Probably a minor point, but I'm curious about the use of "fast" in the article. Wouldn't it be more accurate to say that the affected clock would be running ahead, but not necessarily fast?
To me, "fast" sounds as if the second counter in the clock advances at a rate greater than 1 tick per canonical second. In this case, the fast clock would diverge from actual time (whatever point of reference, it doesn't matter) such that the concept of being 17 seconds away from NTP time would be an accurate statement only at a single instant in time.
It seems that the article is not discussing the accuracy of clock counters, so I am ignoring the fact that no clock is perfect for now. It seems like 17 seconds ahead is much more accurate to capture the idea discussed, that the clock in question advances its second counter at the appropriate rate, but the value of that counter is 17 greater than that of NTP time, and will continue to be so until the next leap second.
I'm not sure there's a single, standard usage. Among (mechanical) watch enthusiasts, who certainly have an axe to grind on the semantics of time, a "fast" timepiece is one that advances too quickly, usually on the order of a few seconds to a minute per day.
The watch enthusiasts are technically correct, but the colloquial meaning while not technically correct is also a phrase that people can use to effectively communicate the idea to a great many people. So they're technically wrong but language-ly correct.
To me, "fast" sounds as if the second counter in the clock advances at a rate greater than 1 tick per canonical second. In this case, the fast clock would diverge from actual time (whatever point of reference, it doesn't matter) such that the concept of being 17 seconds away from NTP time would be an accurate statement only at a single instant in time.
It seems that the article is not discussing the accuracy of clock counters, so I am ignoring the fact that no clock is perfect for now. It seems like 17 seconds ahead is much more accurate to capture the idea discussed, that the clock in question advances its second counter at the appropriate rate, but the value of that counter is 17 greater than that of NTP time, and will continue to be so until the next leap second.