If you order McDonalds, you probably know what McDonalds should taste like. If it meets 100% of your expectations, why would you rate it lower?
Similarly, if I get a meal from Michelin Star rated restaurant and it fails to meet my high expectations, should it get a better rating than McDonald's just because the restaurants meal was still likely much better than McDonald's even if it failed to meet expectations?
In practical terms it dilutes the value of 4 and 5 star reviews for other restaurants. There's some cognitive dissonance involved when McDonald's gets a 4.6/5 and a mom and pop haute-cuisine restaurant gets a 4.6/5 as well. These are two very different dining experiences but they are being judged using the same system and feels inaccurate.
For what McDonalds is they probably deliver a 5 star review the way their customers see it. You probably have to divide restaurants into different categories and compare within them. I beat a lot of people will rate MCDonalds higher than haute cuisine restaurants even in direct comparison. There is no absolute rating for restaurants.
There is no sensible way to compare fine dining to McDonald's.
The rating should be based on how well they deliver what they purport to deliver. If McDonald's delivers the absolute epitome of a fast-food burger and fries then it should get a 5/5, even if tastier food can be found at other restaurants.
Haute cuisine with a mediocre execution should get a 3/5 even if it's tastier than a McDonald's hamburger.
From a utility standpoint, I mean, yeah. You should definitely rate McDonalds lower than a restaurant with food you liked better, if people want to search for "how was the food" divided by "cost of the food" that's obviously a sliding scale that depends on the person. Maybe it should be something you can sort by, but I don't see how you could possibly reconcile ratings from different people if they all give five stars for any restaurant that meets their expectations.
Experience vs. expectations is the only reasonable way to interpret user-based scores even if they are not astroturfed. The problem is that you don't always know what the average user's expectations are and how they compare to your own and there is no basis for comparing scores between different objects.
Similarly, if I get a meal from Michelin Star rated restaurant and it fails to meet my high expectations, should it get a better rating than McDonald's just because the restaurants meal was still likely much better than McDonald's even if it failed to meet expectations?