The two-current model of superconductivity Standard text books (Landau and Lifshitz etc) explain that the correct macroscopic theory of a superfluid is two-fluid hydrodynamics. A superfluid is made of two components, a viscous normal fluid and a non-viscous superfluid. The total current is the sum of normal and superfluid contributions, and in thermal equilibrium the density of the normal component is not zero. This picture is supported by microscopic theories, and by many experiments, most notably the observation of second sound. 
Then, presumably, the correct theory of a current carrying superconductor is the two current theory. Again, in equilibrium the density of the normal component is not zero, and a generic current should always contain both components. Why do we (almost) never discuss this theory? Is the resistivity of the normal component so large that it is quickly dissipated away? Or is the normal density typically too low to be relevant? Where is the second sound mode? (Is it screened? Or damped?) 
As a corollary, what is the right way to think about a persistent current in a superconducting loop? After some time, the normal current is presumably dissipated away. Does that mean that the supercurrent state is not in thermal equilibrium? If yes, why does it persist? (Is it meta-stable?)
 A: If we have a circuit with a resistor $R$ Ohms  in parallel with an $R=0$  resistor, all the current will flow through the zero resistance branch. This is what happens in a superconductor --- the supercurrent short circuits any  normal current that might otherwise be carried by thermally excited quasiparticles. I expect this is what you mean by "is dissipated away", so your supposition is correct. 
Even at zero temperature, however, if the frequency is such that $\hbar \omega> 2\Delta_{\rm gap}$ quasiparticles will be excited and dissipation will heat the system. In this case we can think of the heating as being due to  a normal current.
A: I will make a lame attempt to answer my own question. I found a paper by Jim Bardeen titled ``Two fluid model of superconductivity'', Phys.Rev.Lett. 11, 399 (1958). Bardeen states

Because of the very high attenuation of the normal component from interaction with the crystal lattice, phenomena such as second sound
would be difficult to observe in superconductors.

Unfortunately, Bardeen does not really explain what this statement is based on.
Note that if the damping of these modes behaves like collisional damping of MHD modes,  $\gamma\sim\jmath^2/\sigma$, then the problem is that the conductivity is too low, and second sound modes would be easier to observe in a dirty superconductor.
Bardeen does indeed state that the free energy goes as
$$
 F = \frac{\jmath_s^2}{\rho_s} + \frac{\jmath_n^2}{\rho_n} 
$$
where $\jmath=\jmath_n+\jmath_s$ are the normal and superfluid components of the current, and $\rho_n$, $\rho_s$ the associated densities. This clearly shows that in equilibrium both components must be present (if the free energy is minimized with $\jmath=\jmath_n+\jmath_s$ fixed, then both $\jmath_n$ and $\jmath_s$ are non-zero).
This means that a persistent supercurrent is not an equilibrium state. It should be possible to estimate the barrier for the decay of the supercurrent
states, based on flux quantization and the expression for the free energy, but I have not tried to put in realistic numbers.
