The Peres-Mermin Magic Squares game at https://en.wikipedia.org/wiki/Quantum_pseudo-telepathy defines a simple circuit that outputs legal winning game outcomes. An "ad-hoc" circuit that does this is
OPENQASM 2.0;
include "qelib1.inc";
qreg q[6];
creg c_alice[3];
creg c_bob[3];
h q[0];
cx q[0],q[2];
x q[2];
measure q[1]-> c_alice[1];
measure q[2]-> c_alice[2];
measure q[0]-> c_alice[0];
cx q[0],q[3];
h q[4];
cx q[4], q[5];
cx q[3], q[5];
measure q[3]-> c_bob[0];
measure q[4]-> c_bob[1];
measure q[5]-> c_bob[2];
Runing this six-bit circuit on a NISQ device results in accuracy of 70-85%. However, there are machines with, for example, 27 qubits. How can this additional capacity be used to get more accuracy? How can you combine qubits - it seems there is no way to "average" them.
