The issue of how far to enchant Magicbane has been debated endlessly on rec.games.roguelike.nethack. Conventional wisdom says that +2 is the optimal balance of special effects with damage, but recent research and math by some of that newsgroup's top hackers has cast this into doubt.

The most damning argument is that Magicbane, enchanted to +7, just plain deals more damage on average than +2, even accounting for the greater probability of special effects. (When Magicbane's special effects go off, each one adds +1d4 damage). The second argument is that Magicbane's effects are actually only minimally useful. Cancellation appears at first glance to be the most useful effect, but even at +0, there's less than a 10% probability that any strike with Magicbane will cancel its target. Furthermore, Magicbane's cancellation only works at melee range. This makes it pretty marginal against any foe you'd actually care about cancelling. Probing is only insightful about 10% of the time, and that's out of all probe attempts, which have at best a 40% chance of happening. When a probe isn't insightful, its only effect is to deal 1d4 damage. This is somewhat useful, but this is also the predominant effect at +7. As for scaring and stunning, these might be useful, but even at +0, they're uncommon effects. The general consensus is that it's better to destroy a foe in 2 hits than scare it in 3 and have to make a kill shot with a spell or hurled dagger.

So, the new opinion is that it's better to enchant to +7 if you're using Magicbane as your primary weapon. Older wisdom held that it wasn't a good idea to do this - after all, athames aren't weapons meant for major fighting - but in practice, relatively few weapons fare better in combat than Magicbane. In fact, only Grayswandir, Snickersnee, Excalibur and Stormbringer are consistently stronger. Frost Brand, Fire Brand and Mjollnir often, but not always, are.