Kevin Laland, an evolutionary biologist with a focus on animal behavior, in his book Darwin's Unfinished Symphony, shows that a key tool in nature is the ability that many animals have to learn from each other by copying behavior. He shows that what we may call "culture" arises in many species, not just in humans. In the resulting societies, animals that master strategic copying fare better than those that don't.
Humans not only copy the behavior of those around us (we certainly do that), but we also learn from people whom we have never met and may even no longer be alive. Laland points out that learning is enormously magnified, compared to lower animals, by the emergence of language, writing, teaching, and the cultural practices surrounding these. He argues that it is primarily this cultural magnification that distinguishes humans from other animals and enables a society comprising billions of individuals overwhelming the planet.
We have built a huge cultural infrastructure around such learning. Most people in the developed world devote most of the first 20 years of their life to such learning, postponing even procreation, the biological evolutionary imperative. Our institutions, schools, universities, libraries, magazines, newspapers, television, blogs, and social media are all centered on the conveyance of information between people who have never met.
For a society to work, it has to be built on shared information and shared beliefs. We all believe in the objectively meaningless pieces of paper that we call money, for example. The idea of money is a shared belief, and a society at our current scale could not possibly work without such shared beliefs. Most aspects of our lives depend on shared cultural values that are learned through our information exchange mechanisms.
Until recently, spreading information was relatively costly, requiring investment in printing presses, distribution channels, television and radio broadcasting rights and equipment, etc. As a result, most sources of widespread information were institutional, and the institutions built reputations, good and bad, and business models around those reputations. As a society, we have developed a largely shared value system that trusts some sources of information over others.
Over the last few years, we are witness to huge explosion in information exchange. Technology has democratized publication and broadcasting to the point that any individual can reach every individual. The cost of distributing information globally has dropped to (essentially) zero, and the number of sources of information has exploded. In this world, it seems that trust in certain sources of information becomes more critical. More than ever, we need a respected press, respected thought leaders such as scientists, and political leaders we can trust. Sadly, this is not the direction we are headed in.
On July 16, 2017, I was watching MSNBC's The Rachel Maddow Show, a daily news show that regularly airs strong critiques of the Trump administration. Maddow revealed in this show an event that truly shocked me and made me realize just how devasting the Trumpian revolution may become. Maddow talked about a classified document that was "leaked" to her via her website that revealed some potentially devastating dirt against Trump and his cronies. She did not reveal what that dirt was because she was convinced that the document was a clever forgery designed to trick her into airing a news story that could then later be exposed as "fake news."
Donald Trump's persistent message is that the media and other authority figures in our culture (scientists, academics, and political leaders) cannot be trusted. Had Maddow bitten the bait, her show would be over, and we would have lost a respected (by me, at least) source of information about our world. I have no idea whether this really was a trap and whether Trump or his supporters were involved in it, but one thing I am sure of: Trump would have approved of this trick had it worked. Trump is in favor of discrediting the media in any way possible, thereby systematically driving humanity back towards the cultural mechanisms of lower animals.
Trump himself, the elected President of the United States, overtly lies. He has been unequivocally caught many times, and he sloughs it off. He makes himself the prime example of the political leaders that cannot be trusted. Trump supporters often state explicitly that they do not trust information from the media, which may explain how they can continue to support him in the face of so many revelations about how unfit he is to be president.
For the media to be effective in a democracy, it must convey information. But what is information? In my own book, Plato and the Nerd, I review the modern notion of information, due to Claude Shannon, where the amount of information in a message depends on the likelihood of whatever is described by the message. A message like "the sun rose yesterday" contains very little information because it is very likely (certain, even) that the sun rose yesterday, whereas a message like "the sun was eclipsed by the moon yesterday" contains quite a bit of information because the event is more rare.
I also examine what it means to "learn" from a message. What is the information conveyed, vs. the information contained in a message? I describe an interpretation, credited to the 18-th century English Reverend Thomas Bayes, where the information conveyed by a message depends on our subjective prior assumptions about the likelihood of what is described by the message. More critically, what we learn from a message depends on the extent to which the message changes our prior assumptions. If our assumptions are unchanged, the message conveys no information, no matter how much information the message contains.
