IMO If you have control over whether or not the risk manifests, than it doesn't manifest, which for all practical purposes means it doesn't exist.
Every electrician ever has control over whether or not their work is improperly done and thus could cause an electrical fire. Yet there is still a risk of fire from improperly done electrical work. If almost all houses almost always avoid electrical fires, does that mean that for all intents and purposes electrical fire risk is 0? Or is it still entirely there, we're just very good at keeping it contained?
Fukushima shows just how disastrous bad nuclear maintenance can be. Yet the vast majority of nuclear reactors do not have Fukushima-level disasters--not even close. Does that mean there is
no risk at all? Or does it mean that the risk is small and we can do stuff about it?
Regardless, as others have noted, any time I make a hard move I
could do something pretty nasty--and a hard move on Cast a Spell can happen even to an expert at spellcasting (rolling 2 or 3 on 2d6+3 still gives you 6-.)
But my point was about
badly-constructed spells, not casting existing spells badly. Waziri don't just randomly throw formulae together to see what sticks because things that
don't stick have a tendency to blow up in your face, literally or figuratively. That's why their magic takes so much
time, and why they're so ridiculously jealous over any discoveries they make (well, that and the whole "publish or perish" academia issue)--they have to be
absolutely sure that it works as intended, practicing with just teeny-tiny droplets of power. I trust the PCs to avoid accidentally blowing themselves up on any established spell (unless, as noted, they fail the Cast a Spell roll)--but for developing
brand-new spells, the Waziri method is dangerous but extremely effective when it works. Other traditions are much more limited in what they can achieve by learning; they have to work within their tradition's limits, or try to fuse together multiple traditions and struggle with the difficulties thereof.
I think
@EzekielRaiden 's point is it happens a lot less than 5% of the time, so having it be part of the standard spellcasting isn't reasonable. It's something the DM includes when they choose to.
Saying "that means there's no risk of it happening" is a bit like saying "because there's no dice roll the players can make that will result in a dragon attack, there's never any risk of dragon attacks and dragons effectively don't exist in the game."
Precisely. It's
more rare than the dice can represent in their chunky way. It can still happen. The party as seen the consequences of other people messing such things up. (It isn't pretty.)
Would you as a player ever consider it fair if, appropou of apparently nothing, a spell you cast blew up in your face?
In D&D? Probably not. It would be a
very significant strain on my relationship with that DM unless they had done
very significant prior effort to prove that they can be trusted with such a violation of the rules.
In DW? Absolutely...up to a point. That's one of the possible things a miss can do, so it's perfectly within the rules. All sorts of effects could happen ("Turn their move against them", "Show them a downside of their <class/race/etc.>", "Deal damage", "Split the party", etc.), whatever serves to push the action forward in an exciting way.
So, he created one for our party's Wild Magic Sorcerer. And some of the Wild Surge results from his brewed table were hilarious. 
