> As it turns out, doing the experiment 15 times and taking the
> majority (plugging 7 into S(n)) will give you the correct answer
> 99.4% of the time. Doing things 35 times gets you to five nines
> of accuracy... completely reasonable in my books.
If it were a standard electronic computer, you'd return it for a refund and never buy anything from that manufacturer again.
Five nines sounds good, because for an *uptime* figure, it is good.
But for computational accuracy, that's abysmal. At a speed of only 1 GHz, which isn't exactly a screaming speed demon these days, your computer would be making ten thousand mistakes per second. For traditional computing applications, that makes it so much useless scrap. If our computers make one computational error per *hour*, we complain about how horribly buggy the software is, because we have come to expect that our hardware won't do that to us.
The other poster's suggestion, of using the quantum computer to suggest candidate answers and a traditional electronic computer to check them, makes much better sense. There are whole categories of computing problems where checking the answers is trivial, if you can in the first place get a candidate answer that's at all likely to be correct.
That, and obviously this is a research prototype and the accuracy could potentially be improved in the future, just as we expect that the performance level will be improved by further research.