I think that if you do this and few others do, the main result will be to confuse your readers or hearers -- and of those who are confused, when you've explained I fear that a good fraction of those who didn't already agree with you will pigeonhole you as a crank.

Which is a pity, because it would be good for far more published work to be universally accessible than presently is.

A possibly-better approach along similar lines would be to find some term that accurately but unflatteringly describes journals that are only accessible for pay (e.g., "restricted-access") and use that when describing things published on such terms. That way you aren't redefining anything, you aren't saying anything incorrect, you're just drawing attention to a real thing you find regrettable. You might or might not want a corresponding flattering term for the other side (e.g. "publicly accessible" or something). "There are three things worth reading on this topic. There's a book by Smith, a restricted-access journal article by Jones, and a publicly-accessible paper by Black."