Newsgroups: alt.ufo.reports
From: kymhorsell@gmail.com
Subject: the annual "who knows something about UFOs" poll

[uploaded 23 times; last 28/09/2024]

Any data is grist to the mill.  As some of you that hang around my
website's BB know I follow what goes on at the site fairly
closely. Well... not me... a program.  And at this time of year I like
to look back to see which type of web accesses best predicted UFO
activity in the days following.  Does *someone* out there know
something?

And as in some past years when we cared to look... Yes, someone does
know something. Apparently.

This year I put into the mix-master a bunch of keywords that people
come to my website have followed. The site has been on the air in one
form or another for 40 years. I like to think it hasn't changed in all
that time. :)

In past years we've looked at keywords that appear in the relevant
URL's for the web pages accessed. But today we'll add in bits of the
remote host that is doing the accessing. Will any particular topics
and/or hostnames stand out from the crowd?

The procedure this time will be to use a validated regression of the
"lasso" type. Basically a lasso regression uses some normal cost
function like the R2 or sum-of-errors between the predictions and
observations, but adds to that some term that is based on the
coefficients to be estimated. The pure lasso uses the sum of the abs
values of the coefficients as an "extra term" on the function to be
minimised/maximised (R2 are normally maximised and RMSE minimised).

This has the startling property that "most" of the coefficients become
set to 0 and the remaining coefficients represent the "most important"
variables in the set the procedure was given to match up with the
target variable.

"Validation" is an extra wrinkle on top of that. The idea is to try to
make the final model be a "prediction" rather than simply a
"parroting" of the data it was given. We probably have all heard of
GIGO -- garbage in garbage out. Validation tries to ensure a
statistical model actually learns something about the data rather than
just parroting it back to you. We take part of the data and set it
aside.  The rest of the data goes to the relevant (in this case
"lasso") procedure then after the coefficients are estimated from that
we check to see how well the estimated model predicts the data in the
part the procedure never got to see. If it performs "just as well" we
have something interesting.  A model has learned to "generalise" the
data it was given so it applies to situations it never got to see to
start with. This gives us some confidence if it sees yet more and
newer data it might do equally well at predicting that, too.

So I set a program running with the 30 keywords and hostname
syllables of interest and asked it to find the "best" validated model
allowing for the model to match web refs today with UFO sightings
today, upto web refs today against UFO sightings in 5 days time.  How
far into the future could the web refs predict? If it all?

It chugged away for a few mins and came up with the "best".  It turned
out to be best in 2 different ways. It minimised the relevant RMSE as
well as maximised the R2 ("explanation power").  And -- surprise -- it
turned out to be a model that used web refs today against UFO
sightings in 3 days time. It found a model that shows "someone" out
there knows something about the short-term future of UFO activity.

The UFO community always suspected some part of the world's military
is playing around with advanced aircraft that might be mistaken for
something "strange". So if it makes you feel better, pretend that is
what we are looking at. Some others might like to remember that models
that simulate simple UFO piloting strategies between dozens of asteroids
*also* predict UFO sightings. Not just 3 days into the future, but as
far into the future as predictions of the asteroid motions can be
calculated with any accuracy (sometimes decades).

Anway. The best model we find this year based on 2023 UFO sightings is:

~ ~ ~ snip ~ ~ ~

 REWEIGHTED LEAST SQUARES BASED ON THE LMS
 *****************************************
     VARIABLE     COEFFICIENT    STAND. ERROR     T - VALUE     P - VALUE
   ----------------------------------------------------------------------
QUANTUM           5.01783         1.46663       3.42134       0.00099
yandex            3.18755         0.45075       7.07164       0.00000
bot              -0.00164         0.00548      -0.29994       0.76501<-
google           -0.23670         0.22975      -1.03029       0.30602
radiation         0.00306         0.00943       0.32420       0.74664<-
crime            -0.05297         0.04745      -1.11638       0.26764
CONSTANT         -4.52730         2.67134      -1.69477       0.09406
 WEIGHTED SUM OF SQUARES =       960.85211
 DEGREES OF FREEDOM      =        79
 SCALE ESTIMATE          =         3.48750
 COEFFICIENT OF DETERMINATION (R SQUARED) =        0.47936
 THE F-VALUE =       12.123 (WITH   6 AND   79 DF)   P - VALUE = 0.00000
 THERE ARE    86 POINTS WITH NON-ZERO WEIGHT. 
 AVERAGE WEIGHT          =         0.81905
 RMSE = 5.10728

