FIRST, we have to understand that there are several DIFFERENT Time "Standards".....and in many [not all] cases, entering "LEAP SECOND" corrections has been a MANUAL Process that is prone to HUMAN ERROR:
GPS, UTC, and TAI Clocks
The World-Wide accepted
"ATOMIC TIME" [aka TAI] is a particular "ATOMIC CLOCK(s)" located in Paris, which continually counts the number of Exceedingly Accurate Frequency Oscillations ever since it was "adopted" as the WORLD TIME STANDARD. It is presumably NEVER "corrected".
OTOH, the PRIMARY way time is distributed around the World is via U.S. GPS [and Russian GLASNASS] Satellite Navigation Systems. "GPS TIME" was calibrated to
ATOMIC TIME before launch of the FIRST Satellite and has been free-running ever since WITHOUT CORRECTIONS. Although the
GPS TIME Clocks are synchronized to each other, the Frequency Standard in the chosen MASTER isn't as accurate as the ground-based ATOMIC CLOCK, hence the current
GPS TIME is now AHEAD (Earlier) of
ATOMIC TIME by 19 seconds, as shown in above link.
The time used by MOST people is actually
UTC TIME, [Universal Coordinated Time] as kept by the U.S. Navy Observatory in Wash. D.C, and distributed from Ft Collins, CO by WWVB (VLF Freqs to 1-ms accuracy, received by many Wall Clocks), WWV (HF Freqs to about 10+ ms accuracy), Dial-In Phone Announcements and various time.nist.gov Internet Webservers (1-ms accuracy).....the latter is how most PCs get ACCURATE Time.
NOW it gets INTERESTING: Every several years (or NOT), UTC is CORRECTED by infrequently occurring LEAP SECONDS and [mostly] predetermined LEAP YEARS so that it aligns with the actual motion of the Earth (which is slowly spinning slower) so that Apogee and Perigee always align up with the ACTUAL SEASONS.
So
UTC TIME is currently Earlier than
GPS TIME by 18-seconds...and Earlier than
ATOMIC TIME by 37-seconds.
Which doesn't explain the 10-sec difference YOU saw [maybe they decided to split the difference???]. Many (esp. old) GPS Receivers ONLY Output
GPS TIME, so Leap Seconds had to be (mostly MANUALLY) entered into the Interfaced Software. And some (later) GPS Receivers can be setup to Output and/or Display GPS or UTC...with Leap Seconds and perhaps also Leap Years being MANUAL entries in most deployed GPS Receivers [I would continue to implement it AUTOMATICALLY in the S/W that Uses
GPS TIME].
Since Leap Second is mostly a MANUAL operator action, it is way too easy to forget to do the correction each and every time it is announced....and from my experience repeatedly fighting this issue with my ever changing group of local cable "engineers", it seems to be intermittently entered as they only occur every several years, it's easy to FORGET to DO the requisite procedure.
Also note that the LOCAL Cable "UTC" Guesstimate [which determines DVR Start/Stop Times] is usually DIFFERENT than the "UTC" Guesstimate that the INDIVIDUAL Networks are using to drive their PROGRAM Start/Stop Times...which I THINK also determines LOCAL Ad Insertion Times.
Additional info for the Uber-GeeK:
http://digital.ni.com/public.nsf/allkb/C5BBF6AC7036CDAA8625733000668351
https://confluence.qps.nl/display/KBE/UTC+to+GPS+Time+Correction
https://www.nist.gov/pml/time-and-frequency-division/nist-time-frequently-asked-questions-faq
http://www.oc.nps.edu/oc2902w/gps/timsys.html
http://tf.nist.gov/tf-cgi/servers.cgi