So they improved the decoherence by a factor of 10. This is nice, but no reason to abandon your RSA keys just yet. The real problem with quantum computing is not decoherence (i.e. the losing of superpositions due to uncontrolled entanglement with the environment) - its quantitative imperfections.
A quantum computer is basically an analog device. As you cannot observe states, there also is no way to "refresh" slightly inaccurate states, as a normal digital computer does. A NOT has to be exactly 180 degrees and not 179 or 181. No problem in toy or laboratory setups, where you only do a handful of gates and keeping your system isolated is the (currently) much bigger problem. 1% error might seem quite good in this setting.
But for any meaningful computation, you will require many millions of gates and your experimental accuracy will have to keep pace with that - in addition to keeping your system from decohering (which - at least in theory - can be mitigated by quantum error correction). Rotation angles would have to be not 1% but 0.0000001% accurate.
The problem is too remote to get much consideration now, but I'm sure that it will prove to be the final (and probably insurmountable) roadblock for any real-world use of quantum computers.