So. What is safe? For example, can a gun be safe? A gun is designed to kill yet many easily fired modern weapons include a safety switch. The gun designers were able to make guns very easy to fire and so came to appreciate that accidents happen and that "safe" is a spectrum - a safety switch made sense.
Here the Luge Officials are claiming the track is safe. Yet, for that statement to be true, you must assume that there will never be athletes who make mistakes or miscalculate curves -- no accidents. The New York Times has an excellent explanation of the accident which shows (to my amatuer eye) how easily this sort of accident could take place.
The Olympic Officials have also modified the track, adding a simple wooden wall -- which in all likelihood would have saved him from the critical strike -- claiming it is merely for "emotional support" of the athletes.
It's the same old problem that designers everywhere fail to grapple with. Improper usage. Now, some accidents are difficult to conceive and plan for. (Heart-attacks or strokes while driving cars). Nevertheless, the sport itself is only inherently dangerous when there is an accident. Cars, for example, are perfectly safe until humans make mistakes (not withstanding Toyota). The track designers should take into account the severe danger the human bodies are being exposed to and take action to mitigate it when something invariably goes wrong.
The lesson here is when you design something it is quite important that you consider what's the worst that could happen. Clearly, when safety is defined in this way, the track was not safe. The specific choice one might call into question: By adding the roof over the luge track, they needed support columns which are obviously an extreme blunt force obstacle should a rider be ejected from the track.
For more on this, I recommend: http://blogs.wnyc.org/radiolab/2009/11/03/helicopter-boy
And the best design doesn't end after construction -- it continues with evaluation (where the track had already been evaluated by plenty as being prone to rider accidents). All those steps were missed.
May Mr. Kumaritashvili rest in peace. May designers learn. When you create, what's the worst that can go wrong? Learn the lesson of the Helicopter Boy. (Toyota I'm talkin to you, too)