There is a peculiar contradiction in TRIZ. It is a contradiction between the complexity of the process, which it tries to formalize and the simplicity of the algorithm, which it touts for this purpose. The process is creativity. And the algorithm is ARIZ.
If one takes a closer look at ARIZ, he will notice that it is almost a linear procedure: all steps are arranged in one line. There are just one or two loops and roughly the same number of branching points. From the algorithmic point of view, ARIZ is very primitive. Can it be adequate for modeling such a complex process as creative problem solving ? Are our mental processes have just two loops and two forks ?
For comparison, take a look at algorithms of chess playing. They are extremely complex, although to play chess is, probably, simpler than to invent.
When one solves problems, his thought moves very intricately. He asks himself various questions. Depending on answers found, he asks new questions, etc. The ability to solve problems depends on the ability to ask good questions, in good time, and about the right things.
To Altshuller all these questions and answers seemed to be unsystematic trials and errors. He did not recognize any algorithms behind them. Moreover, he did not believe in them.
He was not right. The process of asking questions is not a process of unsystematic trials and errors. Were it so, the bad problem solvers would ask innumerable questions. Whereas they ask just a few (and not the right ones) after which they are stuck. They do not know what else questions to ask. No further trials and errors take place.
On the contrary, the good problem solvers know which questions to ask, in which order, and about what. They know patterns (or even algorithms) of asking right questions. Paradoxically enough, they are very often unaware of this their knowledge.
That is why the right approach to formalizing creativity is recognizing these patterns and algorithms (what "Thought Guiding Systems" Corp. is working on). It is not a step back from TRIZ to the lists of control questions (or checklists). It is a step forward from the current TRIZ to the algorithms of asking right questions when solving a problem. At present ARIZ mostly consists of orders: do this, do that, do not think on your own, just follow the instructions. (That is why many people do not accept it.) The investigative portion of the problem solving is almost absent in ARIZ. Whereas the real problem solving process is a balance between inquiry and applying some tricks. In the current TRIZ this balance is violated.
That is why it does not fly. Or it "flies" to the same degree as all planes of the XIX century. When it is supported by something (say, by a crowd of TRIZniks), it "flies". But as soon as the support is removed, it falls. Needless to say that this is also the reason of why it does not soar.