The best way to get CoD (or any other) corrections into the 'system' is to publish a paper containing such. It's really hard to get such a paper published in a refereed journal, since the referee says "there's no science here". There is another way, however. If a particular star or stars are already in SIMBAD (a much bigger number now that the complete Tycho catalogue has been integrated into their database), then if you can show where the error lies, send a note to them (question@simbad.u-strasbg.fr) saying what you've found. I've done this quite a lot (many thousands of stars and other objects in the last few years), and as long as there's convincing evidence (a paper trail in the literature is helpful, if such exists), then the fix will get made within a few days. Often, the "errors" are merely objects that are given several times in SIMBAD with different names, and by simply linking them, the problem is solved. Examples are red variables with IRAS sources, or an O or B subdwarf in the PG catalogue with a source detected in the UV. Be aware also of the typical positional uncertainties in various catalogues. The comparison between the CPD and CoD is a good example: the CPD stars are usually within 5"-10" of their true location (often much better), whereas the CoD positions can be an arcminute or so off (although usually within a few tenths of an arcminute). Many of the examples Mati cited are probably real stars, but ones that have crummy CoD positions. Comparison with the sky usually allows one to sort out the problems. Remember also that the CPD was done on blue-light plates, whereas the CoD was done visually and goes to much fainter yellow-light magnitudes, so favors red stars to a great degree, even more than the BD/SBD. Taking all this into consideration, my experience working on identifying stars in southern spectral surveys is that there are not very many true errors in the CoD in the sense of missing stars and such. Cases such as Mati found where he linked a large proper-motion star with its position at the epoch of the CoD should certainly be reported to SIMBAD if the name is not already linked there (you have to check that, too, because somebody may have already noticed it!). \Brian