Blackwood and Roman Key Card Blackwood

Four no trumps asks a question. The hard part is knowing when not to ask it.

Get it for iPhone & iPador play in your browserboth free

Regular Blackwood: Counting Aces#

What defeats a slam is almost never a shortage of points. It is two aces missing, cashed off the top before declarer has played a card that matters. Blackwood exists to check for that, and for nothing else.

4NT asks how many aces partner holds, and the replies run up the ladder:

The 0-or-4 ambiguity in the first step is not the problem it looks like. A hand that has heard partner ask for aces and holds all four will not be mistaken for a hand holding none by anyone paying attention to the auction so far.

Regular Blackwood is the ACOL default in Rase and can be selected in either system from the Blackwood row in System & Conventions.

Roman Key Card: Five Cards, Not Four#

Roman Key Card Blackwood counts five key cards — the four aces and the king of trumps — and squeezes in a message about the trump queen as well. It is a straightforward improvement, because at the twelve-trick level the trump king is worth precisely as much as an ace, and a missing trump queen is what turns a laydown slam into a guess.

Two response schemes are offered. They differ only in which of the first two steps carries which answer:

Rase defaults to 1430 in 2/1 and regular Blackwood in ACOL. Whichever is set, the robots use it consistently on both sides of the table, and tapping the 4NT bid during the auction shows you the scheme currently in force.

One requirement worth stating plainly: key-card Blackwood needs an agreed trump suit. "The king of trumps" is not a card that exists until the partnership has settled on a strain. Without agreement there is nothing for the fifth key card to refer to, and 4NT is not a key-card ask at all.

Reading the Answer#

The ask is the easy half. What you do with the reply is the half that decides whether the convention made you money, and it reduces to arithmetic:

• Two or more key cards missing: stop. Sign off in the cheapest playable contract below slam. Two cashing losers is not a slam, however many points are facing you.
• One missing: bid the small slam. Twelve tricks with one loser off the top is exactly the hand Blackwood was invented to find.
• None missing: bid the small slam, and consider the grand if the tricks are there to count.

Rase's robots follow that ladder literally, which means asking is safe: the engine will sign off below slam on a bad reply rather than blast on. If you have watched a robot bid 4NT and then subside in five of a major, that is what it was doing — the ask was worth making precisely because the answer was allowed to stop it.

The Hands That Should Not Ask#

Blackwood is over-used, and the auctions it wrecks look the same every time. Four situations where the answer cannot help you:

• You hold a void. The count of aces is now the wrong question. An ace opposite a void is not a trick you are missing, and a reply that says "one" tells you nothing about which — which is the only thing you needed.
• Five of your suit is not safe. Asking commits the partnership to the five level. If you cannot stand to play there, you cannot afford the question, whatever the answer turns out to be.
• Your problem is a suit, not an ace. Blackwood counts first-round controls across the whole hand. It says nothing about whether their long suit is stopped or whether you have a second-round loser somewhere specific, and asking a question whose answer you cannot use costs a round of bidding you might have spent on a cue-bid.
• No trump suit has been agreed. Over a natural no-trump sequence, 4NT is far more useful as a quantitative invitation — "bid six if you are at the top of your range" — than as an ace ask. In ACOL, Rase uses Gerber (4♣) for ace-asking directly over a no-trump bid for exactly this reason, leaving 4NT free to mean what it should.

Slam Economics in Rubber Bridge#

Duplicate players bid slams on a different calculus from rubber bridge players, and it is worth resetting before you sit down at a rubber table.

The case for being bolder: a vulnerable small slam pays a 750 bonus on top of the trick score, which is a large fraction of an entire rubber. When a slam is likely and you are vulnerable, bid it — the premium is doing more work here than the equivalent matchpoint swing does in duplicate.

The case against: a game made is a game towards the rubber, and the rubber is what you are actually playing for. Four spades making beats six spades down every single time, with no field to compare against and no consolation for having been unlucky. If the exploration itself might push you past a contract you were certain of, the exploration is the risk.

Put together: ask when the slam is genuinely in prospect and the five level is safe, and do not go looking on hands where a comfortable game is already in the bag. That is a narrower window than most players use, and it is close to how Rase's robots behave — they will ask for key cards on a nine-card fit with slam values, and sign off the moment the reply says two are missing.