What OSR games are you talking about? What indy games are you talking about?
To throw out some examples, Monty Cook's 'Numenera' rules seem to openly encourage Illusionism through the concept of DM Intrusions, and in the playthrough of FATE by one of the's systems co-creators on Wil Wheaton's youtube channel, the decision of the GM to decide that the cloaks the mooks were wearing were actually independent sentient monsters seemed to me as a viewer to be a clear case of Illusionism in that the existence of the monsters seems to have come about as a result of the party's success and the GM's decision that it's just more interesting if the PC's don't succeed easily having overcome the initial challenge.
I'm not ignoring what you wrote. I asked what is the tradeoff. Is flexibility what you meant? Is that all? Are there any other tradeoffs?
In terms of hard encoding the processes of play of a system? No, primarily I mean flexibility. Any other problem just is the ubiquitous problems you get with rules smithing, extended into the realm of processes of play (that normally aren't described and so normally don't have the issue). To give you a specific example of the later, I was going to give this as an example in another thread, but consider the Dungeon World SRD when it tries to codify the processes of play writes:
"Always be honest. If the rules tell you to give out information, like the Spout Lore and Discern Realities moves, do it. Don't lie or give half truths; be open and honest —generous, even."
Now, on the surface that sounds like good advice. GMs shouldn't be trying to screw over the players and they should reward success and even be "generous". But if we take this as a rule rather than a guideline and follow it rigidly, then you run into a problem I've already tangentially referenced in the thread. Maybe, if you closely read the thread and are familiar with the rules of Dungeon World you might have even noticed. There are times when there are truths about the situation that it isn't possible that the player character could know, and instead the player character as a member of his society would know as "truth" only what typical knowledgeable and learned members of his society believe. For example, I cited the case of the King having been usurped by his twin brother at some point in the past and having been turned into a toad. If you take the above agenda as a rigid rule and not a guideline, then a novice DM is like to say, "Well, the PC successfully Spouted Lore about the King, so I need to tell him that the King is in fact not actually Louis the heir but a twin brother whose existence is not even generally known outside of the royal family." I don't think that's actually the intention of the writer of the "rule", but merely he hadn't considered the sort of edge cases that are going to come up (which is a common problem with a lot of tersely written rules) where the useful facts and known information everyone believes is actually wrong. If Prince Humperdinck plans to kill his bride before the wedding night, the useful information that Prince Humperdinck is a master tracker who can track a falcon on a cloudy day, a superb fencer, and is planning to wed Buttercup in three days' time is false, but it isn't wrong to give IMO. Later investigations might uncover the truth behind what everyone knows to be true, but it's not wrong to hide some of that truth until the player actually has some way of uncovering it.
Your OP, as has been pointed out, takes the standard definition of Illusionism and then redefines it more broadly, to include "players believing they are playing one game but actually playing another, and also GMs thinking they are running one game, but actually running another."
That's redefining the term.
I disagree that I have redefined the term. See the prior discussion a bit above this and an example of why I think my broader definition is the more accurate one. However, for the purposes of the sort of statements you are making, I don't need my broader definition and can just happily use the "standard" definition of the term.