Thanks Byung-Soo,

What do you think are the essential features that have to be realised in the the "next horizon" D-Wave system to make it useful or at least to answer the question whether it is a promising direction?

D-Wave under no circumstances promises a "universal quantum computer". This latter beast is the holly-grail, and unfortunately we are far far away from having it. There is a huge debate even if D-Wave performs ANY quantum computation, however recent results tend to say that it does. On the other hand, as some people mentioned, for the price of a D-Wave machine, one can build a classical computer cluster that will solve the same optimization problem faster than D-Wave, so right now I wouldn't bet my money on D-Wave. There is a huge debate about D-Wave's computational power on Scott Aaronson's blog, that may be of interest to you  http://www.scottaaronson.com/blog/?p=1400

Regarding your second question, I believe that poly-time factoring for example will prove you without doubt that you have at least a candidate for a (universal) quantum computer. The optimization problems D-Wave attempt are really hard to decide whether they are making use of quantum speedup or not.

FYI, Google had recently recruited UCSB's John Martinis, a leading scientist in the experimental quantum information community, to join Google's Quantum Artificial Intelligence Lab. This venture will be closely linked to the work done at D-Wave, as both companies started collaborating at the middle of 2013. 

So, I think it's safe to say that we can expect big news in this field soon enough...

Assaf