User blog comment:HeresyHorus/Critical Strike Chance/@comment-1330314-20140415041810/@comment-1330314-20140416081219

I don't actually think crit is overpowered. I do think, however, that it's very hard to properly gauge how powerful the stat is because of its random nature. Switching to a persistent damage amp, which would leave DPS unchanged but change per-attack damage, would help figure this out.

You also seem to think marksmen are weak by comparing them to casters. This makes about as much sense as saying supports are weak because of their terrible jungling potential. Casters are, by nature, burst-based, and can thus deal much more damage in a short span of time than most marksmen. Marksmen are not burst-based: they're autoattack-based, which means they specialize in dealing a lot of damage over a larger time window than casters and will significantly outdamage the latter given time. That's also why they shine at different levels of the game: casters are usually extremely strong in the mid game because that's when they fill up on their core build, which typically relies primarily on AP/AD and cooldown reduction. Most marksmen scale off of AD, attack speed and crit, which means more items to purchase and thus better late game prominence at the cost of a weak early game. This is not true of all casters and marksmen, but this is the model they generally play around. If you're playing a marksman like a caster and you're not, or , you're playing them wrong.

I also disagree with multiple aspects of your second point: first off, the game is not balanced around the unreliability of crit. The stat has been untouched since its creation, and a non-random increase in power due to crit would by nature leave the damage of the champions building for it unchanged. Caitlyn dealing +25% increased damage on every autoattack has the exact same DPS as a Caitlyn with a 25% chance to deal double damage on every autoattack. The other part of this is that forcing players to stack large amounts of a single stat to actually make use of it is not good design, particularly in a game with a strong competitive scene where maximizing gold efficiency is required to win. Counting unreliability as part of a mechanic's balance is also poor gameplay design, as it at best takes gameplay out of the player's hands, and at worst doesn't solve the problems it's trying to avoid. I don't think Riot is counting on the unreliability of crit to balance it, and I certainly don't think randomness has anything to do with balance.

Regarding your last point, the existence of champions like and  shows that it's perfectly possible to have crit on melee champions and not crash the system. You imply that these champions would have massively increased burst when turning crit into a damage amp would reduce their crit-based burst, not increase it. If there really was a 1 in 4 chance for to guarantee a kill, then not only would every player pick it on these burst champions despite its unreliability, but making the damage amp constant would get rid of the abusive case where the item would provide too much burst. While all of these champions incorporate autoattacks into their build, and some even synergize with crit, their strong AD ratios and spell-based damage would make them synergize less well with crit than Yi or Yasuo, who would make far better use of it. Maybe AD casters would itemize for an AA damage amp if it were made reliable, but I don't think it would increase their burst at all when compared to the raw AD or armor penetration they could otherwise purchase. It would certainly give them more sustained damage, but at the cost of their burst, not to its benefit. You bring up as an example, but as it stands it doesn't give as much raw AD as, say, a  or  (both of which are significantly cheaper), and even when factoring in the damage amp it might provide, the item is gold inefficient until you itemize for more crit. For that same amount of gold, you could get a lot more AD, and therefore far better burst potential.