How close are mankind to creating a quantum computer?

[eextreme]

I read about how a quantum computer would work, but how close are we to making one. If it isn't possible what would prevent it from being created?

 

[per.sundqvist]

The decoherence time of a quantum state in a qubit is short, say 1ps-1fs. After that time the memory is destroyed, any you cannot read. You need low temperatures as well, perhaps below liquid helium ~4K. Algoritms for quantum computers exist, but they are simulated on normal computers. In principle its possble to make a prototype in a cleanroom, but i doubt that it will work properly. Heard about stories that NASA has one, but i am not in the feld so i cannot tell you more precise.

 

[f95toli]

The decoherence time of a quantum state in a qubit is short, say 1ps-1fs. After that time the memory is destroyed, any you cannot read.

It is short, but is not THAT short. The coherence times of e.g. atomic/ionic spin systems can be quite long (ms or longer) but the problem is that it is very difficult to scale such systems up to practical computers. Another problem is the time it takes to perform a gate operations is relatively long (remember that it is not only the coherence time that counts, it is how many operations you can perform during that time)
The limit so far is only about 8 qubits, but that has still been enough to demonstrate most of the algorithms. I.e. quantum computers DO already exists, they are just still too small to be practical.

Solid state QC are -at least in principle- easy to scale up to large sizes but they have much shorter coherence times that atomic systems. But there has been dramatic progress over the past few years. The firsts qubits (1999-2000) had coherence times of only a a few ns but that time is going up rapidly, mainly because we are now starting to understand what is limiting their performance (2-level fluctuators in the material etc). The current record is something like 10 us (for T1, T2 being a bit shorter due to pure dephasing) using a superconducting flux qubit.

One advantage with solid state qubits is that they are easy to control and the gate time can be less than 1 ns (using microwave pulses) so you can perform many operations before the qubit decays. The main problem is that the coherence times drop very quickly as we start to couple qubits together and so far only 2-qubit systems can be fully controlled (although coupling has been demonstrated with up to 7 qubits).
This is very much work in progress.
Note that even a 20 qubit QC would be very interesting, not so much for algorithms (too small) but for simulating other quantum systems; it is difficult to simulate e.g. large molecules with more than a few atoms on a classical computers but a QC would -at least in theory- be well suited to this task. I.e. the idea would basically be to use a QC as a "toy system" to test ideas about complex quantum systems.