# Adventures in clock operation

The Kings got a little hometown boost from the clock operator last night and scored with (supposedly) 0.4 seconds left to beat the poor Blue Jackets. If you watch the highlight video from the 25 second mark to the 35 second mark, you’ll see that the clock freezes with 1.8 seconds left. The NHL has admitted that more than 0.4 seconds elapsed during the freeze (so the goal probably shouldn’t count), but they aren’t going to change the outcome.

What was the Kings’ defense? From Kings’ GM Dean Lombardi (written to ESPN):

Those clocks are sophisticated instruments that calculate time by measuring electrical charges called coulombs — given the rapidity and volume of electrons that move through the measuring device the calibrator must adjust at certain points which was the delay you see — the delay is just recalibrating for the clock moving too quickly during the 10 – 10ths of a second before the delay — this insures that the actual playing time during a period is exactly 20 minutes. That is not an opinion — that is science — amazing devise [sic] quite frankly.

Let’s say the delay was 0.5 seconds (it was probably even longer). Lombardi is saying that this delay represents the amount by which the clock ran too fast during the previous “10 – 10ths of a second” (i.e., the previous second). So, the clock got ahead by half a second in only one second of game time? Science or no science, that seems like a pretty volatile clock. What if the clock ran that fast for the last second of the game? It’s not even worth checking this with a stop watch.

In search of the real answer and not well-versed in coulombs and calibrators, I checked with Brother Evan, who’s working on an engineering PhD. He referred me to the Wikipedia page about quartz time pieces (these are extremely common and plenty accurate for sports, so the Kings better have a good excuse if they use something else). We decided to be generous and assume that the 0.5 error was built up over the whole period, not one second. Evan says:

The claim by Lombardi is [generously] that the game clock sped up 0.5 seconds for 20 minutes.  A Timex (typical quartz wrist watch) will gain about .0069444 seconds in a 20 min period.

Uh oh. Under the “Accuracy” section in the Wiki, we see:

Some premium clock designs self-rate and self-regulate. That is, rather than just counting vibrations, their computer program takes the simple count, and scales it using a ratio calculated between an epoch set at the factory, and the most recent time the clock was set.

This kinda sounds like what Lombardi is talking about, and Evan assures me that a simple quartz clock would never have a “recalibration” delay. So, he concluded that either:

1. The fancy clock software messed up completely (goal shouldn’t count)

or

2. The fancy clock was off by at least 0.5 seconds (enough to calibrate), which means that this “amazing device” is worse than a \$5 quartz wristwatch (goal should count and it’s time to get a new clock)

The only other reasonable explanations we could come up with are that additional electronics between the actual clock and the displays created a delay or that the stops and starts during the game build up error. Evan suggested that maybe the timekeeper’s computer did a Windows update at a bad time and claimed too much RAM. If the clock truly needed a correction due to built up error, the correction should have been done sooner, at a lower threshold than 0.5 seconds. It sounds like Lombardi needs to talk to his tech team about their clock and about the email they wrote for him.

### 2 responses to “Adventures in clock operation”

1. John Arnold

This was a much more interesting and considered take on the event than Deadspin’s (http://deadspin.com/5881862/the-la-kings-explanation-for-last-nights-clock-discrepancy-is-bullshit-heres-why). Thanks to both you and Evan for going the extra mile with your analysis.

2. Thanks! Their video is pretty convincing, but it doesn’t answer whether there was some delay earlier in the period that needed to be corrected (this is why I didn’t bother to rummage around for my stop watch). Still, the fact that they run perfectly in sync from 10 seconds to 1.8 seconds remaining and then get off suggests that there really was an erroneous clock delay. The numbers in the article don’t quite match the video either . . . strange.