Basically the trouble with the tests that isolate factors and fragment situations is that they are derivatives of Gutenberg (albeit subliminal). These procedures won’t touch the realities of the all-at-once electronic world of configurations. Our Ford seminar simulcast was a crude sample of new approach but via older method, like skis on grass. The whole subliminal side of the Gutenberg era now comes home to roost, as it were, in the tester’s hay mow. (…) You see, the [Dewey] general theory leads to batch of mechanical model experiments. Can we get from my general theory that medium is the message to a similar set of mechanical models? Why not?
In his following letter to Skornia 2 days later (June 10, 1959), McLuhan set out the same point, again in relation to Dewey: “These matters are easy of test and valid for investigation. But they are not mechanical models of testing. They are nearer to Dewey’s ‘learning by doing’ pattern” (than to mechanical models).