A theory with mathematical beauty is more likely to be correct than an ugly one that fits some experimental data.