For MMOs, fair enough. I can see the problem of the believability of the setting if everyone are running around as humans.

I thought we were mainly talking about smaller/local games like tabletop rpg in which the DM or settings creator are annoyed at players mostly preferring humans in their settings.

The EOS-LRP system in the UK was planned to be a 10-year project, and it died within 4 events because of this exact problem.

The designers created a game that was largely factionalized by race(1) - elves, orcs, goblins, humans, undead, and a few other player options. The idea was to build a dark, gritty "survival" setting where different factions would compete over limited resources, and the game story would be mostly driven by player-vs-player conflict. They kept prices for attendance low by running an extremely small crew.

This has been a successful strategy for many larp systems over the years, player-driven conflict is extremely valuable in keeping your players engaged, because NPC-driven conflict is expensive to run... if all of your game content is being delivered by your crew, you need a large crew in order to be able to keep the players interested and engaged, and this means high prices. If your game content largely stems from player-vs-player conflict, then you can potentially run a game with thousands of players using a crew of 20-50. I've been involved in several of these in the past.

So what happened with EOS? Well, the costume requirements for playing anything other than a human were extreme (this is a common requirement in larp systems that want a high quality immersive experience.) - we're talking full-head makeup, prosthetics, masks, etc etc. 80% of the players in the first event rocked up as humans, and because they were allied, they managed to wipe out the other factions completely. Some of those players went home, some of them rolled new characters, and got wiped out again, and went home.

By the third event, 100% of the playerbase were humans, and allied to each other. The game crew was six people, and they were unable to create any credible threat to the players. Because everyone was part of one monolithic faction, there was no conflict, and the players rapidly became bored, with nothing to do.

The designers tried to fix this by first banning players from rolling more human characters, and second introducing some overwhelmingly powerful hunter monsters to pick off isolated players. When characters died, the game admins told them they had to roll non-humans if they wanted to continue playing, and in response those players quit. Their friends followed quickly, and the game collapsed.

EOS had an interesting setting, with a lot of good design ideas, and some really cool handles for roleplay and conflict. They talked a big game, and promised an exciting, fast-paced, dangerous competitive game. Players were drawn to the events because of what the design brief promised, but in choosing to all play the same race, everything promising from the design brief was undermined, and the game died.

(1) There is, of course, a second, highly problematic issue with drawing your lines of conflict purely on "race" grounds, which is an uncomfortable issue all by itself. Modern fantasy gaming design is moving away from this, for reasons that I hope are well-understood.

This has literally killed games that failed to deliver the reality of their brief to their players. Promise one experience, deliver another, and people quit.

Could you elaborate on what you mean by that in relation to people generally preferring to play as humans/vanilla experiences?

Maybe people are downvoting your replies because this is a commonly discussed and well-studied issue in design circles, but you're failing to understand the problem and dismissing it as a "misguided" concern.

No, when I was writing my comments it was only the person I was talking to that was downvoting, votes are public so you could easily check. At that point, just tell me you don't want to discuss the topic and I'll stop replying.

Just because YOU don't think it's a problem doesn't mean it isn't a problem.

I'm arguing it's only a problem in the mind of the creator. For the ones it actually do matter for, the audience/customers, it's not an issue.