Recently Mark Barkan presented an interesting account of his experience with Invention Machine. In 1991 at a design fair in Chicago he stopped by a booth of Invention Machine Corp. He got interested and with the help of the machine (and the representative of IMC) he arrived at all the inventions made by his firm in the recent years. In addition to these inventive solutions the machine also generated many others which seemed idiotic to Mr. Barkan. Nevertheless, he bought the machine. But without a representative of IMC the machine did not work. He called Simon Litvin (who at the time worked for Invention Machine Corp.) who duped Mr. Barkan for the second time by saying that in order to work with the machine one has to first know TRIZ! The story ends here and Mr. Barkan did not elaborate on how study of TRIZ affected his ability to work with the machine.
The account is interesting in two respects. Firstly, I am unaware of another such an account published by a victim of the Invention Machine Corp. Usually they remain tight-lipped or praise the software in public. The only similar account was reported by a former IMC employee, who wrote to me that the companies in France that bought IMC software stopped using it in two weeks after purchase (for the same reason as in Barkan's case).
Secondly, the story clearly demonstrates the strong and weak points of TRIZ. It is strong at obtaining those solutions which were already obtained without it. Or more specifically, at recognizing the past concrete solutions in the vague abstract solutions, which TRIZ generates.
I can imagine what really happened to Mr. Barkan in Chicago. The representative of IMC helped him to transform his past problems into technical contradictions. Then contradiction matrix was invoked to produce a set of vague phrases pretended to be hints for possible solutiuons to the problems. Mr. Barkan was quick to recognize those solutions in the hints which he already knew. The rest seemed idiotic to him.
However, when he tried to find solutions to as yet not solved problems, "the machine refused to work". NO KNWOLEDGE OF TRIZ WOULD HAVE MADE IT WORK in spite of Litvin's suggestion. Because TRIZ itself is only good for generating vague hints of no use until solution is found.