This is a reply to Chris Tweed's recent attempt to solve the problem of "nearly convergent knowledge" and thus defend a binary account of knowledge against a contrastivist alternative. Ingenuous as his proposal is, it still does not solve the problem.