Lying with Statistics and Algorithms

“Word for word… got that?”

House Republican Caucus

      

“Well, now WE are in power… and if we did what we should be doing, it [ending the Senate filibuster rule] would IMMEDIATELY end this ridiculous, Country destroying ‘SHUT DOWN.’ If the Democrats ever came back into power, which would be made easier for them if the Republicans are not using the Great Strength and Policies made available to us by ending the Filibuster, the Democrats will exercise their rights, and it will be done in the first day they take office, regardless of whether or not we do it. In addition to all of the other things we would get, such as the best Judges, the best U.S. Attorneys, the best of everything, this was a concept from years ago of then President Barack Hussein Obama and former Majority Leader Harry Reid in order to take advantage of the Republicans.”” 

Trump on Truth Social

Lying with Statistics and Algorithms

There is a strategy to effective lying, especially in the political world. Sure, politicians are not viewed as the most truthful people on Earth, but when there is a toxic major political consistency to the lies, where fitting in with your peer group requires accepting a distorted parallel universe, that “A house divided cannot stand” Lincolnian admonition becomes a true threat to democracy. Add demonization/dehumanization and blame to the mix, and democracy goes on “life support.” We have two branches of government (legislative and judicial) failing to uphold their obligations to provide constitutional guardrails against executive excess and a President who has stated that ending the Senate filibuster rule would instantly allow a simple majority vote both to end the shutdown and to enable Republicans to pass 100% of their agenda without opposition. Project 2025 could become law.

The basis for many voters is their perception of the state of the economic, political and cultural world around them. And as the world increases in both uncertainty and complexity, there has been a tendency for people to outsource their perceptions or simply latch on to what they perceive as a credible leader, searching for certainty and hope. Autocracy is an excellent form of government, efficient and quick, but only as long as the autocrat is right… which almost never remains true. Democracy allows the people to vote when a leader seems to be making too many mistakes. Repeat a lie enough, and it often begins to be treated as a fact. It’s called gaslighting.

Still effective leaders generally make decisions that they can justify with reference to “facts,” but when the facts go south, insecure leaders can fire the purveyors of facts, create statistical lies to support their position, and most will select statistical models that suggest how successful they are and how wrong their opponents have been. The greatest statistical manipulative tools are to use averages and metrics which change based on press releases. I have railed about this in the past, but today, I will cite a November 3rd FastCopmpany.com an overview by noted economist and former Comptroller of the Currency, Gene Ludwig:

Americans keep hearing that the economy is strong. Unemployment is low. Wages are rising. Growth is steady. But for millions of families, those headlines feel like a cruel joke. The cost of rent, groceries, and healthcare keep climbing while steady, well-paid work remains out of reach. The disconnect isn’t just perception—it’s baked into the way we measure economic success.

Throughout history, when governments fail to fully appreciate the realities faced by their people, it leads to crisis. The United States may be on the brink of such economic and societal unrest. The unrest that led to the French Revolution and the economic imbalances preceding the Great Depression are both cases in point.

In the late eighteenth century, the oppressive economic situation facing the French people went unacknowledged by the royal family for decades. The French ruling class considered the truth about the nation’s fiscal crisis to be nefarious—a threat to their power. Marie Antoinette, when told the peasants had no bread, replied, “Let them eat cake!” Whether or not the remark is literal or legend, it captures the ruling class’s indifference. Soon after, the Revolution erupted, bringing turmoil and suffering to French citizens of every rank and station.

The same narrative arc applied a century and a half later when the Great Depression loomed. In both instances, economic data that could have set off alarm bells was available—more accurate figures that would have revealed the risks emerging—and this perspective might have prompted action that could have softened the blow, if not avoided the crises altogether. But the data was either confusing, confounded with other contrary data, or affirmatively hidden. The effects were catastrophic…

The unemployment statistics our government releases monthly are misleading. If someone is looking for full-time employment but finds nothing except a single hour of work in a week, they are considered “employed” in the eyes of the government. For purposes of official government statistics, this one-hour employee is in the same category as someone secure in a full-time job.

This logic extends to wages. Someone who works full- or part-time for a salary that falls below the poverty line (around $25,000 a year for a three-person household) is classified the same way as someone earning $1 million every month... The government reports on “median wages” every quarter. The idea behind their metric is simple and straightforward: If you line up all full-time employees in order of their weekly earnings, the person directly in the middle earns the median wage.

But this statistic only considers the wages of people who are currently employed full-time, overlooking millions of part-time workers and unemployed job seekers. So, the moment a low-wage factory worker receives a pink slip, her salary is deleted from the sample altogether. The moment a farm worker’s seasonal employment ends, his salary is similarly deleted.

OK, firing the head of the Bureau of Labor Statistics and using the stock market as a metric of good economic policy fall under this mantra of statistical manipulation. But there is another manipulative technique that has only become more pervasive and destructive with the advent of artificial intelligence: manipulative algorithms (effectively built-in software filtration programs). And, as Enrique Dans, also writing for that issue of FastCompany.com, explains, Elon Musk is the master of this distortive use of algorithms to replace Wikipedia with his own version:

“This is not Musk’s first experiment with truth engineering. His social network, X, routinely modifies visibility and prioritization algorithms to favor narratives that align with his worldview. Now Grokipedia extends that project into the realm of structured knowledge. It uses the language of authority, such as entries, citations, and summaries, to give bias the texture of objectivity.

“This is precisely the danger I warned about in an earlier Fast Company article: the black-box problem. When AI systems are opaque and centralized, we can no longer tell whether an output reflects evidence or intention. With Grokipedia, Musk has fused the two: a black box with a bullhorn... It is not that the platform is wrong on every fact. It is that we cannot know which facts have been filtered, reweighted, or rewritten, or according to what criteria. Or worse, we can have the intuition that the whole thing starts with a set of commands that completely editorialize everything. The line between knowledge and narrative dissolves.” We would do better if Americans would do meaningful research, looking at different sources before accepting the dictates of the supremely biased… but…

I’m Peter Dekom, and asking average voters to engage in true research is unrealistic, but outsourcing to an unreliable or highly biased source, or one where algorithms guarantee distortion, is, simply both naïve and stupid… the resulting decisions can often be profoundly negative on one’s personal life.