(I'm keeping it generic here because this goes beyond US politics and is true for any democracy that has a party-like system.)
Further, even if by some miracle all the fraudulent parties completely cancelled each other out, it would still have a corrosive effect on the ability of the population to trust the outcome of the voting. First of all, how exactly does one go about proving to the populace that all parties committed offsetting amounts of fraud, when all parties are obviously motivated to hide what they've done? And second, it is rational of the voting population to be concerned that even though by some miracle it cancelled this time, and by some miracle they've been satisfied about that fact, it doesn't mean that will happen next time. If fraud is permitted, it doesn't take much playing with the game theory to discover that fraud is almost inevitably cheaper and more effective than anything else, and over time, the incentives inevitably align to encourage fraud to completely crowd out real voting.
And the recent chatter that corroding confidence in the voting system is bad is correct. Where that chatter goes wrong is in somehow thinking that is a primary goal, when in fact it must be built on the foundation of having a reasonably fair voting system in the first place. Having a fraudulent voting system but convincing the population it is valid is taking out debt with an catastrophic and random interest rate; it works until it without warning explosively doesn't work.
I don't agree with Trump's politics in the slightest, but if there's anything that can be gleaned from the Clinton leaks, it's that her campaign and the DNC at large is willing to play fast and loose to get her elected. There is undeniable documentation that they attempted to subvert the primary election (and succeeded), and it stands to reason that they may again attempt something else in the general election. While that doesn't mean that they'll hack into voting machines, it would be prudent to scrutinize the results.
The fact both can do it says nothing about whether or not their efforts to commit voter fraud would be equal. It makes no sense to assume number of fraudulent votes from either side would offset.
You're probably committing the Gaussian fallacy, which is to assume that all events are always drawn from a Gaussian distribution. Drawing a series of events from a Gaussian distribution centered around zero will tend to at least stick close to zero, sure, but there's no reason to assume that fraud distribution will be Gaussian. Especially in the face of hostile and intelligent entities who know you're making that assumption and will abuse you with it. Very little math or math intuition is structured for dealing with intelligent hostile exploitation. (Game theory, of course. And a lot of math involves assuming the worst of your counterexamples, but that doesn't really come from an intelligence with goals, even if we sometimes are tempted to think otherwise.)