Here is a more likely scenario (that my friend the materials engineer gave me):

Gorilla Glass is made by "stuffing" ions into the empty spaces so that the glass is under very high compressive stress. It's really known as "High ion exchange glass" This in turn, gives it very high compressive strength. However, it does have a significant downside that never gets mentioned: Shock absorption and scratches can cause later cracks and shattering. This is just a fact of life. When you place the outer layer under compressive strain, and then begin to compromise part of the outer layer, or begin to stress it, you can eventually cause release of those compressive forces. Once partially compromised, it may seem to be good, but can fail instantly.

Completely agreed. That's exactly why I said "shattering from thermal shock is absurd, and the glass must be under some kind of stress from how it's secured to the phone" in the parent comment.

To further your explanation a little, the reason the outermost layer is put into compressive stress is that, when it's scratched, the stress will cause the crack to "close up" and so the glass doesn't shatter. Unfortunately, the interior of the glass is under tensile stress as a result, and so if the scratch happens to be deep enough to reach that tensile-stressed layer then the glass will shatter. Putting the glass in a case which alters this carefully managed stress state, i.e. by trying to attach it to a rim that's acting to pull it slightly apart, is what can cause this kind of failure.

But this is all speculation on my part, and even if true the reality might be more subtle.

Edit: I just found a good article from AnandTech that explains some of this. Take note of the first figure in particular.

http://www.anandtech.com/show/4225/the-ipad-2-review/13

http://en.wikipedia.org/wiki/Prince_Ruperts_Drop

I took a set of keys, a fork and a pocket knife to the front and back glass panels of the Nexus 4 and couldn't get a scratch to show up anywhere. I also slipped and dropped the phone while pulling it out of my pocket this week, and it showed no signs of my fumble.

Until I see a video, I'm not really convinced, tbh. If he has a granite counter-top, and didn't set it down as gently as the author claims, it's completely probable that it cracked. People are incredibly clumsy with phones. Glass doesn't exactly have a high thermal conductivity (nor does granite/linoleum/plastic), I find it difficult to believe the glass would expand/contract enough from body heat/kitchen counter-tops.

[1] http://www.wired.com/reviews/2012/11/google-nexus-4/

Which seems like a pretty good argument for not wrapping them in glass.

Too often in technology we build things that are fragile, and then blame the user when she breaks them. Tools should be designed to work with the user as they are, rather than as we would wish them to be.

Normal window glass is Mohs hardness 5.5 or so, as is unhardened steel (your fork and pocketknife); hardened steel (e.g. a metal file) is more like 6.5. That means you can scratch regular steel with glass and vice versa, with some difficulty. (Try it. I just did.) Brass (your key) is around 3 or 4, so it won't scratch glass.

There are lots of things with Mohs hardness above 5.5, including most common metals (except brass, aluminum, copper) and most rocks.

Sand is almost never made up of glass particles. It's often made up of crystalline quartz (Mohs hardness 7) which can scratch regular glass easily. Gorilla Glass has Mohs hardness of about 9, so quartz won't scratch it.

http://www.corning.com/docs/specialtymaterials/pisheets/PI23... gives its Vickers hardness as 701 kgf/mm^2, which http://en.wikipedia.org/wiki/Mohs_scale_of_mineral_hardness#... would seem to place in the 5-6.5 range of the Mohs scale, nowhere close to 9. However, Vickers is a scale of resistance to plastic deformation, while Mohs is a (highly nonlinear!) scale of resistance to abrasion, so perhaps take that with a gram of potassium chloride.

Other random places suggest sapphire's Vickers hardness is around 1800-2200 kgf/mm^2, which is substantially harder than the number Corning's brochure gives for Gorilla Glass.

As for Corning, in fairness to them they were instrumental in the development of this type of glass and own a fair few patents, so it's not just marketing. They developed it originally in the 60s, though I imagine a lot has changed since then.

From what I understand, Gorilla Glass is only scratch-proof, not tough.

This is just a theory, but I've discussed it with People Who Know Things (TM) and they all have agreed that it's more than likely a small sliver of plastic or glass (or maybe just a tiny dirt particle) has become lodged in the space between the glass and the phone during fabrication. 9/10 phones don't have these defects, but when you're manufacturing millions of phones 90% still leaves a lot of bad phones.

In any case, IIRC, speculation is that only the screen side of the iPhone uses Gorilla glass, but I'm not sure if that's changed with recent generations. But even without Gorilla glass, you can still use "standard" toughened glass that, correctly mounted, should never shatter just from the sort of thermal shock described in the article.

Maybe I just got lucky, but in my limited experience, the case is not needed to protect the glass.

But the OP is talking about breakage due to thermal shock, not to physical shock.