Do not trust the political polls, inside or outside of the margin of error.
You know that old saying, “Fool me once, shame on you; fool me twice, shame on me”.
Is the United States of America going to be truly shamed in the 2020 presidential election?
We have not fully reconciled with the cyberwar that influenced the 2016 presidential election,
According to most news sources, real ones like the major networks and newspapers, the Republican administration has not taken significant steps to protect the American election system from interference by our adversaries. I can feel the shame growing already.
We certainly know a great deal about what happened in 2016, the perpetrators, platforms, the mechanics of digital fake news. A recent article from NPR is a good summary.
NPR What You Need To Know About Foreign Interference And The 2020 Election September 1, 20197:00 AM ET
But the real problem is not the extent of the evidence, but the lack of insight into the subtleties of how information cyberwar, what I call information terrorism, has been applied to the political environment.
There are three basic elements of political cyberwar, and they all need of updating, regulation and education, and education may be the most difficult to achieve.
1. All political polls use commonly accepted survey practices relating to sample size, methods of contact, and question-wording. And all political polls blithely present the results with the caveat that the results are to be used within the margin of error. There are known flaws in all types of surveys, and in particular political polls, but I want to address one issue that is common to all surveys and polls: we are being fooled by nominal standards.
A large sample size does not confer reliability. Proclaiming the margin of error does not excuse sloppy question writing or intentionally leading questions. We can’t regulate surveys and polls for every sector of business, politics or social issues. But we can set reasonable standards for a new class of polls and surveys that meet clear, consistent and comparable standards of design and execution. The polling practices may still be flawed, such as the inability to interview people who don’t want to be interviewed, but the flaw should be common to all political polls.
Starting standards would include:
· Minimum sample size of at least 1500 people.
· Specific contact method design to be used and considered adequate to reach potential respondents by telephone, in person or online. All standard polls would have to use the same mix of contact methods. Each method has its biases, but all surveys should be comparable with the same biases.
· Standard wording of critical political questions. Creating clear, benign questions about political preferences is definitely not rocket science. But attempts to create clever and guided questions to achieve an outcome has become a science.
2. Providers of political news and polls should enroll in a new accreditation process to use agreed-upon standards, provide complete transparency of their methods, and open to periodic audits. Accreditation of political polls and news would include a specific designation, i.e. a seal of approval, that the organization is recognized as a real source, not a fake or front, and has agreed to follow certain standards. Our society has accreditation and oversight processes from airplanes we fly to the chicken we eat (most of the time it seems to work). We don’t usually trust providers to act solely in the best interest of the customer. Why do we let news of any type, and particularly political news, which affects every aspect of our life, including the eventual regulations about chickens, operate without any oversight?
3. The real problem with cyberwar and politics is that there is still a failure to recognize the cleverness and leverage of the actual tools of cyber propaganda and information terrorism. For decades we have been worried about enemies, internal and external, hacking into voting machines to change votes or eliminate voters from the rolls. But that seems so primitive now, because physical hacking is detectable. The real problem is the recognition by our adversaries of how to change the outcome of an election by highly targeted micro-communications, often fake, hidden within the enormous range of public news. Our political adversaries do not have to hack into the machines, they do not have to eliminate 30% of the electorate, they don’t have to change the minds of 40% of the population. Our adversaries understand the reality of our political polarization and the extensive range of conflicts that are easily inflamed by hyped-up real and fake news. And the friend of fake news and cyber war is our electoral college system. Here are the key points:
a. The great majority of American voters are set their preferences and won’t be swayed by real or fake news. There are various estimates, but for the sake of this discussion, I’ll accept that approximately 40% of the electorate is committed to either the Democrats or the Republicans, which includes the “leaning “people, so that’s 80%. At least 20% claimed to be independent, undecided or uncertain up until the day they vote. In the 2016 election, adversaries did not have to change the minds or votes of 20% of the people, they only had to change the minds and votes of less than 2% of the people in a few key areas that ultimately decided the electoral college votes. You know the states — Michigan, Wisconsin and Pennsylvania — all decided by less than 1% difference between winner and loser of the electoral votes of the states.
How hard is it to change the minds of 1% or 2% of the population in critical areas, especially when there is a natural polarization, significant preference for third-party candidates, which can be up to 10%, and everybody believes that there is a natural margin of error of +/- 3.5%? That is a lot of cumulative range and margin of error to work with. So how does this combination of statistical weaknesses operate within the polls?
b. Change the minds of 20 people, change the course of history? We’ll stay with the assumption that 80% of the population is committed to their political preferences, regardless of whether or not they recognize or are comfortable with the fake news. So, the fake news, or just the hyper-exaggerated legitimate but divisiveness news, is intended for the roughly 20% who are the most susceptible to their self-induced uncertainty, low information process, and the so-called lemming effect and peer influence.
To oversimplify, if some people who are uncertain and undecided tend to go with popular opinion, as presented in the polls, we have our prime market for political cyber-manipulation. Very simply, in a typical political poll of 1000 people, if cyber propaganda can influence political intentions of just 20 people, which is 2% of the poll sample, than a close contest can go from being 51% versus 49% to a shift of 49% versus 51%, shifts accepted because they are within the margin of error of 3% – 4%! If adversaries can nudge just a few people in the desired direction, just a fraction of a percent in a few polls, we suddenly have a new “trend” that can influence a group of people to vote based on false information.
We need standards for political polling; we need an accreditation process to help identify legitimate from obviously illegitimate information sources. But I don’t know what we can do about people who prefer to remain uncertain, prefer to be low information unless we do more to help them figure things out.
One of the reactions that I encountered during the writing of my book was the defensiveness of people who thought I was saying that American adversaries are smarter than the United States in cyberwar, digital technology etc. etc. That was how my proposition was sometimes interpreted before I explained the evolution of the current state of cyberwar. The late 1990s when the Internet began its evolution into the pervasive, familiar communication vehicle of our lives, was the pivotal point in the cyberwar of today.
Before the Internet, the primary American adversaries, Russia and China, were at least a generation behind us economically, militarily and technologically. But on Day 1 of the Internet, every country, every government, was starting from the same point – there was no prior history of digital communications’ experience, everybody was starting with essentially the same skill set, or lack of skills to be more accurate.
What has happened since then is obvious in retrospect, but I saw it developing during 1990s when I was working on some of the early Internet-based research systems. At the very beginning of the Internet age, Russia, China, and other countries recognized the power of digital platforms to further expand their control with digital propaganda, to reach everyone quickly, cheaply and 24/7.
Our adversaries recognized the capabilities of the Internet and weaponized digital communication. What happened in the United States and Western countries?
We put most of our efforts into commercializing the Internet. American capitalism developed Facebook, Snapchat, Instagram, we’ve got mail.
So, the first important accounting in the new digital war happened in the 2016 presidential election and clearly illustrates the difference between weaponizing and commercializing.
In 2016, Russia and China controlled the outcome of the presidential election. Unfortunately, the metaphorical mail we are now receiving is a uniquely clever form of warfare, for which we seem to have no way to combat until we learn how to be digital warriors.
The full diagnosis of the history of the US vs its adversaries and the war we started but might be losing is in my new book, The Margin Of Error War on Amazon.
September 23, 2019