If you've got a wonderful stable relationship, but you see a way to cheat without any risk of every being caught, or any other overt negative consequences, and you sincerely want to do so, should you?

My contention is (unsurprisingly) that you should not – but more interesting is that I think this is fundamentally an instance of a Newcomb-like problem in decision theory. See my earlier post about corporate regulatory environments also being Newcomb-like for an example of what I mean, but briefly:

Newcomb's problem is about what you ought to do if offered two prizes: a transparent box with $1000 in it, and an opaque box with $1,000,000 in it if and only if the present-giver predicts you'll only open the opaque box and leave the transparent box completely alone. For good enough predictors, it highlights a confusion some people have – if the money is already in the boxes, it's already decided, no matter what you do, so why not take both? You get whatever they previously put in the opaque box plus a guaranteed $1000. But, of course, if you open both boxes, it stands to reason that the good predictor knows this, and all you ever get is $1000, which is a far cry from the million you could have easily gotten.

Relationships are the same way. I don't think it's possible, or reasonable, to really check in on partners enough to know if they're faithful. But of course, if you didn't trust them, you wouldn't commit to them, and they wouldn't commit to you. You're looking for someone you can have a $1,000,000 [metaphorically valued] relationship. And there might well be many opportunities for $1000 indulgences. And if you know you couldn't get caught (because the Newcomb prompt asks us to be immune from consequences as well), it matches the setup quite exactly.

Obviously, we should avoid regretting other people correctly predicting our actions. It should be completely obvious that we get a lot of value from being the type of person we'd want others to trust. Don't open both boxes is, ultimately, the only sensible strategy.