I wrote this back when Realism was the topic of the day, but didn't publish it because it didn't seem to add anything new, but what the heck, work is eating my brain so have my 2 cents on realism:
I think when people talk about Realism what they are talking about is specifics (how many people live in this village?) that are actual-- match up with the best we know from historical records (in England of 1100 the village would have between x and y villagers). Actual might drift into plausible for space or hero games, but what we consider plausible is based on our actual experiences anyway.
Specifics communicate the imagined world we're sharing (You're on a cobblestone road. You can see a village with thatched roofs in the distance). Without enough details a player might only have a hazy sense of the world, might not even know what they can do or where they can go.
Genre helps a lot with this by acting as a shorthand for a shared set of assumed specifics. If I say we'll be playing a Western game like Boot Hill, it does a lot of work for me as a DM because we have a shared set of cultural tropes we can draw on. I don't need to tell you we'll be in Kansas in 1870 for you to have images of six- shooters, saddles, and player pianos.
Many of the tropes we associate with particular genres are inaccurate, or at least very vague in the sense of the time they happened historically. D&D's weapons and armor and equipment, for example, are pieces that appeared at different time periods in history.
Accuracy can become important when specifics break player expectations. If we all have the sense that our campaign is viking-style, maybe circa ~900, full suits of plate could seem jarring and weird. If you aren't playing a more gonzo-type of game, modern ideas and philosophies- like druids that are conservationists- might seem weirdly out of place.
Actual details can also be important just because it's so much work for DMs to invent specifics from scratch. If you want to create a new world you need to come up with the clothing, customs, languages, weapons, religions, and history. The list goes on and on. This is a hobby of its own, but takes an enormous amount of time and investment. Instead, I can read about medieval villages, how they're laid out, who lived in them, etc. If I read enough, I can internalize a sense of a "typical village" that makes it much easier to create a new village on the fly or field any questions players have while visiting a village.
With this sense of confidence that knowing about actual villages gives me as a DM, I can make villages that are larger than normal or odd in different ways. I can make it interesting. But more importantly, I can confidently provide details that help players "see" the place and make choices.
One problem is when specifics get in the way of actual gameplay. Examples of rules being more specific than needed are numerous: the idea that because weapons have length and weight that combat should incorporate reach and weapon speeds, or because weapons damage in different ways there should be damage types for blunt, piercing, cutting, etc. All this detail adds steps to combat resolution and thus time. The more specifics the longer and slower combat becomes. Specifics have a price when resolving rule subsystems.
I think the biggest problem with too much emphasis on actual details, being really concerned with historical accuracy, is that very few people are knowledgeable about actual details in a way that will help them see the game world more clearly or give them more choices as players. So, if you spend enormous amounts of time determining how a game set in Kansas 1870 will be different than Missouri 1880, it might help you as a DM with confidence and specifics you can use to describe the world, but most folks will only have those hazy Western tropes in mind. You can't expect players to know these specifics any more than they know the specifics of you fantasy land you entirely made up. We all aren't history majors. Heck, even history majors specialize in certain eras. If you set up the game to require them to know these facts, you might be setting up a lecture more than a game. At the least, players won't see choices that you might consider obvious "You should have asked for help, the way feudalism works means . . ."
I think the reason D&D has been largely a pseudo-medieval setting is because doing that manages a balance of these two: we have images in our head of what things are like, but we don't have to have PhDs in history to play the game.
Everything I do is trying to add the flavor of the real-- whether with trade, toxins and drugs, traveling the wilderness-- without bogging down the fun of the game. I think this is counter-intuitive. I imagine as long as there are rpgs someone will be trying to add weapon length to the combat and devising population density algorithms for hex crawls.
I'm sympathetic with folks that care intensely about actual specifics because I like learning and also because I completely understand someone looking to the actual to help them know a range of possibilities. I create my baselines-- what are the 20 fundamental potions?, or rings?-- for the same reason: to help me, as DM, make rulings and describe my world. And DMing is hard, so if someone spends tons of time reading up on vikings to help them run games, that's cool.
There comes a point though, where a person has to actually be obtuse to not observe how long things are taking in a game or what players are actually getting out of play. I am ultimately on the side of players here and I have little patience if after thirty years of this game being around you still get hung up on detailed encumbrance or price lists because how the game works "just doesn't feel right to you."
My advice: read to learn and enjoy, communicate clearly to players the genre and tone of your game, and include as much actual detail as you can without bogging down play.