The primary reason for a form of entertainment to be insufficient for someone, other than being bad, is when it presents a sphere of interests that does not harmonize with other things in this world that one may encounter. Failure to adjust to, or account for these other attitudes means loss of relevance. As mentioned before, other countries have significantly different priorities than the United States or other developed Western countries. Not accounting this means losing those audiences and anyone connected to them. The Grind is just one example of an unacceptable mechanism that may seem acceptable to a country used in leisure; when in reality most people in this world have economic progress or skills specialization to make in real life. Just to remind that the chance of death on an everyday basis is another significant reality and attitude difference; it is important not to forget these things when thinking of the world.
There must be many ways to increase framerate when many players must be rendered. One of them may be to transition to display of standardized textures for different styles and converging on certain hues when there are large numbers of players in one area; but I don't know if this would really be a computation issue, or just a memory one. However it has the additional benefit of reducing extraneous flashy characteristics and presenting an opposing army as more homogeneous and cohesive, especially if names are not displayed resulting in no 'large red cloud of names' effect so graphics must be simplified when in large numbers so someone can understand a situation faster without being confused or annoyed by distracting details. If polygons is the big issue, then it might be simple to also just have alternate low-polygon models that use the same or the simplified textures, and these models are swapped out when there are large numbers of people, with preference on swapping out those further from the camera.
It may also be possible to use filters, or even weather/lighting effects on the scene as a whole, with the player themself possibly staying unfiltered in this case and maybe nearby friends too, so only 'interesting' people take up normal computer resources. Mostly, large numbers of people nearby whether friendly or enemy should have a graphical effect to promote and reinforce the psychological effect of this environment. This is in line with treating numbers of people differently in other parts of the game too, such as aggro radius and 'sneaking past' something, and it also leads to another function as follows.
Having a conflict extend its effects past line of sight. At a distance, of course the distant sound of arms would be audible, if the game is tracking that large numbers of people are engaged in combat in a vicinity. Also a visual component would be nice (can't help thinking of whenever it must look like in PvP video) such as a flashing red border, or hopefully not something so lame/obvious. Weather effects may change as far out as this too, or anything to give a sense of tension and anticipation. This may manifest in a different way when the outcome of a battle has been decided; for example, someone 200m away may hear or receive some alert only when someone has died, and may not hear the preceding struggle. It might be very faint and only by investigating can they know what happened.
This 'beyond line of sight' idea would enhance the perceived significance especially of conflicts in an instanced area, which otherwise do not individually seem important.