How well would a monkey with a typewriter lead?

I posed this question over two years ago at https://bridgewinners.com/article/view/bridge-leads-for-typewriting-monkeys

I have 15 million+ deals from ACBL BBO events in 2020 and was curious how my monkeys would do compared to real life ACBL players.

I live in Georgia, home of 159 counties, so a few days ago, along with my good friend Dorothy and her dog and wearing my new shiny silver shoes and golden cap, I drove up north to Gillikin County to rent some monkeys again. The smartest monkeys in the US are the Winged Monkeys, first described by Frank Baum (no relation for former ACBL CEO Jay Baum) in 1900 and most of them are in Gillikin County.

Unfortunately, the prices were a little high, and I was told to head west to Winkie County where the old and blind monkeys were sent to retire. Affectionally known as "Winkies", Nikko let me rent some of his Blind Winkies (BW) to see how well the experiment would go.

I set up all 15,357,763 deals and, using exactly the same contracts as played by ACBL players, asked the BW to randomly pick a card for each contract. Not as laborious as you might first think because several deals were in similar contracts with the same person on lead. Nevertheless, I instructed the monkeys to pick a random card every time.

3,798,358 (24.73%) of these deals were immaterial according to double dummy, so in reality I only had 11,559,405 unique boards.

The ACBL players collectively had managed a DDOLAR of 79.16% and ADDOLAR of 72.31%.

If you are not familiar with DDOLAR/ADDOLAR see the terminology page.

How well do you think my blind winkies would do? The BWs would be picking a card at random to make the opening lead (OL). Pick a DDOLAR and ADDOLAR value before you see the results.

How close were you to guessing these values?


Why, if I had a brain I could have realized that I could ask the monkeys to select every possible card for each contract and record the results.

When the BW selected every possible card, they had a DDOLAR of 65.878% and ADDOLAR of 54.666%. There was a difference at the third decimal place compared to the random picking of a card. This was probably the result you expected. The poor monkeys have now picked over 100,000,000 cards.

But, we know now how well a typical BW will do.

Is it possible to train the monkeys any better?


Well, I could no longer use the Blind Winkies so had to upgrade my monkeys to the seeing kind. My first attempt was BANNER leads: Black Always Nine, Never Eight Red. If you have a black nine, lead it, otherwise pick any other card as long as it is not a red eight. I have had BANNER leads on my convention card for some time and they work well in a single dummy setting. I wrote up and explained the rational behind BANNER leads in the URL above.

Do you think BANNER leads will be an improvement to what the BW could do?


Winged Monkeys come in two colors: silver and gray, the gray being generally regarded as the smarter of the two.

Accordingly, I first hired a group of Silver Winged Monkeys (SWM) and gave them the simple instruction set:

1. If it is a suit contract, pick any card.
2. If it is a NT contract, and you have an AK, then lead the A. If you have multiple AK combinations, use the lowest (clubs < diamonds < hearts < spades)
3. Lead fourth best from the longest suit. If multiple suits of the same length, lead from the lowest.
Do you think there is an improvement? If so, what do you predict the new DDOLAR and ADDOLAR will be?


The Grey Winged Monkeys (GWM), in general being more intelligent than the SWM, were given an additional step and to lead from AK against suit contracts as well.

1. If you have an AK, then lead the A. If you have multiple AK combinations, use the lowest (clubs < diamonds < hearts < spades)
2. If it is a suit contract, pick a random card.
3. If NT, lead fourth best from the longest suit.

Do you think a GWM is an improvement over a SWM?


My, monkeys come and go so quickly here. I was starting to run out of monkeys. The most common monkey species found in animal research are the grivet, the rhesus macaque, and the crab-eating macaque. Unfortunately, they were all busy with other customers. Nikko suggested that I try some monkeys that looked like Bridge players. From the Family Pitheciidae of the New World Monkeys, he suggested I try the titis, sakis and uakaris. Titis, sakis and uakarie, oh my! The uakaris certainly looked like a typical Bridge player:

Courtesy: Wikipedia.

I started with the titis and added some extra rules.

1. If you have an AK, then lead the A. If you have multiple AK combinations, use the lowest (clubs < diamonds < hearts < spades)
2. If you have an KQJ, then lead the K. If you have multiple KQJ combinations, use the lowest (clubs < diamonds < hearts < spades)
3. If it is a suit contract, pick a random non-trump card.
4. If NT, lead fourth best from the longest suit.

What do you think the improvement, if any, was?


Next, I tried the sakis but with more rules.

1. If you have an AK, then lead the A. If you have multiple AK combinations, use the lowest (clubs < diamonds < hearts < spades)
2. If you have an KQJ, then lead the K. If you have multiple KQJ combinations, use the lowest (clubs < diamonds < hearts < spades)
3. If you have an QJ10, then lead the Q. If you have multiple QJ10 combinations, use the lowest (clubs < diamonds < hearts < spades)
4. If it is a suit contract, pick a random non-trump card.
5. If NT, lead fourth best from the longest suit.

