Engineering Logic


This Engineering Logic theme concerns both the exploitation of logic in the future of engineering disciplines, and an approach to exploitation and automation of logic which considers this an engineering problem rather than a fundamental research problem.

Engineering and Logic for our present purposes are to be construed liberally. Engineering is about getting things done, generally building things which realise some preconceived purpose. Logic is the sphere of formal a priori truth, encompassing mathematics, and crucially for engineering, all that supports the construction and exploitation of abstract or mathematical models. Engineering is conceived as a discipline which is to be increasingly dominated by modelling techniques which permit the construction and evaluation of a design prior to physical fabrication of its implementation.

The increasingly dominant intellectual content of engineering problem solving, the business of modelling, is at bottom pure logic. Software supporting these intellectual activities will be more effective when it is built on solid logical foundations.

This prospective future development may be related to the digital revolution which we are all now expecting or experiencing. The logical revolution, as yet scarcely anticipated, flows from the same underlying imperatives about the way in which information must be represented if we are to be able to manipulate it effectively.

Digitisation is a prerequisite of information being processed by computers. What can then be done with the information by computers depends on how the information is represented. A static image represented as a bit map can be displayed but can be manipulated less effectively than a representation of the data which contains more structural information. A movie of a dynamic three dimensional experience can be represented as a sequence of bitmaps, but to permit interactive navigation more sophisticated representations are required. Ultimately we will need computers to understand the data which they manipulate and to be able to reason about the behaviour as well as the appearance of the system described. To represent a system in a manner which is adequate for the purposes of many different kinds of software which may be required to work with it, and which is open-ended in terms of the functionality which may be delivered, a logical approach will prove beneficial.

Logicisation is a natural stage in representing information in ways which permit open exploitation and manipulation. What may now be thought an exotic and improbable development will in due course be recognised as an economic imperative, not only for engineering purposes but also in education and entertainment, where models are equally ubiquitous.

While this engineering logic theme does not encompass artificial intelligence, is proposed as a foundation for it, and this possibility is an important part of the motivation for suggesting an important role for logic in engineering.

Alongside logic, key components include secure networking technologies supporting trustworthy widely distributed processing through market funded fully automated subcontracting.

Artificial Intelligence

Artificial Intelligence doesn't really come into my Engineering Logic scenario, it belongs in Intelligent Logic. The reason for mentioning it here is that Engineering Logic is intended to provide a suitable basis on which to develop intelligent capabilities. Engineering Logic concerns that narrow domain in which the correctness of a solution to a problem will be formally demonstrable.
Checking such a demonstration will be a routine algorithmic computation requiring no intelligence, but discovery of a solution and its demonstration will in general be much harder because of the size of the search space.
It is the role of intelligence to find a solution in spite of this difficulty. Engineering logic encompasses all the methods of finding solutions which do not depend upon intelligence, and provides a platform on which to implement intelligent logic which operates within the same problem space but seeks to match or exceed human intelligence in the solution of this class of problems.

Formal Maths
Real formal methods people might think that mentioning formal methods here is a bit of a con. The kind of formal methods I have in mind are not much like the "specify, refine, verify" variety which is typical of this approach to engineering. However, in a broad interpretation formal methods might be considered to be any method which involves formal notations and proofs.
Which sounds to me just like applied logic. There is one other reason. Engineering Logic shares with formal methods a concern for assurance of correctness. The methods proposed are advocated as offering assurance and productivity, rather than assurance at the cost of productivity.

Engineering Logic is all about engineering applications of logic, so logic is the very center of the enterprise. Certain very specific doctrines are involved. Philosophically the scoping of logic which comes along with identification of logic with necessity, analyticity and a priority, is important.

To maximise the extent to which intellectual problems can be undertaken on our behalf by computers it is essential first that we can communicate the problem and its solution precisely, and secondly that we can trust the computer to reliably check the correctness of the solution.
Formal logics offer both precision and trust. In addition to the need for trustworthy problem solving, distribution of the problem solving across a network must be possible without loss of assurance.

Distribution cooperative problem solving involving a flexible combination of human intelligence and computer assistance is intended. Distribution is beneficial in large applications, and in educational and recreational use. Economic factors ultimately will determine whether any technology can be developed and deployed. The internet provides the only marketplace in which thoroughgoing exploitation of logic technologies can be realised.

UP HOME © RBJ created 1996/4/15 modified 1998/8/26