I'll protect my chips with vinegar, but all the same, will I still need to wear my aluminium foil anti-positron cap? You see what's on my mind! Dissemble no more!

Cosmic rays may come in batches, I don't know, but we are told that a "cosmic ray" is basically a particle.
So, given that the ray detector is different from the processing unit that it is supposed to be "protecting", what guarantee is there that the particle that caused the detector to go off actually hit the processing unit ?
In other words, what is the use of detecting that a fault has been caused by a cosmic ray ?
Wouldn't it be better to detect that there was a fault in the first place ?

"Cosmic rays are actually not really rays at all" ? 
Kind of misleading. They are particles, and they are also waves (as described by Schrodinger's equation), and both propagate in a ray-like manner. Rays have an origin and a direction. So yes, they are actually rays.

The car example is borderline ridiculous - I mean you'd expect your car electronics to be quite robust, at least the more basic and critical ones. It's not like your ABS system needs to do calculations in the teraflops range, so I see little sense in having the chips to do with that being manufactured in the 35nm process. Will serve only to make them fail (as in break up) that much quicker.

Besides, running two systems parallel and recalculating if the outputs differ would be a much better way anyway - i.e. having redundancy, rather than second-guessing some mysterious random ray effect may or may not have on the calculations.

The technology may have it's merits and uses, but I'd rather not have it confusing my ABS.

Phillipos, satellite detectors are certainly affected by cosmic rays, more so than on Earth because the atmosphere blocks most of them from reaching the ground. I used to work in the field - we would buy (and extensively test) rad-hard memory chips for the satellite on-board memory to prevent exactly this kind of thing. The technology has been around for a long time.

Camera sensors are quite often affected by "cosmic rays". In fact, according to Sony, most dead pixels in camera sensors are due to cosmic rays.
Also, commercial satellites and space instruments use very primitive chips (compared to what's new), so they won't be easily affected by cosmic rays. Some new insulation might work wonders in the field, allowing the lastest cpus on board.
After all, with Nasa prepaing a mission to mars within the next 50 years or so, ther'd better be some extra protection after all.

I think it shouldn't say "to protec humanity" since the article is about to protect the effects of cosmic ray to electronic chips. When i first read the title I was like "when did intel enter the health business?"

I have my quad core Intel pinned down by one great hunk of aluminum and copper heatsink, I would have thought it the blocks of alloy cooling modern processors would go a long way to shielding expensive silicon from all but the very highest energy particles we get at the bottom of the atmosphere. In light of this, why not just redesign existing heatsinks to be a more effective shield for high energy particles?
You could magnetize a steel pc case to improve cosmic ray shielding.

I mose definitely dislike this snide remark implying that free software crashes computers while proprietary software's propensity to crash goes unmentioned:

"So aside from saving the world from the rare and relatively harmless computer crashes not caused by free software, ..."

You should have omitted the word "free". All types of software cause computer crashes. I won't try to argue which kind is more liable to crashes. My objection is to the implication that free software somehow is more liable than proprietary software to cause crashes.

The effects of cosmic rays are or at least were one of the main reasons for applying ECC to large DRAM arrays. Memory corruption attributed mainly to alpha particles was perceived as a serious problem in the early '80s when 64K bytes was quite a lot of RAM. Perhaps modern memory devices are less succeptible to such effects given that an ordinary PC can contain several gigabytes of DRAM and never appear to suffer from memory failure.