Terrence Chan website had a discussion about playing a hand.
47.1% of the people played the arrangement on the left, which ranks number one by nlsim calculator.
51.0% of the people played the arrangement on the right, which ranks number three by nlsim calculator.

Is nlsim or the majority of people correct?
Terrence Chan's forum
This iPad/iPhone tool is still under development and needs graphics, but it is good enough to use. My tool
has the ability to play two arrangements of the same hand at the same time. I played 100 times.

People's arrangement is tagged as "player" and nlsim arrangement is tagged as "nlsim".
The six numbers on the lower right corner are statistics for two arrangements.
I play 100 times. "player" gets 50 "correct" arrangements and nlsim gets 65 "correct" arrangements. The
term "correct" means it is the arrangement that wins most points or loses least points out of ALL possible
arrangements. Remember that two different arrangements might be both "correct".
Nlsim "mistake" after 100 deals costs it 81 points.
People (or player) "mistake" after 100 deals costs them 133 points.
Each player starts out with 500 points. After 100 deals, Nlsim has 474 points and "player" has 422 points.

Hey, NLsim plays better on this specific hand.
What is the deal here? Go back to the top two images, one for Nlsim and one for People's choice. And
remember, we play 1-3 here.

For Nlsim arrangement, the probability to lose 3 sets (top|mid|base) is 0% (almost). The probability to win 3
sets is 42%.

For people's arrangement, they are 0% and 26% respectively.

I think people's arrangement has a weak middle hand (a pair of J), which prevents it to win all 3 often.

Any way, it is not for me to decide. My calculator did. ^_^.