Anselm's main mistake is in his idea of God itself, not in his proof.
Moreover, Anselm cannot escape the difficulties of his Neoplatonic idea so easily as Barth seems to imply when he argues that Greatness is not a description of deity but a mere rule for our way of thinking about deity, a mere negation that anything could conceivably be greater. For the positive nature of God which escapes us not only cannot be known by us to be necessary, it cannot even be necessary. The only rules that make sense about 'necessary' imply that its referent must be abstract, so that even the 'description' which God Himself could give of His 'essence' must be logically weak. Hence the admission of divine accidents is obligatory.
There is another consequence of these rules: if God, merely as such, merely as omniscient and the like, can be necessary because--though only because--of His extreme abstractness, then so can 'world as such', as merely 'whatever, other than Himself, God knows'. For this is equally abstract. Thus the 'contingency of the world' would become that of this world, not of there being some world or other.
The question of the ontological proof is simply this: what way of thinking about contingency should we adopt, and what are the consequences of this adoption? Our decision will implicitly determine an entire metaphysics. Neither Anselm's way, nor that of most of his critics, has much to commend it, so far as this writer can see. We need a fresh start.
If a man is thinking about fairies and upon being asked what he has in mind replies, "a mere idea," he may be further