A USP Example: Kickback
The Blackwood convention, as originally formulated, violates the USP. Suppose that the agreed trump suit is spades. After the Blackwood "asker" bids 4NT, "teller" can convey four separate messages without bypassing the safety level of 5♠ – four aces or none with 5♣, one ace with 5♦, two aces with 5♥ and three aces with 5♠.
But what if the agreed trump suit is clubs? Suppose that asker and teller each have one ace. Then, after 4NT, teller bids 5♦ to show his ace, and the partnership has to play 6♣ off two aces (or possibly 5NT, which could be worse than 6♣, if it has the machinery).
The problem can also occur when the agreed trump suit is diamonds, although it is less likely because there is more space available for responses than when the agreed trump suit is clubs. But if the partnership is using Roman Key-Card Blackwood there can be similar problems. Suppose that hearts is agreed, asker has one ace and teller has one ace plus the king and queen of hearts. Asker bids 4NT and teller bids 5♠ to show two key cards plus the trump queen, and the partnership is again too high.
The problem is that Blackwood ignores the USP. The lower in rank the agreed trump suit, the more space that is needed if the partnership is to stay at or below a safety level.
The Kickback ace-asking convention deals with the problem by adjusting the asking bid according to which suit is agreed as trump. The ask is always one step above four of the trump suit. So, if clubs is agreed, the ask is 4♦; if diamonds is agreed, 4♥ asks; if hearts, 4♠; and if spades, 4NT.
The responses to the ask might be similar to Blackwood, but instead of associating a specific suit with a specific number of aces, the responses are in terms of the number of steps above the ask. If spades will be trump, 4NT is the ask, and then 5♣, one step, might show zero or four aces, according to partnership agreement. If diamonds will be trump, 4♥ is the ask, and then 4♠, one step above the ask, might show zero or four aces.
The effect is to allocate bidding space where it's most useful in the context of the convention. If clubs is agreed and each partner has one ace, asker bids 4♦ and teller bids 4♠ to show one ace. The partnership can now easily sign off in 5♣.
There is a cost, of course: the partnership that plays Kickback loses the ability to cue-bid the ace of the suit above trumps. That is, assuming that hearts will be trumps, asker can no longer bid 4♠ to show first round control of spades: that would be the Kickback asking bid.
The solution is to use 4NT to show a first round control in the Kickback asking suit. With diamonds agreed, 4♥ is the Kickback ask, and 4NT shows the ♥A or, if credible in the context of the prior bidding, a void.
The agreement that 4NT is a cue-bid still entails a cost, but Kickback users argue that there is a net gain. For example, with clubs agreed, South would bid 4NT to show a first round control in diamonds. This bid not only bypasses the Kickback ask (4♦), but also prevents North from cue-bidding 4♥ or 4♠. Kickback users believe that the gain in space from adjusting the ace-ask outweighs getting in the way of partner's cue-bid.
Notice that the Gerber convention, the use of 4♣ to ask for aces when NT is the likely final strain, is really a special case of Kickback.
Note also that the foregoing is meant only to illustrate the USP. It describes neither additional understandings that Kickback can accommodate, nor the special problems that can arise (for example, the question of which is the agreed trump suit).
Read more about this topic: Useful Space Principle