It seems that a hotel which survived the big earthquake, came down on a smaller quake. This is the same as Christchurch, and results from either the smaller quake being closer and having a higher peak ground velocity (PGV), or the building has had its seismic capacity (in terms of PGV) greatly reduced by the first quake.
In Christchurch it was a combination of both. Here, the hotel was badly cracked in the first quake, and still they jammed it with aid workers. Thus arises the contentious issue of the role of ground motion duration in determining damage. The issue is confused by the fact that nearly all earthquakes are over in a couple of cycles, but on soil basins the cycle (sinusoidal motions) can go on for 30 seconds or more. Yet the soil greatly amplifies the PGV. So, can duration override the PGV in determining structural damage? My head hurts at the complications....
We do know that the first few cycles can damage the structure. At this point, the building becomes softer and can gather energy at the lower frequencies. As well, the relative-displacement-sensitive points have moved. This can be the pull on rebars, or the way the floor hangs on untied lips. Thus, buildings can survive in this state, and only need a puff to knock them over.
An engineer evaluating this building is left to hang in the breeze. Unless he knows exactly how crappy the building is (made of sand with no steel) he cannot determine the critical displacement to collapse. A good, ductile building can easily take cracks. There is tremendous political pressure not to red-tag every building in sight.