Clayton Christensen, a Havard Business School professor, is a prolific author on innovation (see his company web site at www.innosight.com). His books "Innovator's dilemma" and "Innovator's solution" attract numerous readers.
I turned out to be not an exception and have also read some of Mr. Christensen writings. Namely, the widely cited article "The process of theory-building" (available for download from his web site).
Amongst others, it states that "a theory is a statement of what causes what, and why."
A few theories, ranging from the classic mechanics to TRIZ, immediately crossed my mind and I could not immediately grasp why, for example, the second Newton law F = ma is a statement of what causes what and why. Manuals from my car seemed to be a better fit: e.g. "if you switch the ignition key on, then engine will start", etc. And they explained why this would happen. Are manuals theories then ? - I wondered.
After some exercise in phrase-busting, I eventually rephrased the second law of Newton as follows: if you apply force F to a body of mass m, then it will start accelerating at pace a. But why ? Newton laws definitely do not answer this question and, hence, do not fit the Christensen's definition of a theory.
Further study of the afore-mentioned paper revealed how theories get improved. First, an anomaly that contradicts theory gets discovered. Then theory gets rectified to include the anomaly.
But do not computer programs get improved in a similar fashion ? First, unexpected/abnormal behaviour under some circumstances/inputs/data is discovered. And then program gets rectified to either eliminate or include this anomaly. The same, by the way, is applicable to any machine, not computer programs only. They also get improved in response to abnormal/unexpected behaviour.
In short, under the influence of that paper, I came to a conclusion that whatever definition of theory scientists and philosophers propose, it turns out to be more suitable for machines ! Moreover, whatever theory of theories they propose, it also turns out to be more suitable for machines ! For example, Christensen teaches that a distuinguishing feature of theories is that they get improved via anomalies discovery and their elimination. But machines also do pretty much so ! Are they, hence, theories ? Or Christensen theory of theories is not a theory of theories but a theory of something, unclear only of what !
Interestingly enough, with regard to theories and machines TRIZ promotes the opposite point of view. I remember paper titled "Theory as Machine". It argued that theories are very much like machines and subject to the same TRIZ laws of machines evolution.
This makes any innovator bewildered: are theories like machines, or machines are like theories ? Two respectable schools of thought cannot agree on that. And this creates a real innovator's dilemma: whom to listen to ?
To this dilemma, there is, of course, an innovator's solution: listen to Anti TRIZ-journal !