Re: [SystemSafety] Logic

From: Michael Jackson < >
Date: Tue, 18 Feb 2014 17:14:51 +0000

Perhaps the discussion about scalability needs to be tightened up a little.
  1. Questioning 'scalability' assumes, I think that work 'on a larger scale' is in some respects different from work 'on a smaller scale'. Are there such differences, and, if so, what are they?
  2. Are we discussing the applicability of particular formal methods such as X, Y and Z, as described in their respective books and papers? Or the applicability of disciplined formal logic and mathematics generally?
  3. Is the question whether the putatively applicable method or discipline is locally and partially applicable (as, for example, arithmetic is locally and partially applicable to architecture), or is globally applicable (as, for example, COBOL was at one time claimed to be globally applicable to the whole of a data processing system)?
  4. Self-evidently we expect logic to save us from logic errors in development. Are there other kinds of error? If so, what saves us from them, and how is it related to the use of logic?
  5. Is it true that formal expression of any requirement, design or other development artifact is superior to informal expression? If not, how are informal and formal expression related in a development?
    • Michael Jackson

At 16:48 18/02/2014, Martyn Thomas wrote:
>On 18/02/2014 16:10, Derek M Jones wrote:
>>Some of the points in this article on QuackWatch provide good advice
>>to anybody who wants to evaluate the claims of formal methods
>That feels to me like abuse. If you want to discuss the merits of
>formal methods, let's do it professionally.
>I see this table in the article that you reference. I think that the
>work on formal methods by computer scientists meets the definition
>of Science in the table (except that we are not discussing physical
>processes, so mutatis mutandis), whereas the criticisms of formal
>methods seem more closely to correspond with Pseudoscience. I can
>give you examples to match most of the boxes in the left-hand
>column. Can you do the same for your claim that formal methods do
>not scale to more than the simplest problem?
>Science Pseudoscience
>Their findings are expressed primarily through scientific journals
>that are peer-reviewed and maintain rigorous standards for honesty
>and accuracy. The literature is aimed at the general public. There
>is no review, no standards, no pre-publication verification, no
>demand for accuracy and precision.
>Reproducible results are demanded; experiments must be precisely
>described so that they can be duplicated exactly or improved upon.
>Results cannot be reproduced or verified. Studies, if any, are
>always so vaguely described that one can't figure out what was done
>or how it was done.
>Failures are searched for and studied closely, because incorrect
>theories can often make correct predictions by accident, but no
>correct theory will make incorrect predictions. Failures are
>ignored, excused, hidden, lied about, discounted, explained away,
>rationalized, forgotten, avoided at all costs.
>As time goes on, more and more is learned about the physical
>processes under study. No physical phenomena or processes are ever
>found or studied. No progress is made; nothing concrete is learned.
>Convinces by appeal to the evidence, by arguments based upon logical
>and/or mathematical reasoning, by making the best case the data
>permit. When new evidence contradicts old ideas, they are abandoned.
>Convinces by appeal to faith and belief. Pseudoscience has a strong
>quasi-religious element: it tries to convert, not to convince. You
>are to believe in spite of the facts, not because of them. The
>original idea is never abandoned, whatever the evidence.
>Does not advocate or market unproven practices or products.
>Generally earns some or all of his living by selling questionable
>products (such as books, courses, and dietary supplements) and/or
>pseudoscientific services (such as horoscopes, character readings,
>spirit messages, and predictions).
>The System Safety Mailing List

The System Safety Mailing List
systemsafety_at_xxxxxx Received on Tue Feb 18 2014 - 18:15:05 CET

This archive was generated by hypermail 2.3.0 : Tue Jun 04 2019 - 21:17:06 CEST