Let me know if someone's actually done this.
Experiment: Use undergraduate schoolstudents as test subjects. Give each subject a shuffled deck of playing cards and ask them to sort it by suit and rank as quickly as possible. Time how long each subject takes to complete the task.
Prediction: A minority of computer science students will markedly outperform everyone else.
I think many people will hit on the ~optimal solution of laying the cards out in a 4x13 grid on the table, even non-CS students. I bet you'd still see spikes for different algorithms, though.