This myth was created by immigrants from the former USSR to facilitate selling TRIZ in the US. It was taken by English speakers without a grain of salt and repeated in various books and papers on TRIZ in English.
The readers, of course, presumed that "patents" that Altshuller analyzed were similar to American patents and the number 1,000,000 impressed them immensely. But some started doubting how could Altshuller have analyzed this huge number of American type patents when reading and analyzing just one of them takes a lot of time.
The truth is that Altshuller never analyzed patents in American sense and they
even did not exist in USSR.
What he analyzed were just concise abstracts of inventions, consisting usually
of one phrase.
The Soviet State Committee for Inventions and Discoveries published
such abstracts in its monthly bulletins.
Abstracts were written according to the following template:
The idea of an invention could be easily extracted from such an abstract in a matter of minutes. Thus, it is quite plausible that Altshuller indeed analyzed 40,000 such abstracts, as claimed in his book "The algorithm of inventing".
But in the US this number was inflated first to 200,000, then to 1,000,000 and eventually to 5,000,000 patents in the American sense (not invention abstracts of the Soviet style) ! No wonder that serious people refuse to believe this numbers.
This allegation is impossible neither to prove nor refute because most inventors never described how they arrived at their inventions. When in the rare instances they did, it always looked like random trials, which gave Altshuller the grounds to claim that inventors, conversely, always invent by random trials and errors.
But suppose that some inventors indeed knew certain methods but never disclosed them. In this case Altshuller's work amounted to discovery.
He also undoubtedly invented some methods, such as IFR, SuField Analysis, etc, which definitely nobody used or knew before him (even subconsciously/intuitively).
Thus, his work was a combination of discovery of known but never disclosed methods with inventing new ones (which nobody knew before him) with collecting the previously published ones (such as inverting, dividing, combining, etc).
The above misconceptions are widely spread nowadays and very damaging to reputation of both Altshuller and TRIZ.