By the standards set forth in this paper, only one gravitationally lensed quasar qualifies as a golden lens. Taken by itself it gives a Hubble constant with a formal uncertainty as small or smaller than that obtained with Cepheids. But prudence demands at least one more golden lens before declaring that lenses do as well as (let alone better than) Cepheids.
The remaining systems with time delay measurements are not without value. The central concentration degeneracy may reasonably be lifted by appeal to other (non-variable) lensed systems for which the central concentration can be measured, or alternatively, by appeal to measurements of gravitational potentials of nearby elliptical galaxies. The present author would guess that the systematic errors associated with either of these assumptions would introduce an error of perhaps 5% in the Hubble constant. This would rehabilitate those optical quadruple systems that suffer only from the central concentration degeneracy.
Optical double systems suffer from both the central concentration degeneracy and the paucity of constraints. As they have no internal redundancy, their usefulness will depend upon the scatter in H0 values observed from a good-sized sample of such systems.
In the case of B1608+656, the outstanding question is whether the data are so good that, merging lenses notwithstanding, all plausible models give the same value of H0. What is needed now is the effort of a "loyal opposition" to search the far corners of model space.
New telescopes and upgrades to existing telescopes are certain to produce new gravitational lenses, and with them we will eventually see a Hubble constant that is less uncertain than that obtained from Cepheids. Even those who take it as a matter of faith that the universe is perfectly flat may find such a direct measurement of H0 interesting.
The author gratefully acknowledges support from the US National Science Foundation under grant AST02-06010, and thanks the members of the organizing committee for their good efforts.