Melee combat abstractions: Time

Games necessarily use abstractions. Abstractions are neither good nor bad, although there can be good or bad implementations, and when choosing levels of abstraction for different elements, there can be good or bad choices.

Some things true of pre-modern armed combat which might not be obvious to everyone are:

1. When combatants are in reach/measure, strikes can be delivered very fast, and 
2. Combatants can spend a lot of time out of reach/measure, sizing each other up.

Many tabletop game designs get this backwards. There is almost never an incentive to stay just out of fighting range. And strikes in reach are either

1. Modelled individually, in which case the system almost always permits too few of them; or
2. Modelled abstractly/collectively, which usually works better.

But in both cases, game systems tend to neglect the necessary corollaries.

What do I mean by that?

Strikes modelled individually

Go and watch a video of HEMA sparring or historical fencing and count the attempted strikes. For a competent fighter using plausible weapons and trying to defend at the same time, you will see upward of two strikes per second. For a skilled fighter using a quick weapon like a one-handed sword, you will easily see four per second.

GURPS at least comes close to getting this right (depending on choice of rulesets). But its one-second-action time scale is still not fine-grained enough to properly model individual strikes!

Strikes modelled abstractly/collectively

A lot of games use one resolution mechanism (like an attack roll) to model a section of combat, summing up a series of attempted strikes.

Games like modern DnD and its various clones and pseudoclones claim to do this: one attack (or two or three, for skilled fighters) in six seconds represents a more complex section of combat. The claim doesn't hold up, though.

If an attack roll represents a series of attempted strikes, there are at least four important corollaries, which apply even to unskilled combatants:

1. It must be possible for on-hit effects like poison or sundering to be delivered multiple times per turn, even for an unskilled combatant;
   
2. It must be possible to (sometimes) hit multiple different foes in one turn, and increasingly likely for larger power disparities;
3. At a point in battle where a character finds themself in front of multiple helpless foes and no other threat, it must be possible to stab all of them - certainly at least six per turn;
   
4. Where a magical item or something similar delivers an effect on an object it hits, it should do so numerous times for every attack roll, to represent all the parries.
   

In general none of these are true for 5e and comparable rule sets.

(As an aside, the one attack roll representing numerous attempted hits clearly doesn't apply to ranged attacks, where one piece of ammunition is consumed per shot; there are other bigger problems with timing there, at least in modern DnD, where a windlass crossbow can somehow be shot and spanned in six seconds but throwing two throwing knives that are already in your hands takes twelve seconds.)

Is there a solution?

At least three that I can see. You can have an even more abstract set of rules so that none of these things even come up. You can change to a preposterously fine-grained time resolution. Or you can model multiple strikes with a single resolution mechanism but take pains to get all the corollary rules right.

I personally favour the last option. Abstract semi-tactical combat is fun. But it's really common in fantasy gaming to have a blade-breaking sword, or be faced with magically incapacitated foes, or be playing an extremely skilled fighter against a very large number of very weak foes. These kinds of cases should be supported!