In case the criterion is the real world (leaking or not leaking?), the "bolt-up" ASME VIII theory is conservative, exactly as MoverZ says.

In case the criterion is the ASME VIII/ "Taylor-Forge" theory, I would say that the operating gasket load is nonconservative taken. In fact in the original Taylor-Forge bulletin 502 one can find that "the gasket load is 2b(PI)GmP where gasket factor m relates the required gasket stress at design pressure [...] For example an m factor of 3 means that the residual gasket stress at P must be at least 3P for the joint to be tight". True, however evaluating stress with a gasket load "at least" a given value, you risk to find a value of stress flange "at least" the result of calculation.

A paragraph below, Taylor Forge says that the gasket reaction load is generally assumed to decrease as internal pressure is applied. True again, but why we evaluate stress with a gasket load which increases as internal pressure is applied?

For these reasons, I avoid qualifying flanges which do not pass integrally ASME VIII calculation; that theory is weak enough (and in the same time it provides good quality flanges) and I don't need to improve it relaxing the steps of its calculation.

Best regards.