seek other avenues to bring this work into the fore field.
The point is that we as a community of knowledge and computer scientists might work on the intellectual issues with a clear mind, and then express this clarity within a new programming language that allows the type of openness that XML and Topic Maps need. This new programming language might be Python 3.