I find this project interesting, because in a sense I am pursuing exactly the opposite agenda, that is, I believe that our knowledge of how software will actually behave is empirical, not mathematical. However, I strongly agree with the critique of what you call "empirical methodologies" that appear, cause hype and disappear again. What we need to control the hype is properly designed empirical investigations that test the actual effectiveness of a method.
Thank you for finding this project interesting.
You strongly agree with the critique of the empirical methodologies that appear, cause hype and disappear again. On the other hand, you apparently disagree with the “Linear Software Models” project agenda.
Let me propose that you are not pursuing the opposite agenda, but rather a complementary one.
My point of view is that of the natural sciences. These have two complementary components to acquire knowledge and test its validity:
1- measurements in a laboratory – these are the properly designed empirical investigations that you suggest;
2- software theory – this should have a rigorous mathematical basis, as I suggest;
When testing the laboratory measurements against the theory, there may be:
a- agreement, meaning that the theory is a suitable description of the empirical investigations;
b- disagreement, implying that either the theory needs refinement, or the measurements need improvement;
In any case these two components are necessary and complementary for a deep understanding of what has been done – say, good software system design, reliable measurements, and so on.
The importance of the mathematical theory is:
a- to trigger new questions that one could not naturally ask, without the theory;
b- to provide a concise formalism to precisely describe the outcomes of the empirical investigations.
Finally, if you think that software is not amenable to natural science, I recommend reading the very nice book by Herbert Simon, “The Sciences of the Artificial”, in which it is claimed, among other things, that one can analyze the flight of an airplane or that of a bird by the same aerodynamic equations, independently of whether the airplane is a human artifact or the bird is a natural being.
Yes, I've benefited a lot from reading Simon. On the other hand I tend to think of him as more of a social and behavioural scientist than a natural scientist. I don't know much about aerodynamics, I suppose that for certain purposes one can sensibly abstract from the differences in motive power between mechanical and biological flight, and for other purposes one probably needs to take account of those distinctions. I'm sure it will always be helpful if we can find a mathematical basis for a theory, but I think that there are important discoveries to be made about the causal realm constituted by software, which are liable to be overlooked if one treats the software artefact as if it were just the same as the mathematical object on which its design has been based.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Software