Data manipulation, the new frontier in power usurpation.

You should usually default to the assumption that the mean remains the mean and the norm remains the norm.

But these are not normal times.  Clearly something was amiss in the 2020 election.  Given that the majority of the electorate assume so, it can't simply be dismissed as fringe conspiracy theorizing.  Were the bad practices sufficient to swing the election?  I don't know.  With the astonishing expansion of mail-in ballots, which have historically always had high fraud rates for states which used them, it is easy to anticipate that there was greater than usual.  But enough?

One of the compelling things to me all along has been that Biden won in the popular vote, significantly outperforming Clinton but he had no coattails.  Down ballot Democrats did poorly. 

But the big question is what might have been the mechanisms of fraud?  Was it just mail-in ballots or was it mail-in ballots combined with the fraud ridden practice of ballot harvesting?  Was it ballot system program tampering?  Hard to tell.

From Millions of WIMP Votes Shifted Nationwide by Ray Blehar sheds some light on how it might have happened.  Basically, most the margin of Biden's victory is covered by dramatic shifts in the results from shifting write-in and minor party ballots through ballot adjudication.  His analysis is not conclusive but is highly suggestive.  Interesting to read.  

Simultaneously, there are the first whiffs of monkey business in the Census numbers.  Just as with Blehar's analysis of voting, the issue is not so much about the absolute counts per se but about how different the ratios are from the past.  From Why Did Biden Census Bureau Add 2.5 Million More Residents to Blue-State Population Count? by Stephen Moore.

The Census determines not only how many House seats a state receives but a great deal of federal money is allocated based on headcount.  The more population you can document, the greater your share of money.  It is quite lucrative to overestimate the number of people.

There are all sorts of mechanisms to hold over-counting in check, but there are a lot of pressures to do so as well.  

There is something very fishy about the new 2020 Census Bureau data determining which states picked up seats and which states lost seats.

Most all of the revisions to the original estimates have moved in one direction: Population gains were added to blue states, and population losses were subtracted from red states. The December revisions in population estimates under the Biden Census Bureau added some 2.5 million blue-state residents and subtracted more than 500,000 red-state residents. These population estimates determine how many electoral votes each state receives for presidential elections and the number of congressional seats in each state.

Interesting but not compelling.  If red-states were on growth spree, is it not reasonable to assume that small increments of overestimation year by year in a continued pattern of growth might have generated a material over-estimate which needed to be corrected?  Sure, that is conceivable.  Likewise with blue states with a decade of decline overestimating that decline (though this is somewhat less probable as the incentives are so against errors in this direction) incrementally year by year over a decade.

Adjustments are not unexpected.  Adjustments in a uniform direction and at magnitudes greater than in the past do, however, raise some red flags.

The original projections for the Census reapportionment had New York losing two seats, Rhode Island losing a seat and Illinois perhaps losing two seats. Instead, New York and Illinois only lost one seat, and Rhode Island lost no seats. Meanwhile, Texas was expected to gain three seats, Florida two seats and Arizona one seat. Instead, Texas gained only two seats, Florida only one and Arizona none.

[snip]

The evidence is now only circumstantial, but when errors or revisions are almost all only in one direction, the alarm bells appropriately go off.

Here are some of the strange outcomes in the Census revisions just released:

No. 1: New York -- We've been tracking the annual population/migration changes between states since the last census in 2010. Over the past decade, New York LOST about 1.3 million residents on net to other states. (This does not include immigration, births and deaths.) Still, this is a population loss that is the equivalent of two, maybe three, lost congressional seats. But the final numbers ADDED approximately 860,000. That's roughly twice the population of Buffalo and Rochester -- combined. This is the state that has lost by far the largest population over the past decade.

No. 2: Many deep-blue states had 2020 census numbers significantly revised upward from their December estimates: Connecticut, Hawaii, Illinois, Massachusetts, New Jersey, New York, Rhode Island and Vermont.

No. 3: Many red states had 2020 census numbers lower than their 2020 estimates: Arizona, North Carolina and South Carolina.

No. 4: Going back to the 2010 Census, the final head count in every state was within 0.4% of the original estimate, and 30 of them were within 0.2%. This time around, 19 states were more than 1% off, 7 were more than 2% off, NY was more than 3.8% off, and NJ was more than 4.5% off.

No. 5: Virtually every one of the large deviations from the estimates favored Democrats. Just five states in the 2020 census were within the same margin (0.41%) that all states were within from the 2010 census.

Maybe the 2010 estimates were abnormally accurate, or maybe the 2020 estimates were abnormally inaccurate. The Census Bureau needs to tell Congress why these revisions under former President Barack Obama were so much larger than normal and so weighted in one direction: toward the blue states. 

 To me, point number 4 is the most suggestive.  Proof?  No.  Highly suggestive?  Yes.

This past year we have had three, perhaps four major data analytics disasters on the part of the government in which the results favor a particular blue worldview.  While not conclusive of nefarious intent or practice, the more incidents there are like these, the more the bonds of trust are eroded.  

• Covid-19 deaths - Poorly defined, poorly tracked, poorly communicated from the beginning.  Stripping away all the ambiguities and we only have a ballpark number of how many people avoidably died from Covid-19.  Adding in people who died with Covid but not from Covid has been one problem.  Including those who were already so laden with comorbidities as to dictate their death within six months has been another.  On and on.  The definition and tracking of Covid-19 has been the antithesis of a scientific approach to understanding the world.
• 2020 Election - see above about all the doubts raised by both the known activities in some states as well as the clear deviance of outcomes from the past.
• 2020 Census - see above for the surprisingly high error rates and the surprising homogeneity of the directional adjustments.
• Police shootings - It is reasonably thoroughly documented that police shootings and civilian deaths are proportional to crime commission and do not deviate by race.  And yet the year has been dominated by claims and riots based solely on the denial of this factual knowledge.

As a complex modern economy and sophisticate structure, we are dependent on accurate, timely, pertinent data collection.  If we are entering an era when data manipulation or obfuscation is just one more front in political battles, we are in trouble.  We have to have a degree of trustworthiness to which it appears the Mandarin and chattering classes are no longer committed.