Read the introduction of this paper, and you will see the human genome is badly annotated. And if I remember right, 40% of the genes are still of unknown function! Really, an appaling job.
On a more serious tone, all genomes are still bad, because there is still a lot to be discovered. But some are worst than others, the "classic" model organisms - as human, mouse, fruit fly - have the better genomes (in sueqence coverage and annotation), all the others are worst, to a great extent because their annotation is almost completelly automatic, thus import errors from others annotations and introducing new errors.