A Community Devoted to the Preservation and Practice of Celestial Navigation and Other Methods of Traditional Wayfinding
From: Frank Reed
Date: 2021 Apr 27, 11:58 -0700
How would you detect cheating? I agree with Greg and David that it's not likely, and the nature of the competition works against it, yes. But when cheating is easy, people cheat. So how could you detect it?
I have listened to stories from people who have been quizzed after races that require celestial in some categories, Marion-Bermuda specifically. The inquisitors ask all the wrong questions, like details about the sight reduction process. They have no comprehension of the possibilities.
I've put some thought into this in the past few years while "designing" (and then shelving) a sight scenario simulator. I have use for this in my classes. While there isn't a good simulator available right now, it's clear that an excellent one could be created by considering enough variables and options. But a simulation would probably not match the statistics of human sights at the level of detail below the real noise in the observations. Specifically, many navigators become obsessive about recording tenths of a minute of arc in their sights. These are below the noise in nearly all practical navigation (with a few barely practical exceptions, like lunars). And here's the thing: navigators have favorite tenths. When recording tenths of a minute, the scale can certainly be read clearly to +/- a quarter of a minute of arc, so no one would record 10.7' when the correct value is 10.3. But if it looks like it's between 10.6 and 10.8, some navigators will always record 10.6 and some will always go for "lucky seven" and record 10.7. The tenths in real sights will show preferences, and very likely they will be observer-specific, constituting a fingerprint for each observer's work (enter Lt. Columbo, with "just one more thing"). The tenths in simulated sights, by comparison, will be uniformly distributed.
At the level of normal noise in sights, a poor sight simulation sight would include no random error at all. That would be obvious quickly. A bad sight simulation would add uniform errors (for example numbers pulled at random from the range -1.5' to +1.5'). And a mediocre sight simulation would use normally distributed errors. Both of these latter choices would probably be detected in the sight statistics if more than a few dozen sights are available. Real celestial sights have heavier tails to the distribution than a standard normal "bell curve". I find that errors resembling real sights can be generated by a simple double normal distribution. You draw from a normal distribution with a typical standard deviation, like 1.2', for 80% of sights, and then, randomly, for 20% of sights you draw from a wider normal distriburion with an "outlier" standard deviation, like 2.5'. Then errors look real.
Let's throw an albatross into the works. Suppose you're halfway around the globe, sailing fast, looking forward to winning the respect of all who doubted you, and while you're busy making lunch some sunny day, a frigate bird (ok, not an albatross) comes along and, with its unusually strong beak, manages to crack the micrometer drum off your sextant and flies off with it. Maybe you could repair the sextant? Maybe rig up some sort of replacement for the micrometer drum? But suppose, in anger, you have thrown your damaged sextant at that darn frigate bird as it flies off into the distance. What else could you use as a fallback? If you could consult with NavList regulars, I bet you could come up with interesting options inside an hour of the frigate attack. But alone and with no special expertise in celestial navigation? Maybe you would dig out the iPad and cheat.
In the near future, I would bet that races like these will depend on live video streams from every boat if they hope to get sponsors to pay the bills. Then cheating at celestial would become even more absurd. One would have to go through the motions at the right times with nearly correct orientations in azimuth and index arm position then carefully synchronized with simulated sights. Putting on an act like that would be more difficult than just taking real sights. Enter Lt. Columbo again: "Now, Sir, you see, Sir, here in the video, you're facing the right way for every sight... oh yes, you were very careful about that. But you forgot one thing: you left the index arm at zero for every sight... you see, Sir, in the video... in every case, the index arm stays at zero while you appear to adjust the micrometer...".