So it's not just about fairness; it's also about ensuring the algorithm works in the first place.
Notice it explicitly says:
> This would give them better chances. At the expense of students that don’t (or can’t) trade.
But in fact those students are unaffected.
In subsequent years, as you say, it is a problem. But I don't see how to read that part as anything other than denying that when just restricted to that year it is a Pareto-improvement.
You may say "well that's a self-evidently stupid claim; of course they're not making that claim, because that would be stupid". But there's a lot of people out there who really seem to believe in a zero-sum world, and are convinced that when anyone's position improves, that means that everyone else has been harmed. That Pareto improvements are fundamentally impossible. There are absolutely people out there who would make that claim.
I agree that the wording of 'at the expense of' is a bit misleading, but I don't think it is categorically incorrect. The idea of fairness can, by certain definitions, even encompass pareto-superiour situation, if the improvements are not distributed fairly.
I gave this example in another comment:
Lets say I have a cookie, and there are 10 kids in front of me who like cookies. If I give any kid a cookie for any reason, it doesn't HARM the kids who don't get a cookie; they are no better or worse off than if the cookie never existed. However, suppose I decide to give the cookie to the tallest kid - even though none of the other kids are harmed, this is still not a 'fair' situation.
You can do no harm and still be unfair. Whether that is bad or not is subjective, but it is still unfair (in the sense that every kid has an equal chance at the cookie).
Of course, whether being absolutely fair is worth foregoing pareto-superiour improvements is a moral argument, and there is no objectivly correct answer to that.
The rest of your comment just returns to the claim that the whole thing is unfair even though nobody is harmed. But that's not the part I'm claiming is wrong! The part I'm pointing out as a "straight-up error" is the part where it denies that (in the restricted context of a single year where people didn't know that trading would be possible) is indeed a Pareto-improvement. I'm not making any argument as to the broader fairness.
I'm not sure we really disagree on much here.
Your assertion that it is an 'error' is only true if they only allow the trading to happen this year (after the preferences were set), and not in future years. The trades are only pareto-superiour if there are no future decisions to be made; since this is a repeated game, allowing pareto-superior trades this year means that next years game will NOT be fair (by their definition of fairness).
I agree that their definition of 'fairness' is not objective, but it is one they have decided to go with; they decided that the ability to strategize is not evenly distributed, and so allowing strategy is unfair.