Trump is systematically setting out to ensure that messages from the media and from authorities such as scientists and presidents convey no information. If we do not trust the source, our prior assumptions will be unchanged by any message. If you like a little math, below I give a mathematical model that shows how this works. A democracy cannot function without effective mechanisms for conveying information.
According to Shannon's theory, every conveyance of information is flawed except when both the source and destination of the information is a digital machine. For digital machines, however, the information conveyed has no meaning. For information to have meaning, it must be subject to human interpretation, and such interpretation cannot be perfect. Shared information that builds culture, therefore, must ultimately be based on trust, and Trump is undermining that trust.
The Israeli historian Yuval Noah Harari, in his book Sapiens: A Brief History of Humankind, traces the origins of writing and argues that the ability to construct societies that are more than tiny tribes depends critically on written communication. Trump's broad attack on societal trust in such communication is subversive. If he succeeds, he will throw us back to the tribal world of illiterate apes. The only information we will trust is that we directly obtain from the observed behavior of those around us. If your neighbor believes that earth is 4000 years old, you will too. If your neighbor believes that climate change is a hoax, you will too. Anything written about this by strangers, proclaimed by political leaders, or reported in Science will be dismissed as not trustworthy.
I claim that Trump's attack on information is a bigger threat to The American Way than anything that ISIS can muster even in their wildest dreams. What??? How can this be as bad as terrorism? A terrorist attack, by definition, is one where the intended effect goes well beyond the physical victims. The 9/11 attacks killed many people, but that was not their purpose. Their purpose was to undermine our culture, and indeed, they were somewhat successful. The whole experience of air travel, for example, has changed dramatically. And we were drawn into incredibly costly and long wars in Iraq and Afghanistan. But on the whole, they really were not all that effective. American culture is, if anything, more materialistic, inequitable, and discriminatory against Muslims than it was before 9/11.
The American philosopher Daniel Dennett, in his book From Bacteria to Bach and Back, talks about how human communication augments the power of a fundamentally limited human brain.
[H]uman brains have become equipped with add-ons, thinking tools by the thousands, that multiply our brains' cognitive powers by many orders of magnitude. Language, as we have seen, is the key invention, and it expands our individual cognitive powers by providing a medium for uniting them with all the cognitive powers of every clever human being who has ever thought. The smartest chimpanzee never gets to compare notes with other chimpanzees in her group, let alone the millions of chimpanzees who have gone before.
Today, we have spectacularly more effective mechanisms for "comparing notes," but if we lose trust in the notes we compare, we lose this cognitive multiplier and become chimpanzees.
In Darwin's Unfinished Symphony, Laland points out that "teaching" as a practice is absent in all animals except humans. Teaching, according to Laland, has no evident evolutionary purpose, since the conveyance of knowledge from the teacher to a stranger does nothing to further propagation of the teacher's genes. Teaching is evolutionarily altruistic, and therefore inexplicable from the perspective of purely biological evolution. Under Trump, teaching is under attack, in part because of widespread distrust of intellectual authority and its advocacy of the theory of evolution. In the Trumpian revolution, teachers, such as myself and Laland, cannot be trusted. You should not believe us. Ironically, Trump is attempting to throw us back to the early days of evolution, where only biology, not knowledge, made progress.
We are facing an Information Apocalypse. Is "apocalypse" too strong a word? According to Wikipedia, which I believe has not yet been undermined by the Trumpian revolution, the word "apocalypse," from ancient Greek, literally means "an uncovering." It is a disclosure of knowledge, a revelation. In Christian tradition, from the Book of Revelation, also called the Apocalypse of John, the last book of the New Testament, the revelation which St. John receives is that of the ultimate victory of good over evil. It signifies the end of the present age, and today the term is commonly used for any prophetic revelation that leads to an end of time or to the end of the world. The Trumpian Information Apocalypse is based on the revelation that all information is untrustworthy. This is true, at some fundamental level, but we cannot build an American Way or any other cultural value without a shared commitment to common truths. Trump, more than ISIS, is undermining America.
