First off, I have an MSc in optics - but I don't work in auto headlight industry. While there are good and bad HID kits (I have had both), I have to take issue with quite a lot of the information on the Daniel Stern website. This makes me wonder that there may be some bias because of some vested interest of the author.
Firstly and fundamentally, the diagram here comparing arc and filament:-
http://dastern.torque.net/Photometry/filamentarc.jpg
Assuming the arc and the filament are the same or similar lengths, in the same position, on the same axis, then it seems to me that the arc shown will be a fine replacement for the filament (in a projector style headlamp). The 10,000 cd/cm2 bright spots on the ends of the arc are such a small volume of the rest of the 6,000 cd/cm2 arc they should not cause great concern.
Next, I quote from Daniel Stern: "This article will never go out of date, because the problems with HID kits are conceptual problems, not problems of implementation. Therefore, they cannot be overcome by additional research and development". I disagree with this also. I repeat, if the arc is the same dimensions as the filament in the same place as the filament, then in a "projector" style headlamp the arc will project in a similar way to the filament. Hell, the lumps and bumps in illumination in a modern filament headlight are still obvious, and the filament has all those gaps betwen the wires.
Also, I quote: "Probably the biggest issue is HID headlamps' significantly worse color rendering index (CRI), which is in the high-60s to low-70s range. Halogen headlamps' CRI tends to be around 90 to 97 or so."
That 60-70 CRI doesn't fit with the info I can find. HID, being an arc lamp, can give a much flatter spectrum of illumination than a filament. From wikipedia: "The output of a pure xenon short-arc lamp offers fairly continuous spectral power distribution with a color temperature of about 6200K and color rendering index close to 100." EDIT: Turns out automotive HID uses metal halide arc lamps which have a more spiky spectrum- so the CRI could be worse.
I would add that my own experience of a cheap HID kit shows that there are some bad kits out there, and it's not as easy to get it right as the vendors would have you believe. The most important thing is that HID conversion should be done in a projector style lamp, not a reflector style, because the metalised end of the filament contributes to the shutoff in a reflector lamp. But there's no fundamental reason (I can see) which would stop an arc replacing a filament in a projector lamp, if an arc of the same size is situated where the filament would have been (and on the same axis as the filament).