Monday, November 14, 2016

What if the polls were right and the election was wrong?

Now that Donald J. Trump is president-elect and Democrats are fighting like rabid raccoons to lay blame for the disaster, and pollsters are pulling out their hair to figure out what went wrong,  let's consider a possibility.

If thousands of engineers agreed on a bridge design and then the bridge was built and immediately collapsed, who would you suspect made a mistake? The highly trained engineers and their computers or the company that built the bridge?

The engineers checked and re-checked their design, but in practice the design proved terrible.

What's easier to imagine, that the engineers were wrong, or that perhaps the construction company used substandard concrete or steel, or took some other shortcut?

Back to the election. Virtually all of the polls, which change from week to week, were in basic agreement. 

Then the exact opposite happened.  Suppose that the polls were correct and that the election was flawed.

 We know many voters were prevented from voting and were incorrectly told they were not eligible, or were otherwise disenfranchised. And we know with a fair amount of certainty that Russia hacked into various email accounts and then gave the messages (or false messages) to Wikileaks. 

If a hacker can penetrate an email account, how difficult would it be to hack into an election computer and change some totals? The hacker would only have to alter a relatively few votes in Florida, Ohio and Pennsylvania, and maybe for good measure a few smaller states. 

I have no idea  what voting systems were used or how votes were tabulated. Where I voted in New Haven paper ballots were fed into a machine. Did anyone calibrate or check the machine at any point? I don't know.

Some voters in other states used computers to vote, which seems like an extremely bad idea. 

Meanwhile, has anyone started to compare the results with the numbers of votes cast? The number of votes could be below the number of voters, if some chose to not mark their ballots. But the total should never be greater than the number of ballots cast. 

Some enterprising reporters could check into this, unless they are otherwise occupied writing about dogs, celebrities, sports, cooking, relationships, medical miracles, fires, and other things. Sports, to paraphrase Marx or someone or other,  is the Valium of the people. 

What we need, assuming elections continue in the future,  is a system that is as uncomputerized as possible.

Who understands how computers work? Do you? Sure, you know vaguely about 1s and 0s and electricity, and microchips. and programs. But if you had to explain exactly how a computer counts paper ballots, you would not be able to.

 Few people could.  Maybe Alan Turing or John Von Neumann, or Steve Wozniak.

What we need instead is the simplest system imaginable that can still be counted in a few hours, unless people are willing to wait a month or two for results. 

Let's consider a precinct with 1,500 voters. The poll would receive 1,500 1-gram neodymium magnets. The voter is given a number of magnets equal to the number of candidates and ballot measures that are possible. 

If two people are running for mayor, for example, the voter would only need one magnet, because he is not supposed to vote for more than one candidate.

The voter drops his magnet into either hole A or hole B. 

When the poll closes, you remove the magnet bins and weigh them. If the total is more than 1,500 grams, alarm bells are sounded. Otherwise, the votes are weighed and the bins impounded. Seems simple. The most recent ballot would be more complicated. 

A voter could spit a ticket, or not vote for certain candidates, "leaving the space blank," so to speak.
Perhaps the voter is given enough magnets to vote every possible combination of votes and is instructed to place any unused magnets into another "extra magnets" hole. 

Presumably, where I voted, the paper ballots could be checked against the computerized totals. 
Magnets leave no paper trails, but then again, each magnet could weigh exactly 1.14159 grams. Different precincts could use different weights. If need be, someone can count the magnets.

Or something like that. Let's not depend on machines that work in obscure ways, and let's not depend on machines that someone can hack into.

So, who's going to start investigating the vote, and why the results were so different from the predictions?

No comments:

Post a Comment