Why We Use a Classical Baseline Next to QAOA
When you run the Resilience Lab, the page shows two results side by side: the best feasible sample the QAOA simulator found, and the exhaustive classical optimum for the same six-node problem. That second number is not decoration. It is the reason the lab can stay honest about what a six-qubit run actually demonstrates.
A toy problem has a known best answer
The lab asks a small planning question: which two backup links should be added to a six-node chain so that the most pair-connectivity is recovered across five single-link outage scenarios. Because there are only six candidate links and the budget is exactly two, the total number of ways to choose is 15. That is small enough to check every combination exhaustively and keep the best score.
An exhaustive classical solver finishes in a fraction of a millisecond on any modern browser. It returns the true optimum for this fixed instance. The QAOA circuit, by contrast, samples from a probability distribution over the 64 basis states. Some of those samples will be infeasible (not exactly two links). Among the feasible ones, the best observed score can be compared with the known optimum.
The baseline is not a competitor the quantum run is expected to beat
On a problem this small, the classical baseline will almost always match or exceed the best sample the simulator sees in a few hundred shots. That is expected. The point of showing both numbers is not to declare a winner. The point is to make the comparison visible so a reader can see, in the same run, how much of the optimum the sampled distribution actually reached and how many shots were infeasible.
Without the baseline, a high coverage score could look impressive even if it is still far from the best possible selection. With the baseline, the gap is explicit. The lab also keeps the classical result separate: it is never substituted into the quantum sample table.
What the comparison does not claim
Seeing the two numbers side by side does not mean the simulator is being used as a practical optimizer for real networks. Six decision qubits and fifteen feasible selections are a teaching and verification size. The classical method scales poorly if the number of candidates grows, but that is not a result this page demonstrates. The page also does not claim that QAOA will eventually outperform classical methods on this or any larger problem. It only shows what one noiseless simulation produced on one fixed instance, next to the known best answer for that instance.
How to use the comparison when you run the lab
After a run, look at three things: the classical optimum (coverage and which two links), the best feasible QAOA sample, and the fraction of shots that were feasible. If the best sample matches the optimum, the circuit and sampling found a best solution in the shots you requested. If it does not, the table still shows the energy and coverage of every observed bitstring, including the infeasible ones. The downloadable JSON keeps both results and the seed so the same comparison can be reproduced.
The classical baseline is the simplest way to keep the Resilience Lab from overstating what a small QAOA simulation has shown.