Ok, I think you may have misinterpreted some other comments then. The argument was that "proven" in "innocent until proven guilt" is too high a bar for a low-stakes internet discussion.
Just get some background rates, and assume that people are guilty with eg 0.1% probability. (Just a made a up number. Real priors should depend on a lot more context.)
So... base assumption is innocent.
That's all I was saying.