What do you think the improvement, if any, was?


For the uakiris, I gave them an extra rule.

1. If you have an AK, then lead the A. If you have multiple AK combinations, use the lowest (clubs < diamonds < hearts < spades)
2. If you have an KQJ, then lead the K. If you have multiple KQJ combinations, use the lowest (clubs < diamonds < hearts < spades)
3. If you have an QJ10, then lead the Q. If you have multiple QJ10 combinations, use the lowest (clubs < diamonds < hearts < spades)
4. If NT, lead fourth best from the longest suit.
5. If it is a suit contract, lead a non-trump Ace if you can.
6. Pick a random non-trump card.


The titis wanted to have another go, so I added more rules.

1. If you have an AK, then lead the A. If you have multiple AK combinations, use the lowest (clubs < diamonds < hearts < spades)
2. If you have an KQJ, then lead the K. If you have multiple KQJ combinations, use the lowest (clubs < diamonds < hearts < spades)
3. If you have an QJ10, then lead the Q. If you have multiple QJ10 combinations, use the lowest (clubs < diamonds < hearts < spades)
4. If a suit contract, lead a singleton if you have one.
5. If NT, lead fourth best from the longest suit.
6. If it is a suit contract, lead a non-trump Ace if you can.
7. Pick a random non-trump card.


My monkeys were tired. Nikko suggested I try using a Bird for the next test. Accordingly I gave my avian tester one extra step.

1. If you have an AK, then lead the A. If you have multiple AK combinations, use the lowest (clubs < diamonds < hearts < spades)
2. If you have an KQJ, then lead the K. If you have multiple KQJ combinations, use the lowest (clubs < diamonds < hearts < spades)
3. If you have an QJ10, then lead the Q. If you have multiple QJ10 combinations, use the lowest (clubs < diamonds < hearts < spades)
4. If a suit contract, lead a singleton if you have one.
5. If NT, lead fourth best from the longest suit.
6. If it is a suit contract, lead a non-trup suit from Xx where X<=10 if possible
7. If it is a suit contract, lead a non-trump Ace if you can.
8. Pick a random non-trump card.
Do you think there will be a change? If so, how much?

I was running out of birds and monkeys. Nikko told me that the wild Gibraltar monkeys, better known as "Gib Monkeys", were going to be the best choice. But they are a protected species, they are not in the US and I could not use them.

I was just about to give up, when Dorothy interjected, and said, "No one know what's going to happen next." She was right. In the caliginous corner was a cantakerous chimp. I enquired and Nikko said, that is the Smartest Winkie (SW) that I have. A baby has brains, but it doesn’t know much. Experience is the only thing that brings knowledge, and the longer you are on earth the more experience you are sure to get. And that monkey has the most experience of them all. Why, SW even sends me a postcard with the word Bermuda in it, so it must be a very smart and experienced monkey. Heeding Nikkos advice, I hired SW.

SW could provide one last improvement for the test. SW did not understand transfers, or doubling for the lead, but SW would lead a suit if partner had bid it! SW was the cleverest monkey of them all. These were the rules:

1. If you have an AK, then lead the A. If you have multiple AK combinations, use the lowest (clubs < diamonds < hearts < spades)
2. If you have an KQJ, then lead the K. If you have multiple KQJ combinations, use the lowest (clubs < diamonds < hearts < spades)
3. If you have an QJ10, then lead the Q. If you have multiple QJ10 combinations, use the lowest (clubs < diamonds < hearts < spades)
4. If partner bid a suit, lead it if you can. If you have the Ace of the suit lead that, otherwise lead low.
5. If a suit contract, lead a singleton if you have one.
6. If NT, lead fourth best from the longest suit.
7. If it is a suit contract, lead a non-trump Ace if you can.
8. If it is a suit contract, lead a non-trup suit from Xx where X<=10 if possible
9. Pick a random non-trump card.

How well do you think SW did?


I am now back home with pretty Dorothy and her dog. Dorothy is still adjusting to the fact that she is no longer in the mid west and instead living in the wonderful world of GA.

Summary, showing the additions to the monkey's arsenal:

Description DDOLAR ADDOLAR
Random blind winkie (monkey) 65.876% 54.663%
BANNER leads 65.850% 54.629%
Silver Winged Monkey - 4th best against NT 68.344% 57.942%
Grey Winged Monkey - Lead AK 71.232% 61.779%
titis - Lead KQJ 71.779% 62.506%
saki - Lead QJ10 72.050% 62.866%
titis again - Lead Ace 74.117% 65.612%
uakiris - Lead singletons 75.160% 66.997%
Bird - Lead xx 76.168% 68.242%
SW - Lead partner's bid suit 77.276% 70.207%
ACBL players 79.160% 72.311%


With very little training, my monkey is under 2% away from the average for an ACBL player, and only 3.7% away from a world class expert. Out of all ACBL players with 1000+ boards played, my monkey ranks 11,008 out of 13,564.