For a multiple QSO, the
propagation time from the source to the observer varies from one image to
another, and this difference ($\Delta\tau$) can be measured when the source is
variable. In
general, assuming a flat universe without cosmological constant, the parameter
$\Delta\tau$ x $H_0$ depends on the redshifts of
the lens and the source, as well
as the positions and fluxes of the individual images. However, this parameter
is not related to the basic observations (redshifts, angular positions and
fluxes) in a single way. Firstly, the observations of multiple images of the
same source are used to infer a lens model : the source position and the
adjustable parameters that appear in the picture of the deflector (King
profile, etc.), and secondly, $H_0$ is deduced from $\Delta\tau$,
the basic observations and the lens model corresponding to the
lens picture.
A golden system (which is suitable for determining $H_0$) must be a multiple QSO with well resolved images and verifying some properties : (1) visible lensing objects, (2) simple lens with simple picture, (3) measurable source variability and (4) absence of strong short timescale microlensing. The best determined $\Delta\tau$ is that for the famous Twin QSO, which verifies (1), (3) and (4). For this system, the lens is complex (a giant elliptical galaxy plus a cluster of galaxies). In spite of the problems with the choice of a "good" picture, by using recent lens models, a ten percent measurement of $H_0$ is attainable. In this short review, we discuss the present quasi-golden systems and the perspectives in a near future.