What you’re witnessing is a self reported dataset from an international community (read as “unconcerned with The Protestant Work Ethic”) of ship captains and owners.
There are so many edge cases and oddities that the more you look the more you’ll find.
The moral of the story here is that you cannot accept AIS data, or any single data source, to be God’s Truth; you must always skeptical of the data before your eyes, lest you fall victim to “The Analyst’s Fallacy” (believing that the data presented is all the data there is) and create false stories in your mind.
Couldn't a more rational theory be that the data is essentially captured over VHF radio waves and there can be data loss due to things like gaps in capture ranges, faulty capture devices, faulty transponders, or even signal congestion?
I would wager that the data is imperfect and the points are tweened to discrete samples rather than having a high detail point-to-point path. You'll probably see traveling over land more like cutting a corner. Just a hunch.
Some of the (likely) fixes look pretty weird, moving at a ridiculous speed, through land and sea in a straight line, only to reach a certain point and continue as if nothing had happened.
e.g. 8 June 2012 05:00 from western Sahara to English Channel, through Sahara, right the middle of Spain, parts of France.
Another on 9 May 2012 from Southeast Asia, straight through the south of India and right through the middle of Africa to somewhere on the Guinean Gulf in about half a day.
That sounds to me like those ships are missing transponder data for some reason, and the visualisation is extrapolating between to distant data points.
Not sure why this would be the case. Transponder turned off? Data missing for other reasons?
I've never seen a perfectly clean dataset of any significant size. Big datasets have all sorts of weirdness unless there was an extraordinary effort to keep them pristine.
