- The Grand Tour of the Voyagers seems to closely explain UFO
sightings reported over the last 1/2 of the 20th C.
- As probes approached some planets there was an increase in UFO activity
reported on Earth. After they passed reported activity declined again.
- One planet showed exactly the opposite pattern. Like someone was
hiding. See if you can guess which one.
- If some UFO reports were caused by the "buzz" created by TV coverage
of up-coming close approach, this last pattern is odd.
- The positions of the 2 probes with respect to their various
planetary appointments seems to explain about 99% of the UFO data in
the late 20th cent.
There seems to be some disappointment with the public version of the
It might be just me, but some of the official explanations sounded
kinda strained. It's a psychological burden to answer questions you
don't want to answer but have to appear to be cooperating to answer so
you don't cop some punishment.
Some of the brass have undergone interrogation training and some of
the testimony looked like it had clicked into action. In a galaxy far
far away I was a prison shrink here in SE Aus. Some of my own schoolin
clicked into action watching some of the video from the proceedings.
But what I want to talk about here is a reprise of some analysis I
posted elsewhere a couple years back that showed the Grand Tour of the
solar system in the 1970s maybe stirred up some people somewhere.
Maybe enough to have them e.g. start following carrier groups off socal
around like lost puppies.
Here's the setup. The US launched 2 probes to take advantage of some
fortuitous planetary alignments coming up in the late 20th cent.
Voyager 1 and 2 both visited Jupiter and Saturn and "Vijay" went on to
scoot past Uranus and Neptune later.
Now tell me if you heard this, but if -- I said if -- a certain
phenomenon was at least partly based on one or other of these targets,
would we her on lil ol Earth ever get to see any evidence of it?
As I might have pointed out the principles of Datascience and
Forensic Science are similar. In Forensics one of the major
principles is if a suspect visits a crime scene then they leave
evidence at the scene and the scene leaves evidence on them. The
datascience version is -- if 2 datasets share the same spacetime then
information from one is imprinted on the other.
In both cases, finding the fingerprint of one thing on the other is
evidence the 2 things are connected.
So here we setup a little regression problem. For each planet we'll
have a countdown for contact by one or other of the probes as the
calendar rolls on from 1970-2000. For Jup and Sat there are 2 probes
approaching and departing. We have to handle that. So well just sum
the 2 countdowns. After each probe passes the countdown changes from
a +ve number to a -ve number.
If we try to product UFO activity observed on Earth, can we get a very
good match between how the V1 and V2 probes are doing on their tour
and monthly UFO activity reported to the NUFORC (I use this dataset a
lot because of the organisation's dedication to transparency).
The dataset is long so I'll just summarize some lines from the input
file to give anyone that might try to replicate it a leg up to
figuring out what I've done.
Date Jup Sat Ur Nep log(#UFOs)
0 18.6667 22.4167 16 19.5833 5.13391
1 18.5 22.25 15.9167 19.5 3.74762
2 18.3333 22.0833 15.8333 19.4167 4.44077
3 18.1667 21.9167 15.75 19.3333 4.44077
4 18 21.75 15.6667 19.25 4.84623
5 17.8333 21.5833 15.5833 19.1667 7.35854
115 -0.5 3.25 6.41667 10 6.31257
116 -0.666667 3.08333 6.33333 9.91667 6.58083
117 -0.833333 2.91667 6.25 9.83333 6.14552
118 -1 2.75 6.16667 9.75 6.05021
119 -1.16667 2.58333 6.08333 9.66667 4.44077
120 -1.33333 2.41667 6 9.58333 5.53938
594 -80.3333 -76.5833 -33.5 -29.9167 6.31897
595 -80.5 -76.75 -33.5833 -30 6.06379
596 -80.6667 -76.9167 -33.6667 -30.0833 6.35089
597 -80.8333 -77.0833 -33.75 -30.1667 6.49072
598 -81 -77.25 -33.8333 -30.25 6.41836
599 -81.1667 -77.4167 -33.9167 -30.3333 6.48464
The first col is the number of months since Jan 1970. The planet
columns are the sum of the number of months each probe will take to
reach that planet. When the probe passes its count becomes a -ve
number of months. The last column is the log of the number of UFO's
reported to the NUFORC that month. In the 1970s almost all reports
were within the US but we wont try here to deal with that. We'll just
assume the report number is a noisy indicator of UFO activity on Earth.
Now the interesting part.
Is there a linear model of the above that accurately predicts the
last column from the other columns.
Well I wouldn't be posting this unless I knew there was. :)
The model is as follows:
REWEIGHTED LEAST SQUARES BASED ON THE LMS
VARIABLE COEFFICIENT STAND. ERROR T - VALUE P - VALUE
Jup x1 -5170.64990 2068.51392 -2.49969 0.01272
Sat x2 7147.74365 2714.64722 2.63303 0.00870
Uran x3 -3832.30859 1314.56030 -2.91528 0.00370
Nep x4 -121.88642 749.23901 -0.16268 0.87083
WEIGHTED SUM OF SQUARES = 213.64821
DEGREES OF FREEDOM = 549
SCALE ESTIMATE = 0.62383
COEFFICIENT OF DETERMINATION (R SQUARED) = 0.98979
THE F-VALUE = 13303.979 (WITH 4 AND 549 DF) P - VALUE = 0.00000
THERE ARE 553 POINTS WITH NON-ZERO WEIGHT.
AVERAGE WEIGHT = 0.92785
Basically, the program I'm using here (PROGRESS -- you'll need an
archaeology degree to look that up) says there is a very very good
model linking the countdowns for each prob and the number of UFO sightings.
The "scale estimate" if French for how well the model predicts monthly
UFO reports. Remember we're using the LOG of the monthly count so it
translates into exp(.62) = 1.9. IOW the model claims to predict
monthly UFO reports within -2 to +2 of the historical value. Pretty close.
The robustness of the model is given by the F statistic. The greater
the value the more robust it is. Normally an F if 2,3,4 is great. The
F here is 13,000! I.e. there is almost no chance there is no link
between the positions of V1 and V2 and UFO reports.
The R SQUARES says about 99% of the UFO report numbers see to vary in
response to changes in the planetary countdowns.
The T-VALUE and P-VALUE of each coefficient shows that NEPTUNE maybe
not so important, but the others are 99% involved in predicting UFO
activity as seen on Earth.
We see the coefficient for most planets is -ve. Meaning as each probe
is approaching and the countdowns for V1 and V2 (added together for
Jup and Sat) are positive and decreasing there is very close to no
effect on UFO activity, but as the countdowns approach 0 then UFO
reports spike up.
"For some reason" the coeff for SATURN is exactly the opposite. When
the countdowns are +ve and the probs are approaching *then* UFO
reports are high. But as the probes get closer and closer UFO activity
on Earth suddenly drops.
The is exactly the opposite we might expect from a population being
bombarded with news of space probes approaching Saturn. We might have
expected that as news programs tell people Voyager 1 (or 2) is about
to fly by Saturn, everyone would run outside and look at Saturn and
start dreaming and maybe mistake Saturn or a fly on the wall for a UFO
and make a report.
So it seems there is something different about Saturn. But Jup, Sat,
and Uranus are all involved. Neptune not so much.
In a certain branch of philosophy it's argued that if something is
known by one person then it must be assumed it is known by everyone.
So if we "know" this fact based on 50 year old data then we have to
assume the back room boys at the Pentagon -- even if they have no
other evidence whatsoever -- know this as well.
Who knows, it might have been part of the reason V1 and V2 were funded
Click here to read the complete article