There is one sure-fire way to identify a good programmer: their work. In fact, it is by definition the way that we should evaluate programmers. It is the great equalizer, and with web technology, exposed for all to see.Unless you are presuming a team of only a couple of developers doing their own thing, there is a real challenge in determining the extent of the programmer's responsibility for something that is exposed for all to see.
There are a number of other causes that might make a site a developer helped work on look or run like complete garbage. Organizational inertia might have forced the site into sucking, regardless of how good the programmers developing it were. The organization may have one or more bad programmers who were assigned to work on some critical component, screwing the whole site and all the good programmers.
We can wax on eternally about how the good programmers in this situation should've seen the writing on the wall and gotten out as soon as possible, but that's very easy to say seated in armchairs gazing from afar. Maybe they thought things were going better than they really were; maybe they saw things going bad, and thought they could motivate others to turn things around before things become a disaster -- and maybe they failed trying to do that[1]; maybe they have been trying to get out, but they are already tainted by the bad work of the organization that has become part of their image as well.
They could do work outside of their employer, for instance on an open source project, but these can succumb to the same sort of pressures that push profit-seeking organizations toward mediocrity. A single developer, even a really willful one, will have a hard time overcoming this inertia.
[1] At one of my prior employers, I was in this boat; I both tried and failed to get things turned around. I did quit, but probably not soon enough to convince people that the site was not a reflection of me and my standards. The site was absolutely shameful, and I'm sure anyone who used that site to infer things about my skill level would've sworn I had no skill or experience outside of a weekend PHP crash course. Granted, it would vet one organization I would probably not want to work at; but it would represent wasted potential based on presumption, which is a shame.
I really think the details matter, because they are a chance to determine if the work you are using to assess the programmer is spurious input, and reflect an influence largely made up of things other than that programmer.