The American historian of science James Gliek, in his book The Information, points out that very notion of "history" is inextricably tied to writing. Without a mechanism for communicating beyond our personal encounters, there can be no history. He goes further to claim that the very act of "thinking" and our notions of logic and reason are also tied to writing, as is even the notion of consciousness. When writing loses its veracity and its authority, when we lose faith in it, we lose history, thought, reason, and consciousness. This will be the end of the world as we know it. It would be an apocalypse, to be sure, but not the one of St. John. It would be a triumph of evil over good.
Is this apocalypse likely to occur? It is clear that there are forces
beyond Trump, including some Russians, who are systematically working
to undermine trustworthy sources of information. The attack on Maddow is
a harbinger. We have entered an arms race that will require
considerable investment to protect our trusted sources of information.
The first step in this race needs to be to put out of office the
co-conspirators of these attacks.
For the Mathematically Inclined
Let E represent that some particular newsworthy event occurs. Our prior assumption about this event is that it has some likelihood, given mathematically as a probability P(E). The probability is small if the event is unlikely, and larger if it is more likely. Called a "prior probability," this number lies between 0, cannot happen, and 1, surely happens (see chapter 12 of Plato and the Nerd for the real meaning of "probability").
Let M represent a news report that E has occurred. M has its own probability P(M) because a newspaper may or may not publish M. Let P(M|E), the probability of M given E, represent the probability that M occurs given that E has in fact occurred. And let P(E|M) represent the probability that E has occurred given that M occurred. Then Bayes' rule, named after the 18-th century English philosopher Revered Thomas Bayes, tells us that
P(E|M) = (P(M|E)/P(M))P(E)
P(E|M) is is called the "posterior probability" because it is the probability that E has occurred given M, that the newspaper reported that E occurred. If we trust the newspaper, then P(E|M) is a better estimate of the probability that E has occurred than our prior probability P(E). We have learned something from the newspaper.
If the newspaper always reports M if E has occurred, then P(M|E) = 1. If the newspaper also only reports M if in fact E has occurred, i.e. it never lies, then P(M) = P(E). In this case, Bayes' rule tells us that P(E|M) = P(M|E), meaning that our prior assumption P(E) is irrelevant after we read the paper. If M appears in the paper, then E occurred, and if M does not appear in the paper, then E did not occur. This is how it works if we have complete faith in the newspaper. We believe it regardless of our prior probability for event E.
We should, of course, always harbor some skepticism. Even the most trustworthy newspaper can make mistakes and can choose to not report a newsworthy event. We should therefore assume a value for P(M|E) somewhat smaller than 1 and a value for P(M) somewhat different from P(E). This is what a learned reader will do. Put some faith in the newspaper, and we are capable of learning but not completely gullible.
In a Trumpian world, however, the situation is very different. We do not trust the newspaper at all. This means that P(M|E) = P(M); the probability that the newspaper publishes M does not depend on whether E has occurred. Consequently, by Bayes' rule, P(E|M) = P(E). Reading the paper teaches us absolutely nothing about whether E has occurred. Our posterior probability is the same as our prior probability. In this Trumpian world, news reports teach us nothing and humans become incapable of learning from strangers.
Edward Ashford Lee
Edward Ashford Lee
Author of Plato and the Nerd: The Creative Partnership of Humans and Technology
Good stuff! In your opinion, what happens after the Information Apocalypse? Would that be the end of the Anthropocene and the dawn of the next era? Would AI (what ever that is) or more specifically, deep machine learning die with no information, or only very untrusted information?
Have you considered a NY Times op-ed piece?
Actually, a very scary possible outcome is that AIs will be recruited to generate fake information and flood our current sources. Think about "Artificial Trumps" or "ATs." Currently, AIs are used to protect websites like Wikipedia against vandalism, but these techniques may be ineffective against a concerted effort by AIs to inject subtly incorrect information into the system.
Neal Stephenson's science fiction novel Anathem talks about historical events where the "reticulum" (the equivalent of the Internet on planet Orth) essentially collapsed in a flood of fake information taking down much of the society that depended on it. In the novel, trustworthy records of the events were scarce because the very mechanisms for keeping records were thoroughly undermined.
Leave a Reply.