Erm, well considering those are the only two groups of people likely to own them and make them dangerous, probably the most common sense thing to do would be to ban them.
If you mean ban the humans, then yes... Since they're the deciding factor in an overwhelming majority of serious dog attack incidents, the sensible thing to do would indeed be to prohibit the unsuitable kind from owning dogs.
Since banning everyone from ownership would be highly prejudicial, you'd have to assess individual suitability through licencing, which for the majority of cases would essentially just mean making existing dog training classes mandatory. Other factors would easily be taken into account, such as whether the owner is sufficiently able-bodied, or has a prior record of dog abuse. Mechanisms even exist for things like ongoing inspection of living conditions, which would require only a modest increase in enforcement powers, all of which can be funded by licencing fees.
FML I think I need to say it again; this isn't monocausal, that there's influence from a dog's environment doesn't negate that there are behavioural traits associated with breeds.
And here's why the contradition you're bleating about is entirely of your own making...
Association does not equate to causation, which is the mistake you insist on perpetuating.
Association requires that at least two variables correlate, yet for all your witter against monocausality, you focus your assertions on just the one variable.
People assume ancestrally functional behaviours are synonymous with modern breed, but that is only vaguely accurate.
Your further exacerbate the situation with things like this:
Also, you seem quite muddled re: that last part, that these traits vary is literally the point! See the top part of this chart, see the entry for "Akita", see "dog aggression"... note how it is high:
Firstly, you're citing a study limited to pedigree examples as representative of breed, which as you know excludes up to 90% of the wider breed, so is hardly representative.
Secondly, you've seen that within-breen variances are of similar levels to between-breed variances when you include that other 90%.
Thirdly, you're failing to understand the pretty pictures you're waving about - This is about heritability, which you have yet to grasp...
The chart entry to which you refer is a breed-average scoring taken from a study that examines why certain behavours are shared across closed-breeding populations of seemingly unrelated breeds.
Your study admits it didn't even collect both phenotypic and genotypic data from the same subjects!
As for plotted trait variance on the left, this is heritability and dog aggression barely scores over 0.5 between breeds. This means that almost half of that variance is due to environmental factors.
The within-breed variance for that same trait is down to 0.2, meaning roughly 80% of the variance is due to environmental influence.
Thus you have the obvious conundrum you still can't address, you're happy to acknowledge differing behavioural traits when it suits you and ascribe some explanation but then simultaneously try to claim some blank slate position re: XL Bullies.
The conundrum is again of your own making.
Of course I acknowledge that behavioural traits differ, but I do not classify those differences by breed, which is where you are getting confused.
You're further complicating the issue by taking a system of closed-breeding analysis and applying it to a group of dogs that has been very widely bred.
Aside from the phenotypical elements, the most defining aspect of an XLB is how environmentally malleable it is - That's temperament and not a genetic behaviour, nor is it breed specific, even though it's what most influences the dogs' behaviour.
In short, you're assuming that all dogs are True Scotsmen, where most of them are not true and where the XLBs are the very antithesis.