I've been hesitant to engage the mathematical topics until adequate time has been allowed to commence auditing the ballots. However, some things must be said that may not be obvious to non-mathematicians, that inform the 'debate' over election fraud ...
-- for which take the prima facie case to have been made already.
This will be a long but informative
Thread:
This will be a long but informative
Thread:
1/ It seems an uncontested fact that some 'batches' of votes had a very high percentage of Biden vs Trump votes, with precisely zero down ballot voting in the entire batch. This is so utterly remarkable a result it essentially proves fraudulent intent with no doubt whatsoever.
Allow me to explain.
2/ I suppose 'they think we are stupid': that some batches like that must have emanated from very pro Biden districts (likely those target by receiving 100s of millions in USD, in the guise of non-profit donations to 'harvest' ballots, ...
2/ I suppose 'they think we are stupid': that some batches like that must have emanated from very pro Biden districts (likely those target by receiving 100s of millions in USD, in the guise of non-profit donations to 'harvest' ballots, ...
in an election so partisan and divided that it is not surprising that the targeted minorities *only* indicated to the social work a preference for Biden over Trump -- or whatever other guise was used for mass harvesting.
As I shall show, this is a mere cover story.
As I shall show, this is a mere cover story.
3/ In this thread, I will show that the complete absence of down voting in any large batch is utterly damning evidence of intended fraud, on the face of it. Why? Because such batches can only be intended to facilitate counting *MULTIPLE TIMES* without detection.
They are, in effect, a sort of cryptogram, and can be 'solved' by the same sorts of methods that solve any difficult cryptographic method -- Enigma, say.
4/ A highly partisan race with only two parties considered is very special mathematically. Think about what would happen if a pre-printed batch of such fraudulent votes were to be scanned *multiple times* using software that (intentionally) cannot detect duplicate ballots
being scanned twice.
If done naively, a batch of 9950 Biden votes and 50 Trump ones, scanned exactly 5 times, would yield a batch with exactly 5 times 9950 for Biden and exactly 5 times 50 for Trump. The multiples of 5 in each batch would be suspicious.
If done naively, a batch of 9950 Biden votes and 50 Trump ones, scanned exactly 5 times, would yield a batch with exactly 5 times 9950 for Biden and exactly 5 times 50 for Trump. The multiples of 5 in each batch would be suspicious.
5/ If we could see the 'scan order' we would also detect a lack of shuffling between batches, and they would come out in the exact same order. Most suspicious, of course, if all the Biden votes came first, and then the packet of Trump votes. That would show two packets
6/ So, it is necessary (of course) not to be so precise -- to 'mix' a bit of real data, or at least only partially run a batch so the counts don't come out exact integral multiples, with no 'natural variation' between the batches. Scanning errors will introduce some, but not nuff
7/ Let's suppose we try to be a bit more 'clever' with our fraud, and introduce just one downballot race -- for county treasurer say -- and that only for Biden, so always a Democrat one. Now we have three kinds of ballots -- Biden alone, Biden with treasurer, Trump alone, ...
, in say proportions a : b :c.
8/ If we run the mix exactly 5 times, we are 5a : 5b : 5c ratio, which we can reduce to its lowest terms a : b : c and see, like the empirical formula in Chemistry, the underlying 'formula' for the batch, as well as the 5 times weighting.
8/ If we run the mix exactly 5 times, we are 5a : 5b : 5c ratio, which we can reduce to its lowest terms a : b : c and see, like the empirical formula in Chemistry, the underlying 'formula' for the batch, as well as the 5 times weighting.
It is like having a verly large molecule with three elements, and having 5 such molecules.
9/ The problem now is that, if we form the hypothesis all the votes for Democratic Treasurer are also Biden votes, that correlation 'unlocks' b to c ratio for us, as well as the sum of b + c, and so we have to be careful to obscure *that* information as well, ...
by not only making sure we add a few b's and subtract a c's in our 'steganogaphy' batch.
10/ Of course, it is logically possible that someone voted for Trump *and* the Democratic Treasurer, but like 'letter frequencies' in English, that's surely a rare letter, and in real data,
10/ Of course, it is logically possible that someone voted for Trump *and* the Democratic Treasurer, but like 'letter frequencies' in English, that's surely a rare letter, and in real data,
such 'correlations' give the game away eventually. The underlying problem is that we have a set of LINEAR equations, involving N 'alphabet symbols' with base frequency a : b : c replicated by scanning M=5 times.
11/ There's a bit of a complication with races left empty -- that allows really three choices, R, D, and none of the above, and if you don't know for a fact there are never split tickets (in the real world, it happens all the time, esp the headline national race vs local ones,
not to mention third parties) -- but that simply adds a bit of complexity to the problem, the root of which is ENTROPY.
12/ The thing that the Biden/Trump ONLY ballots have, as a batch, is far too little entropy. Measuring the entropy of each batch is key.
12/ The thing that the Biden/Trump ONLY ballots have, as a batch, is far too little entropy. Measuring the entropy of each batch is key.
Another analogy is queries in a search engine. There are 'head queries' which are very frequent, and have very low entropy -- 'ebay', 'amazon', and 'fb login' are surely among the commonest queries Google sees -- they are in the heads.
13/ In contrast, there is a 'long tail' of really goofy queries that are highly improbable in the first place and Google will never see again in its entire lifespan as a service. These have very, very low entropy.
In our election, these may be ballots with 75% of the down ballot races filed in, all for R, but one sole race in a minor office filled on for a third party or D -- just as likely to be a scanning error (noise) as a correctly scanned rare preference.
14/ In between are the *normal* queries that Google can't anticipate (so can't precompute the answer), but neither are they so rare as to be hopeless for the algorithm. In our election, these are *normal* ballots.
Normal data looks normal. It has normal frequencies and correlations.
15/ If we think about the R races, with two choices, then there are 2^R POSSIBLE PREPRINTED ballots -- if you allow some races to be 'skipped' that puts in a third answer, so 3^R.
15/ If we think about the R races, with two choices, then there are 2^R POSSIBLE PREPRINTED ballots -- if you allow some races to be 'skipped' that puts in a third answer, so 3^R.
These 2 or 3 to the R possibilities form the 'alphabet' for the race -- or the number of different print runs for our fraudster.
16/ Not only is it *EXPENSIVE* to print large runs for each letter of the 'alphabet', since even for 10 races [most real world ballots have more] ...
16/ Not only is it *EXPENSIVE* to print large runs for each letter of the 'alphabet', since even for 10 races [most real world ballots have more] ...
you have 2^10 = 1024 print runs (or 3^10 = 59049 different ballots if you allow blanks races) and most of those are 'rares' like 'alternates R and D down the line' and 'only voted for Biden and the State Auditor' and goofy things like that.
Many more combinations than are *likely* even in a real world data set with 10 races.
17/ but even 'just a few' -- 50 say -- distinct ballots *and no others* would leave a very suspicious signature.
17/ but even 'just a few' -- 50 say -- distinct ballots *and no others* would leave a very suspicious signature.
18/ That is why, if you are going to tell a Lie, you should tell a BIG ONE and repeat it often -- *NO ONE*, out of 10,000, bothered to vote down ballot. Oh, and Biden got Stalinesque numbers in the 90th+ percentile for frequency.