Presumably so, yes, and it wouldn't be hard to imagine cases where the numbers shift around a lot, but the post I was replying to assigned a fixed 1-10 number to each site.
You're still making some very rigid assumptions about what kinds of things a scale can mean. I'd be more likely to judge restaurant ratings in terms of how often you'd like to eat at a given restaurant, not in terms of whether 50 meals at one restaurant is better in some sense than 12 meals at another restaurant. You can only ever eat one meal at a time.
Well, it's at least reasonable to ask whether you'd be willing to give up the possibility of ever again eating at the Cheesecake Factory in exchange for eating at Poppy once, even if it's a fairly speculative question—nobody is likely to face such a choice (unless the chef at Poppy wants to kill them, in which case other considerations likely outweigh the Cheesecake Factory). Answers "yes", "no", and "maybe" are all plausible, and they provide a clearer explanation of someone's preferences than just a 1-to-10 scale. You can easily imagine two people who both rated CF 3 and P 10 one of whom says "yes" and the other "no", but who just have very different relationships to food.
As for frequency, I think that might have more to do with variety and familiarity than quality. Surely, even if you agree that Tim Horton and McDonalds have equally terrible food and you hate it, you'd be more willing to be condemned to eat, each day at your choice that day of either of the two, than to be condemned to eat every day at just one of them, chosen today for all time. But that clearly doesn't mean that the union of Tim Horton and McDonalds has food that is less terrible than either of the two, which would be contradictory.
(You could of course have a scale of restaurant variety rather than quality, and in that case surely a PizzaHutTacoBell would rate higher than either a PizzaHut or a Taco Bell. But that's surely not the kind of rating we're talking about here—is it?)
> Surely, even if you agree that Tim Horton and McDonalds have equally terrible food and you hate it, you'd be more willing to be condemned to eat, each day at your choice that day of either of the two, than to be condemned to eat every day at just one of them, chosen today for all time. But that clearly doesn't mean that the union of Tim Horton and McDonalds has food that is less terrible than either of the two, which would be contradictory.
> (You could of course have a scale of restaurant variety rather than quality, and in that case surely a PizzaHutTacoBell would rate higher than either a PizzaHut or a Taco Bell. But that's surely not the kind of rating we're talking about here—is it?)
You seem to be making a fairly strong argument and then confidently drawing a conclusion that is the opposite of the one the argument suggests. The value of food is determined at the most fundamental, objective level by what else you've eaten recently. If you are short on something, your perception of taste will change so that the nutrient you are short on tastes better. What's the argument for variety not being a part of food quality?
If we accept that, then as I suggested in https://news.ycombinator.com/item?id=28613379, the whole project of rating restaurants on any kind of scale is doomed. Do you really think there's no sense in which the entire McDonalds menu is worse than a particular main course at Le Cinq? That the whole idea of Michelin stars is entirely nonsense? That food quality reduces entirely to variety?
I was attempting to separate out the "variety" aspect, which as you say cannot be evaluated in isolation and thus makes no sense to print in a guide, by proposing long time scales. Would you really argue that nothing is left once you factor out variety?
I think your model of how restaurant ratings should work has very little to do with how people decide what or where to eat. I'd rather think about asking someone to choose between two restaurants repeatedly (over time), and see how often they say one or the other. You can give the restaurants Elo ratings if you really want to. That's still not what's happening most of the time, but it's much closer. Or, as I suggested earlier, you could try to work with a satiety model and look at how long someone goes between visits to a particular restaurant.
> [Do you really think t]hat the whole idea of Michelin stars is entirely nonsense?
I think the perspective some other comments have taken, that there's more to food quality than whether you like it, is entirely nonsense. I don't think the concept of rating is nonsense; everyone likes some things more than other things. But I think your attempt to interpret restaurant ratings doesn't look very good.
> That food quality reduces entirely to variety?
No. But I'm happy to state that the union of Tim Horton's menu with McDonalds' is superior to either alone.
I'm saying it doesn't suit your purposes either. You are assuming that all meals at a given restaurant have the same effect. This assumption is untrue, and this means that any prediction you make based on your interpretation will also probably be untrue.
They just made the same incorrect inference that you did, that's all, but they did so more politely. In my response, which I linked for you earlier in this thread, I pointed out one way my original comment took those complexities into account in what you evidently think is my Grand Unified Theory of Restaurant Hedonics.
There's nothing about a 1-10 scale that implies that every visit to the restaurant will be equally pleasurable. Novelty is a factor. I would suggest that a sublime restaurant (a 10) will both make a really great first impression and also get tiresome more slowly (because restaurants like that tend to change their menu more dramatically, use a broader variety of ingredients, etc.). My utils from going to Cheesecake Factory would quickly approach zero after a few weeks of regular visits.
That's a good point. It might also depend on what the alternative is: the Cheesecake Factory might be much more appealing as an alternative to starvation than as an alternative to a diet consisting entirely of milk and potatoes.
You might have to stretch out the time scale a bit to reach stability, as I suggested in my initial comment: would you rather eat at the Cheesecake Factory once a month than Poppy once every two years, but rather eat at Poppy once a year than at the Cheesecake Factory once a month? (I never went to Poppy; maybe you didn't either.)
Maybe there's no time scale at which those numbers stabilize for you, though, so maybe whether you'd rate Tim Horton higher or lower than the Cheesecake Factory depends on how often you've suffered through their ham and cheese croissants in the last month or year. In that case maybe we have doomed the entire project of rating restaurants on a scalar scale, even a purely ordinal scale like the Mohs scale.
