For background - this discussion stems from the bladesinger thread, where a high rolled stat bladesinger can have AC's in the low 20's pretty much from the start. Given the general attack bonuses for monsters of appropriate CRs, the net effect is that the monsters will only hit on high rolls of the d20. Given the extremely distorting effect of disadvantage on low probability rolls (it's inverse-square, not linear). This means the if you normally have a 40% chance to hit, with disadvantage you have a 16% chance to hit; if you have a 30% chance, with disad you have 9%; with 20% it's 4% chance to hit. If you only hit on a 20, you have a 0.25% chance to hit.
In that context, and given that the damage output of appropriate CR creatures always exeeds bladesinger hp by the DMG chart (unless the wizard has a +3 Con bonus, then it tracks), an optimum strategy for high AC bladesingers is to use a round to cast blur -- especially given that most encounters do not start in melee range.
As noted, the problem doesn't really lie with blur, which is awesome in certain cases, but has a utility cost for most casters well below that threshold. The problem really only obtains with high AC, Tier I and II bladesingers, who can internalize the combo. That requires excellent rolled stats, however, which is less common than assumed in the analysis. Still, the corrective action should be limited to the bladesinger ability suite rather than the spell.
A suggested fix was to nerf the concentration bonus from INT to proficiency, which reduces the impact of high stats, and to restrict the advantage to acrobatics to only those checks involving movement, which reduces the additional defense to non-AC attacks (shove, grapple). Those two smaller changes increase the chance of losing blur when struck AND remove the high chance of avoidance for shove and grapple, both of which are counters to disadvantged attacks against AC.