OBSERVED           ESTIMATED            RESIDUAL   NO  RES/SC WEIGHT
 vadjcal             vadjcal

23.00000             9.48788            13.51212    1    3.87   0.0
 3.00000             9.48490            -6.48490    3   -1.86   1.0
 8.00000             8.65320            -0.65320    5   -0.19   1.0
 7.00000             8.72210            -1.72210    7   -0.49   1.0
14.00000             8.17667             5.82333    9    1.67   1.0
18.00000            13.28845             4.71155   11    1.35   1.0
11.00000            13.13952            -2.13952   13   -0.61   1.0
16.00000             6.22572             9.77428   15    2.80   0.0
 9.00000             9.44399            -0.44399   17   -0.13   1.0
12.00000             9.40139             2.59861   19    0.75   1.0
 4.00000             5.06874            -1.06874   21   -0.31   1.0
16.00000             9.35234             6.64766   23    1.91   1.0
 7.00000             8.48458            -1.48458   25   -0.43   1.0
10.00000             8.29299             1.70701   27    0.49   1.0
18.00000             7.54917            10.45083   29    3.00   0.0
 5.00000             5.30933            -0.30933   31   -0.09   1.0
15.00000            13.91042             1.08958   33    0.31   1.0
 8.00000            10.84875            -2.84875   35   -0.82   1.0
15.00000            26.78539           -11.78539   37   -3.38   0.0
13.00000            14.06286            -1.06286   39   -0.30   1.0
14.00000            10.86259             3.13741   41    0.90   1.0
32.00000            26.76885             5.23115   43    1.50   1.0
11.00000             9.52851             1.47149   45    0.42   1.0
 9.00000             9.49382            -0.49382   47   -0.14   1.0
15.00000             9.52675             5.47325   49    1.57   1.0
11.00000            15.03689            -4.03689   51   -1.16   1.0
 8.00000            20.48075           -12.48075   53   -3.58   0.0
 9.00000            26.80557           -17.80557   55   -5.11   0.0
11.00000            26.79644           -15.79644   57   -4.53   0.0
23.00000            23.67085            -0.67085   59   -0.19   1.0
 7.00000             9.53198            -2.53198   61   -0.73   1.0
11.00000             9.79902             1.20098   63    0.34   1.0
 8.00000             9.56605            -1.56605   65   -0.45   1.0
13.00000            13.23491            -0.23491   67   -0.07   1.0
 5.00000             9.21333            -4.21333   69   -1.21   1.0
 5.00000             7.53249            -2.53249   71   -0.73   1.0
 8.00000             7.36132             0.63868   73    0.18   1.0
11.00000             9.37054             1.62946   75    0.47   1.0
 7.00000             7.62354            -0.62354   77   -0.18   1.0
 5.00000             7.16594            -2.16594   79   -0.62   1.0
13.00000             8.16533             4.83467   81    1.39   1.0
 4.00000             6.33881            -2.33881   83   -0.67   1.0
 3.00000            27.96825           -24.96825   84   -7.16   0.0
15.00000             9.08101             5.91899   85    1.70   1.0
 2.00000             9.16837            -7.16837   86   -2.06   1.0
 6.00000             9.21703            -3.21703   88   -0.92   1.0
 8.00000             9.38363            -1.38363   90   -0.40   1.0
11.00000             9.36889             1.63111   92    0.47   1.0
 6.00000             8.37077            -2.37077   94   -0.68   1.0
10.00000             9.29080             0.70920   96    0.20   1.0
10.00000             9.32882             0.67118   98    0.19   1.0
12.00000             9.31875             2.68125  100    0.77   1.0
12.00000             8.98838             3.01162  102    0.86   1.0
 9.00000             9.40790            -0.40790  104   -0.12   1.0

~ ~ ~end snip ~ ~ ~

I've included every 2nd line from the "prediction" part of the output
to show it does seem to be predicting stuff in 3 days time.  Perhaps
the biggest standout you notice right away is one day toward the end
of the data where the model predicts "27.96" UFO sightings in 3 days
time when, in fact, only 3 happened. FAIL! Well. Not really.  In 4
days time there were an unusual 15 UFO sightings in a period were 2,3
or 4 per day was more usual. It was "off by 1 day" we might think.

