On Liu-Chen's idea of how to use contradiction matrix when just one parameter is known

Y. B. Karasik,
Thoughts Guiding Systems Corp.,
Ottawa, Canada.
e-mail:karasik@sympatico.ca

Published in this issue article by Profs. Sreebalaji and Saravanan [1] builds upon the idea of Chih-Chen Liu and Jahau Lewis Chen of 2001 [2]. Liu and Chen started their work with the correct observation that designers often know just what they want to improve and don't know what will worsen as a result of their improvements. What they failed to mention, however, was that this usually happens at the early stages of design and the worsening parameter usually transpires later. Nevertheless, Liu and Chen wanted designers to use 40 Principles of contradictions resolution right away, even though the worsening parameter is not known yet and there is no contradiction. How reasonable was this objective ?

Contradiction matrix and 40 principles of resolving technical contradictions have the following prerequisite to their use: if improving one parameter of a technical system by the known/conventional means results in worsening some other parameter then resolve this contradiction by applying the inventive principles located in the corresponding cell of the matrix.

What are the conventional means ? Suppose that heat loss has to be improved (i.e. reduced). One of the conventional means of doing so is to apply insulation. Any parameter has such non-inventive conventional means of its improvement. 40 Principles are supposed to be used only after such conventional (and less tricky) ways of improving something were tried (or contemplated) and resulted (or were known to result) in worsening something else. They are not supposed to be used before that. Hence, the objective of using contradiction matrix (and 40 Principles) when designers know just what they want to improve is misguided. Liu and Chen were apparently unaware of the above prerequisite.

But suppose that they were simply concerned with the hypothetical situations when conventional (and less tricky ways) of improving a parameter are unknown and one wants to try 40 Principles. How could this be done ?

Liu and Chen proposed to use those principles, which only improve the required parameter and do not worsen ALL others. Do such principles exist ? Liu and Chen do not answer this question. They even do not try to identify such principles. Instead they settle for a "slightly" less: find principles that just have the highest probability of improving one parameter without worsening ALL others !

How great could such the highest probability be ? One may think that it has to be high just by the virtue of being the highest. In reality, however, even the highest probability can be as low as 0.

To evaluate this highest probability one has to accomplish the following:

  1. collect a large and representative sample of inventions;
  2. indentify the inventive principles used in them;
  3. count how many times a particular principle improved a parameter and did not worsen others;
  4. divide this count by the size of the sample.
Suppose that for principle P the count is C(P). Suppose that the size of the sample is M. Then the probability that the principle would improve something without harming something else is approximately C(P)/M.

But Liu and Chen propose another way of estimating this probability. They count occurences of a principle in the cells of the same row of contradiction matrix. The more occurences the higher the probability, according to their logic, that the principle would improve the parameter associated with this row without worsening others.

Here is a simple counter-example to the Liu-Chen estimate. Suppose that principle A occurs in just one cell of a row and principle B occurs in all cells. It does not mean that principle A is used less often. It only means that it is used for resolving of just one type of contradiction only, whereas B is used for resolving many other types. It may happen, however, that the type that A resolves is the most frequent one and occurs in every N inventions out of M. At the same time the types of contradictions that B resolves are rare and occur in every K inventions out of M. If K is less than N then the probability that principle A would improve a parameter without worsening others is higher than that of principle B.

Liu-Chen's method is built on the false assumptions, is useless in practice and contributes nothing to TRIZ.