Well I didn't look at the PDF but that's a pretty worrisome headline.

What's gonna keep us all from being blinded on the highways?? 

Common courtesy?  

This is a marketeering claim. The light from a Metal Halide lamp (which is what automotive HIDs are) is definitely not at all close in quality to that of sunlight. Three main characteristics of any light are its spectral power distribution (SPD, the absolute presence and relative prevalence of the different wavelengths that humans can see), its colour rendering index (CRI, the fidelity with which the light reveals colours, compared to standardised sunlight conditions), and its correlated colour temperature (CCT, applicable only to white light, basically whether the light is "cool" or "warm" in appearance).

SPD:

The visible portion of sunlight is a continuous spectrum from red to violet, with no gaps. The visible portion of a glowing filament (which is a blackbody radiator) is likewise a continuous spectrum from red to violet, with no gaps. The spectrum of an HID lamp is a series of peaks and valleys. The light is superabundant in certain wavelengths (colours), relatively deficient in others, and absolutely deficient in still others. So from the standpoint of SPD, halogen headlamps actually are much closer to sunlight than HIDs. Which is better? Well..."better" is tricky to define here, because it really depends on what exactly we're trying to do with the light we're creating. In general, a continuous spectrum (rather than a peaks-and-valleys spectrum) is better, because it makes it easier to get a higher CRI, which I'll get to in a moment. But that's definitely not an inviolable rule! Sometimes (as for example when driving through fog or snow) we want to filter out a portion (blue to violet, in this case) of the spectrum. And for general illumination, there are many excellent discontinuous-spectrum lights (fluorescents, HIDs, LEDs, etc.), though this is not an either/or situation. The old fluorescent lights and mercury vapour street lamps produced yucky-looking light because of gross excesses and deficiencies (peaks and valleys) in the spectrum, but today's phosphor and halide technologies are giving us fluoro, HID, and LED lights that may have a peaky spectrum, but contain enough of the various wavelengths to produce a good-quality light. It is worth noting here that there is no such a thing as "full-spectrum" light. The term is used by marketers of everything from headlight bulbs to seasonal affective disorder lights to reading lamps to new fluoro tubes for your kitchen, but it means whatever any particular marketeer wants it to mean. There is no standard definition — not even close.

CRI:

Obviously, not all sunlight is the same, so a set of conditions has been standardised. In greatly simplified terms, the conditions can be understood as "noonday sun on a clear day". This is considered to be a CRI of 1.00 (sometimes stated as "100"). There is no light of CRI higher than 100, and a higher CRI is always better than a lower one except in certain very specialised lighting tasks (as for example in photographic darkrooms or in situations where ordinarily-tangential factors such as preservation of night vision, rather than ordinary factors like effective illumination, are the priority). A properly-fed tungsten-halogen filament lamp with a colourless glass or quartz envelope has a CRI of between 0.9 and 0.99 ("90" and "99"). Current-production automotive HIDs have CRI of between 0.7 and 0.74 ("70" and "74"). So, again, from the standpoint of CRI, halogen headlamps are closer to natural sunlight than HIDs.

CCT:

This is measured in Kelvins (not "degrees Kelvin" as is sometimes incorrectly stated), and is directly keyed to the kelvin temperature of a blackbody radiator. In this scale, there is no such thing as "better/worse", just different/same/similar. The standardised sunlight conditions described above are considered to have a CCT of 6500K. Automotive HIDs (real ones, not ones that have been jiggered to produce bluer-than-standard light) are between 4000K and 4500K. Properly-fed tungsten-halogens are between 3100K and 3450K. So, in this respect, automotive HID headlamps are closer to sunlight.

Now, what are the safety performance implications? Enough research has been done to show that the poorer CRI of HID headlamps is of no safety consequence. Stop signs still look sufficiently red, for example, and guide signs still look yellow enough. The SPD might be causing some glare-related problems. Automotive HIDs have a high spike in the blue-violet region, and there's pretty good evidence that just as some people are glare-sensitive and some are not, some people are blue-sensitive and some are not. This is not a medical condition or disability, it's just a human variance like nose size or eye colour. There's also prety good evidence that at any given intensity, headlamp light with a higher proportion of blue light causes more glare than headlamp light with a lower proportion of blue in it. There is competing evidence, however — yes, academic researchers do compete with one another, with theories and studies and data instead of long-jumping frogs or whatever — suggesting that a higher blue content improves certain aspects of drivers' night vision. Scientifically this one hasn't been shaken all the way out yet, and it's possible both effects might exist simultaneously to some degree. From a marketing perspective, the question is moot; the decision's been made to push more and more towards the direction of bluish-white car lights.

