From: https://blackboxvoting.org/fraction-magic-5/
FRACTION MAGIC — PART 5
MASTERS OF THE UNIVERSE
By Bev HarrisMay 12, 2016COMPANIES, ELECTIONS INDUSTRY, MAKE AND MODEL, RESEARCH
5 – PRECISION CONTROL TO WEIGHT RESULTS –
Using the GEMS “double” configuration, users can control election results using amazingly accurate vote percentages. We demonstrate this by assigning vote percentages using a children’s letter-number cipher. A=1, B=2, C=3, etc. For example, H=8 and A=1 so 81.8181% is “hahaha.” We helped Bush/Cheney dominate Alaska’s Cheney Lakes, assigning 38.514525% of the votes, spelling CHENEY.
We used a simple Excel spreadsheet to plan desired results for each location, then injected allocations into GEMS. Smith, a programmer, used a USB stick [Flash Memory Disk] containing a Visual Basic script to interact with GEMS. For Harris, who has no programming skills, Smith created a demonstration utility to read GEMS, display races, candidates and vote subgroups, then inject percentages typed into it to GEMS.
Using exactly the same utility, Harris weighted results in Glades County, Florida against Mark Foley (U.S. Representative) and Charlie Crist (Florida Attorney General) races. In Marin County, California she controlled Measure F to defeat a private corporation named Veolia’s bid to take over a water project, this time using only the absentee votes. She weighted the 2010 Shelby County, Tennessee sheriff’s race to give it to Randy Wade.
Votes everywhere were instantly redistributed to match the planned result with no need to reprogram for different elections, races, precincts, candidates or voting methods. This demonstrates that the fractional vote capability in GEMS enables any private contractor to weight elections in different jurisdictions, even if he lacks programming skills. Wherever he can get the contract he can set the result.
Before:
After:
Planning sheet:
Result:
Planning sheet:
Result:
We used letter-number ciphers only to demonstrate that our control over results is precise and unarguable. Any percentage can be created and you can insert your formula into some or all of the detailed vote subgroups.
Why aren’t percentages exactly the same?
Percentages achieved are remarkably, though not perfectly precise. Variations in the number of total votes can create slight variations in the votes once you reach the hundredth place.
Errors
Use of an inadequate number of decimals will produce errors. An error of ± 1 in vote count addition can be a round-off error, as shown in the following example:
Vote cast = 10
Candidate A receives 2.5 votes which rounds to 3
Candidate B receives 7.5 votes which rounds to 8
3 + 8 = 11 votes, for a discrepancy of one.
Using three or more decimals helps reduce round-off errors:
Vote cast = 10
Candidate A receives 2.501 votes which rounds off to 3
Candidate B receives 7.499 votes which rounds off to 7
3 + 7 = 10 votes, producing no round-off error.
Because round-off errors will only occur with a minority of decimal values in the data, absence of an error will be the more common result. Round-off errors can be positive or negative and some will cancel others out. Lack of error does not mean the underlying vote total is a whole number.
Loss of value can be introduced when rounding downward. For example, 0.3245 may round off to zero, losing one-third of a vote. Loss of value accumulates when using thousands of vote decimals, increasing the chance that small, telltale errors may appear in certain parts of the data. Therefore it is important for election officials to provide interim results in real time, detailing all components of the vote total (total cast, blank, under, over, and candidate) and committing to a breakout by precinct and vote type (absentee, early, at polls), publishing updates frequently.
Because the decimalization feature in GEMS is so powerful, it is possible for programmers to write scripts that reduce or remove errors, but such code typically involves assigning vote fragments into candidate or undervote pockets. Thus, the more immediate the public commitment of detail results, the more difficult it becomes to find a place to assign fragments.
Use of middlemen and private contractors to report results makes it easier to hide round-off and lost value errors.
Public commitment of immediate, detailed, directly sourced vote results makes it somewhat more challenging to alter results successfully. However, current election administration trends are traveling in the opposite direction.
Won’t you see the decimals if you look in the Microsoft Access tables?
The tool needed to remove all floating points (decimal values) completely and permanently also exists in GEMS. Without going into further detail, any Fraction Magic programmer worth his salt will include clean-up commands in the script. We chose not to do so because, as you will see in a later section, we will be demonstrating other concepts by looking at the decimals we have retained.