That erroneous view is a fairly natural consequence of treating NULL as a value which exists in all domains instead of as a marker quite separate from the domain of values indicating the absence of any value. This error was originally committed by Codd, but he corrected it later (and by 1986 was advocating two different NULL markers, one for "Inapplicable and therefore absent" and the other for "Applicable but absent", and a 4-valued logic with values T, F, I and A); unfortunately the SQL crew didn't pick up the correction, still less the idea of two different nulls, and the ANSI standard explicitly states that its single NULL is a special value. A value that denotes the absence of any value is a bit of a loopy concept - presumably it denotes the absence of itself, so it would fit well into Charles Dodgson's works of fiction but not at all well into his textbooks on mathematics and logic.