Up to now, most of the research has effectively conflated CCT and SPD, because of the limitations of the headlamp light sources available for study: Tungsten-halogen bulbs have a high CRI, a continuous SPD, and a relatively low CCT. HIDs have a low CRI, a discontinuous SPD, and a relatively high CCT. This is to some degree an implementational limitation, not a conceptual one, and in my opinion it is likely to be found, eventually, that a blue-rich SPD can cause glare problems but a high CCT can potentially improve seeing performance. That is going to be a tricky balance to optimise, for high CCT to a significant degree goes along with blue-rich SPD. But we're now seeing LEDs that have a higher CCT than HID headlamp bulbs, but without a proportionately higher blue spike. It will be interesting to see what shakes out of this. The marketeers may have to find another tactic, having already painted themselves into a corner using blue paint: up to now, the bogus claim of "whiter" headlamp light has been used to refer to light that is in fact bluer. When it becomes possible to provide headlamp light that is of higher CRI and higher CCT rather than just higher in blue content, that light will in a more real sense be "whiter" than HID headlamp light...but what are they going to call it...?

The Audi R8 as sold outside North America is equipped with the first volume-production all-LED high/low beam headlamps. The Lexus LS600h has LED low beams, and the forthcoming '08 Cadillac Escalade (whatever they're calling the ultra-bling edition) has LED high and low beam headlamps.

Yep...all the light source people (Osram Opto Semiconductors, Philips Lumileds, and a few others) are developing better and better white LEDs all the time, and all the lampset makers (AL-Automotive Lighting, Hella, Valeo, Ichikoh, Stanley, Koito) are hard at work developing new optics to make headlamps with those new LEDs. The Wikipedia "Headlamp" article contains good information on LED headlamps.

Well...two things. First off, MB has traditionally been rather conservative in adopting new technologies, at least in the area of lighting. They were still equipping their top models with 1972-technology headlamps (H4) well after more advanced bulb and optical technologies became available, they were far behind BMW and Audi in the adoption of projector optics for HIDs (really the right way to do it), they utterly failed to take advantage of any of the various changes in US lighting laws to allow better headlamps than the junk they put on most all US models between 1986 and 1994, etc. So, this is not new.

Secondly, LED headlamps aren't yet better than HIDs. Today's best LED headlamps consume more power than HIDs, give beam performance somewhere between halogen and HID, cost a bloody fortune, and bring with them complications and issues not present with HID.

Yes and no. Adoption of LED headlamps will definitely happen, but not to a significant degree any time soon. In terms of prevalence, even in the developed countries, HID headlamps are somewhere between a minority (Europe) and a trivial mini-minority (North America). Now, LEDs do get better and more efficient every time we turn around and blink, as it seems, but we've only just reached the point where white LEDs produce enough light to make a reasonable beam pattern. Give it some time, then give it some more...the cost differential at the production level between HID and LED headlamps is more than ten times greater than the differential between halogen and HID has ever been, even when HID was brand new. Obviously LEDs have major advantages in signaling and marking applications (parking lamps, turn signals, tail lamps, sidemarkers, brake lamps, reversing lamps, daytime running lamps, interior lighting) and their market share in those areas is rising rapidly. Keep in mind, most of us are used to thinking of LEDs in terms of indication (you look at the lamp to get a message) rather than illumination (the LED emits light which you bounce off what you want to see). We've had LED indicators for years on our electronic devices, more recently on our kitchen appliances, on car dashboards, we see them now in traffic lights, etc. We consider LEDs to be ultra-efficient, low-power-consumption devices, but in fact their efficacy (amount of light out per unit of electricity in) is at this point comparable to the best filament bulbs. Compare this with HIDs, which have an efficacy about triple that of a good filament bulb. LED headlamps are quite power-hungry at this point.