So, here's an example.
You sit down with partner and agree to play that a 2NT response to a 1NT opening is a relay to 3C, possibly weak with clubs but possibly a 4441 with slam aspirations. So, you hear a 1NT opening and bid 2NT, ready to go, expecting partner to bid the 3C relay, when suddenly partner bids 3H.
What kind of nonsense is this?
Actually, partner has done that which he should do. Rotely bidding 3C on every hand seems wrong. Sure -- 3C is the "agreed" call, but then there must be room for thought.
So, think. Partner is showing something in a way that seems safe but that is constructive, with a hand that cannot afford your one end-the-auction action of passing 3C. So, he must he clubs.
Give yourself an average bust-type club hand. Maybe 6322 with nothing but honor-sixth in clubs.
What would make partner move opposite that hand?
How about three side Aces and the rest internal honors? Maybe KQx in clubs, with three side Aces? Opposite that hand, you don't need much for 3NT to be right. Consider two possible hands:
Nine tricks. Easy.
OK, so the first step is that partner could easily have a super-accept of clubs. But, why 3H?
Well, if he had the perfecto, he might have opted 3NT. So, perhaps he has a problem? Maybe:
Still 9 tricks, as long as hearts are under control. But, why would 3H send that message? Couldn't you simply bid where you live instead? In other words, bid stoppers up-the-line?
Perhaps. But, there is something else going on here. Partner also needs to cater to your possibility of being 4441, right? So, perhaps with any perfecto he should bid a focal suit for that contingency.
For example, consider a slightly different Axx-Axx-Axxx-KQx hand. 3D could operate as a call that super-accepts clubs but identifies diamonds as a suggested strain opposite the 4441 hand. 3NT, then, would "flag" or identify clubs as a strain suggestion in the event of a 4441 hand (typically 3-3-3-4 or 5-card clubs).
So, let's just assume that 3D flagged diamonds for the 4441 consideration while also super-accepting clubs. How would you proceed under that assumption?
With no interest in 3NT, you should be able to rescue to 4C, but otherwise bidding 3NT to accept the game try (honor sixth) works. But, you would also bid 3NT with a 4441 hand with club shortness (because Opener should have two top clubs, one would think). Does it matter that partner does not know which? Of course not.
What if you had that 4441 hand and diamonds is a good strain? No problem -- bid the shortness, if it is a major. so, 3H or 3S would logically show that 4441 hand with shortness in the bid major, right?
What about a 4441 hand with short diamonds? Again, you may want to bid 3NT. But, you might be missing a 4-4 major fit. Wouldn't partner think of that problem also? If he has THAT hand, he would typically bid a 4-card major before diamonds, if he had 4-4 in the major and diamonds plus clubs super-accepted. So, 3D would seem to eliminate a major out, making 3H and 3S continuations purely shortness calls.
But, partner bid 3H. Could he not, then, have four hearts and diamonds? Or, 4H/4C? Or even 4H/4S?
This seems to call for some slightly different situational handling. With the 4441 hand with short hearts, you would bid 3S and let partner unwind this appropriately, as all will work well. With hearts, you can bid the shortness or bid 4H with short spades! Problem solved, when 3H is the call. But, you cannot bid 4C to show club shortness!!! That shows an escape. No problem -- you now kick into your meta-agreement. If up-the-line, 4D for short clubs, 4H short diamonds, and 4S short spades. If "natural," reverse 4D and 4H.
What about partner bidding 3S, then? Well, you can eliminate having both majors, as 3H would have stood out. So, if he has two suits, they are spades and a minor. Maximizing space for the unwind, you could bid 4D with the "short spades" 4441 (if too good for 3NT), because 4C is again that escape. With a spade fit, then, the shortness calls start at 4H.
Now, all of this sounds like quite a bit to make up at the table. And, it is. And, perhaps you would assume or reach different conclusions from me. Ideally, these situations are discussed, but that is not realistic.
In my opinion, however, it is tough to draw a line as to when the assumptions to make during an undiscussed sequence are "too much." I would expect that a well-tuned partnership might know well how to anticipate assumptions in sequences like this and to get fairly deep into the analysis.
This is, of course, quite dangerous, but then NOT taking advantage of this type of thinking during the bidding is perhaps just as much of a loss, for you end up simply bidding 3C every time and missing opportunities.
An idea might be to run a few odd examples through the partnership. For example, take an auction like this and run with it. Have each partner write down their thoughts independently, and then compare.
A good starter example might be Kokish. You might write down this question:
"Opener bid 2H as the start of a Kokish relay, but Responder whips out 3C. What should 3C show, and how should Opener react to this?" If you have never discussed this, which would be great, see how you and partner respond independently and then discuss this. Allow no "that call cannot exist" lame escapes. This is a theory test, not a bidding test. Play along, and then see where you end up. You might discover a lot about partner's way of thinking, and about your own.