From: kymhorsell@gmail.com   
      
   An email from someone that maintains a paranormal site got me thinking   
   about the old phantom airship phenomenon.   
      
   Back before flying disks   
   were a thing there were a couple of "flaps" where people across the   
   US, Europe and even New Zealand reported what was usually described as   
   "airships" flying over their cities.  As reported in newspapers it was   
   even claimed the airships were typically manned by people. In some   
   extreme cases the airships supposedly tied up to local buildings and   
   pilots came down, asking to borrow buckets of water. They sometimes   
   supposedly claimed to be from Mars.  And a whole fantasy array of   
   other things were reported to have been seen by witnesses up to and   
   including perspiring men powering the machines using pedal power.   
      
   Some researchers have written the whole thing off as total fabrication   
   by the newspapers of the time. One theory goes that reporters of the   
   1890s wrote a bunch of rubbish to fill the pages on the understanding   
   with the readership that most of what they read was for entertainment   
   only, and didn't pretend to make sense.   
      
   But some paranormal researchers are not so sure. Maybe under all the yellow   
   journalism "airship" was   
   just the mental model people in the late 19th cent had for unusual   
   things flying around the sky in much the same way as people in the   
   1940s talked about "phantom rockets" crashing into lakes in   
   Scandinavia.   
      
   Or maybe it's all a big jolly joke perpetrated by person or non-persons   
   currently unknown to science.   
      
   Or maybe it's a big test. Can organised   
   science handle complex things they can't drag back to a lab and stick   
   in a hard vacuum and prod and poke at will? Can scientists have enough   
   faith in what the mass of the population says it's seeing to actually   
   look seriously at whatever it is? Well, of course, we know the answer   
   to that one. :)   
      
   Well my contact sent me some data and I promised to throw it to the   
   AI programs to see what they made of it. I was pretty much   
   suspecting the data would be too little and too noisy to get anything   
   out of.  "Noisy" is a technical term for what they publish in   
   newspapers even now. Maybe moreso in the 1890s.   
      
   The data I have looks like:   
      
   Year Mon Day #sightings   
   1890 1 22 1   
   1890 8 21 1   
   1890 11 24 1   
   1891 9 5 1   
   1891 9 6 1   
   1891 9 7 2   
   1891 9 8 1   
   1891 9 9 3   
   1891 9 10 1   
   1891 9 11 1   
   1891 9 12 1   
   1891 9 19 1   
   1891 11 26 1   
   1892 1 18 1   
   1892 3 4 1   
   1892 3 26 4   
   1892 3 28 2   
   1892 3 31 2   
   1892 5 18 1   
   1892 6 10 1   
   1892 9 10 1   
   1893 7 3 1   
   1893 12 31 1   
   1894 9 29 1   
   1895 9 4 1   
   1895 9 6 3   
   1895 9 7 1   
   1895 9 10 1   
   1895 11 16 1   
   1896 11 18 1   
   1896 11 19 1   
   1896 11 22 1   
   1896 11 29 1   
   1897 2 2 1   
   1897 4 10 1   
   1897 4 13 1   
   1897 4 16 1   
   1897 4 19 2   
      
   Fresh off the production line the AI's say the data has exactly the   
   same hallmarks as modern UFO data. From the numbers supplied -- from   
   newspaper and magazine reports from various countries in the 1890s --   
   they find predictive models based on the position of key planets at   
   least predict a good chunk of them month to month.  The number of   
   sightings in any given month in the 1890s is predicted to within +-.4   
   given the stddev of the sightings is almost +-.6 per month. I.e. the   
   prediction is "skillful".  (Below we multiply the per month rates by   
   12 to get per annum rates that are in a better range for some s/w to   
   manipulate).   
      
   Moreover, they can come up with a planet that is "most likely" linked   
   to the airship phenomenon, and we know it well from prev posts --   
   Neptune.  While all the outer planets are linked in some way to the   
   rise and fall of "airship" sightings over the period 1890-1900,   
   Neptune predicts the data the best. Pretty much just like modern UFO   
   activity.   
      
   The demonstration is long and probably boring, but an outline goes like this.   
      
   For the given dataset create predictive models based on the numbers for   
   each planet in turn. In this exercise only 2 types of data were looked   
   at -- the distance of the planet from the sun and the speed it was   
   moving with respect to the sun for each month between 1890 and 1900.   
      
   To build the predictive model we use a validation technique.   
   I.e. part of the data is used to estimate model parameters (here we   
   build mostly linear models like "y=a+b*x" where x is a planetary number   
   and y is the monthly airship sightings count -- so the parameters are   
   "a" and "b"). Later we use the "withheld" part of the data to   
   calculate the error that model predicts without having see those   
   numbers before. We try some different combinations of factors to try   
   to push the model building program to its limits to see if it will   
   break.  Given it doesn't fall we can -- at the end -- assemble all the   
   successful models and decide e.g. which planet best estimated the   
   airship counts in the months it was not allowed to see when it was   
   building the models.   
      
   It all sounds incredibly tedious, right? Yea. Well I'm a data   
   scientist and that's my mother you're talking about!   
      
   But the summary data shows for each planet and all combination of   
   factors how well the "average model" involving just that planet was   
   able to estimate the airship activity for the "unseen" part of the   
   data.   
      
   Planet	Avg error in unseen part of data   
   neptune 5.05608   
   jupiter 5.10431   
   pluto 	5.9523   
   saturn 	6.099   
   uranus 	6.21247   
      
   For comparison, the stderr for airships sightings for the period was   
   around 7.  I.e. if we just use the "average number of sightings" to   
   guess how many sightings will happen next month then we would be out   
   on average around 7 sightings + or -. But the average model using   
   Neptune's position and speed with respect to the sun is only out +-5   
   per month -- a considerable improvement on using the average (aka   
   "just guessing").   
      
   This simple evidence is enough to convince us -- OK, some of us :) --   
   that Neptune was "most likely" involved in 19th cent "phantom airship"   
   sightings and that in many cases (when not actual fabrications by the   
   newspapers or "witnesses") actually corresponded to what people   
   would call UFO's now.   
      
   --   
   We are not afraid to entrust the American people with unpleasant facts,   
   foreign ideas, alien philosophies, and competitive values. For a nation that   
   is afraid to let its people judge the truth and falsehood in an open market   
   is a nation that is afraid of its people.   
   -- JFK   
      
   Physics Thinktank Proposes Method for Detecting Extraterrestrial Spacecraft   
   Using Gravitational Waves   
   The Debrief, 16 Dec 2022   
   An international team of scientists has written a paper showing how to   
      
   [continued in next message]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)