I work in this field. Some notes here "Unlike ADS-B in a plane, AIS can only be turned off in rare circumstances." AIS can and does get turned off all the tim

I work in this field. Some notes here "Unlike ADS-B in a plane, AIS can only be turned off in rare circumstances." AIS can and does get turned off all the time. Different countries have different laws on this, so you will see very different patterns on where AIS goes dark. Sometimes vessels will turn off their AIS as soon as they leave some EEZ for international waters, sometimes they just turn it off if there's little risk of getting caught. As the article says, AIS is primarily used for safety measures - and is what it was designed for. But most developed countries have since started to use AIS for maritime surveillance, and have in the later years codified in laws / regulated how vessels of certain size must use their AIS transponders. VMS, which has a much greater capability in sending out data (catch logs, for example), has become more and more common. It is also difficult to know whether or not a ship deliberately turns off its AIS transponder, or if it is in an area with poor coverage. AIS data is collected via satellites and land based antennas - some data providers also provide signal strength to give a clue. When presenting AIS data in court, we never, ever say that a vessel has turn off their AIS transponders - only present the data and some probabilities. Bogus AIS tracks is a massive, massive problem. I frequently analyze areas where as much as 100% of the "traffic" turns out to be junk. They _can_ be pretty easy to filter out if there is some pattern of randomness, and a lot of the junk is just that - the AIS messages are just scattered uniformly around the globe, leading to impossible speeds. 99% of junk tracks are either registered as Chinese, or some micro-nations with little to no maritime activity. Spoofing is also a problem. In my work, we've been analyzing the Russian tanker movement closely for a couple of years now, and some of the spoofing has been really sloppy - so sloppy that the public quickly noticed. This is my favorite, because it was so low effort: https://gcaptain.com/blink-and-youll-miss-it-russia-deploys-second-stealth-lng-tanker-to-sanctioned-gas-project/ https://gcaptain.com/blink-and-youll-miss-it-russia-deploys-... (if you look at the data, you'll see that they just spoofed the AIS messages with a constant speed for two week straight, going through the exact same positions multiple times). With that said, the average of the data gives you nice patterns. There's a ton of variance due to all the fake data, missing data, faulty data, etc. - so in a day-to-day job a lot of your work is to figure out whether or not the dataset is a valid one.