The topic of changing a game is always a divisive one at best. The standard arguments tend to boil down to 'this thing makes some choices so obvious that nobody will ever choose choice A and that is bad' vs. 'nothing ever needs to change because I played the game as it is and had fun.' My natural instinct is to balance the hell out of everything and constantly update and alter any game I am playing as I find things within it that are broken. This does lead me astray sometimes as I have an unfortunate tendency to alter the game prior to achieving mastery over it and sometimes I alter it incorrectly.
This becomes a much more serious question when rather than house rules we are talking about a big company putting out updates to a game used by many players. DnD is a great example of this and the DnD Next blog post this week was on the topic of how Wizards intends to handle corrections and updates to Next after it is published and the inevitable problems crop up. Their basic strategy isn't really a departure from the past since they intend to put out FAQs regularly but only make official balance changes once a year or so.
The trick of course is that many changes don't require changes to existing rules but instead are handled by adding on new things. For example in 4th edition the basic game had a terrible flaw in the scaling that resulted in players hitting less and less often as they levelled up. This was handled by putting in a bunch of extremely overpowered feats that gave scaling bonuses to hit that anyone who wanted to be decent had to take. It would have been much better to put in some kind of actual correction into the basic math but admitting that the basic rules got flubbed that badly and fixing it properly was not the route they went.
So I tend to think that the things are outlined in the blog post are pretty irrelevant. After all, they aren't going to wait a year to correct serious problems but will just solve them with kludges like outrageous feats instead of changing the core books. Because apparently the core books being massively incomplete is okay, but having them be wrong is an issue. It is really important to figure out what sorts of things really need correction and which balance issues are actually features. To my mind the core things that need to be satisfied are:
1. All of the options that people really expect to be reasonably balanced must be so.
It is fine to have a Hand Crossbows be total garbage compared to Longbows. People don't have a strong expectation that every weapon will be equal. However, if all caster classes are awesome and all thugs are terrible people will get bitter because they expect some kind of balance across classes.
2. There should be a variety of 'best' builds and choices.
People tend to be frustrated if the only way to be good is to use a longsword, or be a wizard, or cast Invisibility. They want to have a few different ways to play while feeling like they are effective and they want different people in their group to have different options that don't necessarily include being useless.
3. There should be mediocre and also terrible choices.
It isn't much fun if all of your choices are equivalent. Choices need to matter and putting together a good package of options is a big part of the fun. Being powerful is fun but it isn't much fun if you can do absolutely anything and be good. There must be a lot of stupid things you can do in order to make it interesting to strive to be good.
4. Choices should come from small lists.
There are people out there who enjoy perusing lists of hundreds of things to find the overpowered stuff but they are the minority by far. Most people find that sort of thing intimidating, tedious, or annoying. Choosing from small lists where the consequence of the choice are reasonably obvious has the greatest broad appeal.
So the tricky part is what you do when you discover something that really breaks these guidelines. I think the answer is that when you are really sure that there is a problem you fix it. Waiting a year, putting out additive kludges, or hoping that everyone just houserules it in the meantime is just taking forever to pull off the bandaid. Once you know there is a problem you deal with it right away and you don't try to pussyfoot around it. Step up and say what you are trying to fix and then do it.
Of course this strategy is coming from the mouth of a compulsive balancer so most likely the course that Wizards is choosing to take is actually a middle road solution that will please as many people as possible. They are after all in the business of making money rather than making beautiful games and that changes their priorities somewhat.