I've been thinking about this "vs rolls" and the consequence on combat and design... I think the key is "does the attacker failling help the defender (beyond not being hit)".
Let us return to the 4 goblins vs the master swordsman. Now I think in general we can agree that it's a bad idea for 4 goblins to tangle with a master swordsman - they are going to die, it doesn't matter if it's 2e, 4e or in other systems like Troika! or D&D. The goblins are going to lose in this fight, period. That's not the problem.
In a system like D&D or warhammer 2e, the check is either unopposed (the swordsman has static defenses) or sort of (the swordsman can parry/dodge). But in either way there is no real "consequence" for failing to hit the swordsman. The goblins are no worse off for trying to attack the swordsman (in fact in 2e there are significant bonuses for ganging up on someone, so this might actually work!), besides now being in the reach of the swordsman' blade.
However, in 4e (or Troika!) attacking the swordsman helps the swordsman. By attacking the swordsman and failing, the goblins are either granting the swordsman advantage (4e) or allow him to hit the goblins (Troika!). In fact, if the goblins must engage the swordsman in melee, the best thing they can do is just show up and hang out, wave their daggers around but not actually attack! (even more so if they have a dodge action of some kind, but let's ignore that for now). It will take far longer for the swordsman to dispatch the goblins if they aren't assisting him by trying to stab him.
I think there are a lot of benefits to an opposed roll for combat - it certainly helps with WH's whiffiness for example. But as soon as there is a benefit to the defender if the attacker misses (advantage or a counter-attack), we risk creating a system where weaker foes attacking the PCs helps the PCs, which is really bizarre if you ask me.