Regardless of this, the model correctly predicts UFO sightings in 3
days time within about 5 on average. This also sounds unremarkable
except "just guessing" would get to within 10 on average.  So the
model is "twice as good as guessing". You also have to consider the
validation procedure. This is a model that was trained on only the
first 6m of data in 2023. It was flying blind with the 2nd 1/2.  But
it *still* got a result of twice as good as guessing.

As for the leading keywords that seem to be able to predict UFO
sightings 3d in the future. "QUANTUM" is an old page based on various
quantum computer simulations I've done over the past 20y.  Most of
these were unpublished. At least one was published under someone
else's name. :) For some reason "someone" interested in quantum
physics or QC knows something about UFO activity in the future.  As
does the 2nd item in the list -- "yandex". Russia's favourite search
engine. Of course, anyone or anything could probably tap into that.
It is not necessarily the Russian govt, military or ministry of science.

A couple of the remaining keywords have only minimal statistical
significance (marked with a "<-"). And, anyway, some have a -ve sign
in their coefficient.  Somehow MORE of these types of queries align
with FEWER UFO sightings in 3days time. Some kind of "anti knowledge". :)
(In data science we look for good guessers or very bad guessers -- as
long as you can figure which is which they both can be used to predict
something correctly).

Overall, the F-value for the model shows it could not do as well as it
does by chance alone. The probability it is all just lucky coincidence
given it's based on the day-by-day data for 2023 is virtually 0.

So it seems someone out there knows something. They also interested in
quantum physics and use a Russian search engine to search for things.

Some crowd that might be using google to do the same thing has
"anti knowledge" -- the more active they are the fewer UFO sightings
are recorded 3d in the future.

Make of that what you will. ;)

--
Science is facts; just as houses are made of stones, so is science made
of facts; but a pile of stones is not a house and a collection of facts
is not necessarily science.
-- Henri Poincare

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

Climate crimes must be brought to justice
UNESCO 26 June 2023
Climate denial has increased the risk of catastrophic global change. Should
international criminal law be used against those who promote ...

There is something there -- measurable light, multiple instruments -- and
yet it seems to move in directions inconsistent with what we know of physics
or science more broadly. And that, to me, poses questions of tremendous
interest, as well as potential national security significance.
-- Rep. Adam Schiff, D-Calif., 2022 House Intelligence Committee hearing
   on UAPs.

[No Good Deed Goes Unpunished:]
Whether you work in the UFO warehouse at Area 52, are the surgeon who
handles the alien autopsies, or are the designer of the amazing
climate cleaning machine, if your work is classified, you can't blow
the whistle on it for the public good and expect the law to work in
your favor.
-- David W Brown, "How to Blow the Whistle if You Work With Flying
   Saucers and Their Alien Pilots". A letter from Clearance Jobs, an
   organization representing govt workers with high security clearances.

A vast array of our most sophisticated sensors, including space-based
platforms, have been utilized by different agencies, typically in
triplicate, to observe and accurately identify the out-of-this-world
nature, performance, and design of these anomalous machines, which are
then determined not to be of earthly origin.
-- Jonathan Grey, NASIC intel officer, Wright Patterson AFB, 06 Jun 2023

[Secret UFO recovery program blown open:]
I hope this revelation serves as an ontological shock sociologically
and provides a generally uniting issue for nations of the world to
re-assess their priorities.
-- David Grusch, 05 Jun 2023
[Talking to Les Kean et al for The Debrief, Grusch called for an end to
nearly a century of global UFO secrecy and warned that humanity needed to
prepare itself for "an unexpected, non-human intelligence contact scenario"].

The US govt portrays itself as the world's preeminent
superpower, so to acknowledge that there are things in their
airspace, whatever they are, that are faster and more manoeuvrable
and run rings around fast jets doesn't play very well.
So there's the embarrassment factor, and maybe a little bit of
fear that either an adversary has made a quantum leap in
development, which has left the US in a poor second place, or, as
some believe, this really is extra terrestrial, in which case we're
not at the top of the food chain anymore.
-- Nick Pope, 02 May 2023

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
detect extraterrestrial spacecraft using gravitational waves.
[The reason LIGO hasn't been looking for "warp signatures"?
Nobody thought of it].