Wednesday, January 06, 2021

Explaining Natural Selection

#evolution #adaptation .

Nature has two largely independent processes: one to produce variation (using mutation and/or sexual reproduction), and one to eliminate unfit variations (known as natural selection). What we observe at any timepoint in natural history is the emergent consequences of the interaction between these two processes. Over millions of years, overall biological complexity has increased, but a vast number of evolutionary paths have ended.

The long-term outcome of natural selection is often referred to as the Survival of the Fittest, the term introduced by Herbert Spencer, but this is at best a massive simplification, if not actually tautological. Darwin happily adopted Spencer's term in the later editions of his book, but Wallace argued that Elimination of the Unfit was a more accurate descriptor.

The actual mechanism of natural selection works in terms of reproductive advantage. Fit individuals (whatever that means) are likely to survive and pass on their genes to the next generation, while unfit individuals won't. Over time, the standards of fitness may gradually shift.

For example, if predators attack a herd of antelopes, the slowest are likely to be caught, while the faster ones will escape. The predators that are not fast enough to catch even the slowest antelopes will starve. Over many generations, thanks to natural selection, the average speed of both the antelopes and the predators will increase. (We're assuming here that the environment remains more or less the same. Obviously if the sea level rises and the plains get flooded, then the rules of the game change, and different variants will start to have the advantage.)

In order to survive, antelopes and predators don't need to be the fastest of their species, they just need to be fast enough. So we shouldn't take survival of the fittest literally - what this is really about is the survival of the fit.

And as Bateson argued, this term has a double meaning. This is not just about the survival of fit individuals or variants, it is about maintaining the relationship (the fit) between the two species.


During the Covid-19 pandemic, people have become interested in (and sometimes confused by) the way natural selection works among viruses. Professor Rickards spotted a misleading metaphor from the UK Government's Chief Scientific Advisor.

And Magnus Nordborg recently challenged a statement made in the Economist.

I commented that Natural selection is not about the survival of the fittest but the non-survival of the unfit. A deadly virus might not survive for ever, but don't hold your breath. @gcochran99 insisted that I was mistaken, and added

I think this is missing the point, for several reasons. Everything will change/mutate many times before anything ends up anywhere. While there seems to be one particular variant that is currently reproducing itself faster than other variants, new variants may well emerge with significantly different properties. Transmissibility depends on conditions that are not constant and at least partly under human control - including social distancing and wearing masks, as well as vaccines and other medical interventions.

And what I'm mainly arguing against is the idea that we only need to worry about the most efficient viruses because the less efficient ones will eventually disappear. Yes they might, but they might kill a lot of people and animals first. I made a similar point in my post on Viral Pandemic (April 2005). See also Arguments from Nature (December 2010).

In its small way, Gregory Cochran's tweet has gone viral, receiving more Likes than the rest of the thread put together. But that surely doesn't mean his tweet will end up being the one that matters.

P. den Boer, Natural Selection or the Non-survival of the Non-fit (Acta Biotheor 47, 83–97, 1999) 

Dean Keith Simonton, Creative thought as blind-variation and selective-retention: Combinatorial models of exceptional creativity (Physics of Life Reviews 7 (2010) 156–179 - paper recommended by @Tudortweet who writes: worth linking to the theory of creativity as another metaphor

Charles H. Smith, Natural selection: A concept in need of some evolution? (Complexity, Volume17, Issue3 January/February 2012) 


Updated 26 January 2021

Thursday, December 17, 2020

Trolls are like ghosts

On the one hand, trolling messages contribute no meaningful content, being merely tediously predictable responses to certain situations. But on the other hand they are designed to provoke a certain effect - to harass and intimidate.

@adriandaub makes the interesting suggestion that trolls are like ghosts (WTCT p96). Or perhaps automatons.

An aggrieved white guy who has set up an alert for when Sarah Jeong tweets and then huddles over his phone to make some claim about racism and Roseanne using jagged grammar and vertiginous logic is functionally indistinguishable from a bot having been set up to do the same thing.
WTCT p94

Although he complains that practitioners of NLP (Neuro-Linguistic Programming) take the notion of programming literally (WTCT pp 144-5), the automatic response described by Professor Daub seems to involve a form of conditioning that might be regarded as functionally indistinguishable from programming. (As a computer scientist myself, it is not for me to argue with a professor of literature whether it is the practitioners of NLP or its critics who take the notion of programming literally.)

As I think I've stated elsewhere, I regard NLP as a syncretic collection of interesting ideas (strongly influenced by Bateson and others) and dubious snakeoil. Although the snakeoil elements are generally regarded as pseudoscience, I wouldn't want to lose the ideas. Daub mentions two important ones, which Bandler and Grinder didn't invent but did much to popularize - reframing and feedback.

In communication there are no mistakes - everything is feedback
WTCT p 145
Many years ago, I invoked a similar idea (the meaning of a communication is its effect) in a discussion on the signal/noise ratio with the blogger Ernie the Attorney, who had complained that What we have here is a failure to communicate.

In his chapter on Communication, drawing on earlier theorists including John Durham Peters, Daub argues not only that communication often falls short of its potential, but sometimes occupies a space of preordained, deliberately engineered disappointment (WTCT p 89). It's as if the troll actively wants to be misunderstood.

Or even to cease to be a subject. Daub mentions Sontag's interpretation of Freud: human aggression frequently flows from an unconscious desire to become inanimate (WTCT p95).

As I pointed out in an earlier post (November 2018), many of the speech acts that pollute the internet are not propositions but other rhetorical gestures. And even if the trolling message appears to be coded as a proposition, the metacommunication is otherwise. In his 2019 article for Logic Magazine, Peters notes that the aim of trolling is to goad someone else into getting upset, an act known as triggering, and describes the outgoing US president as an absolute master at metacommunicative messing. And of course framing/reframing.

Furthermore, the troll's targets often include the medium itself, as the cultural theorist Mark Fisher once observed.

The elementary Troll gesture is the disavowal of cyberspace itself. In a typical gesture of flailing impotence that nevertheless has effects — of energy-drain and demoralisation — the Troll spends a great deal of time on the web saying how debased, how unsophisticated, the web is.

Andrew Iliadis explains the information theory of Gilbert Simondon in terms that can be linked to the notion of reframing: 
Information is that that which, depending on the way that it comes into contact with another abstraction of itself, unlocks or clicks into another form of reality.

Reality, fiction functioning as truth, or just lulz?

To be a game, the participants have to agree on the frame that this is play. The anthropologist Gregory Bateson once made this point brilliantly. Hazing rituals, he said, were governed by the frame is this play? Trolls like to claim the prerogative to define an interaction as play when their conduct makes that frame completely unclear.
Peters 2019


The play's the thing, someone once suggested, wherein to catch the conscience of the King. But what if the king has no conscience, no soul?

Adrian Daub, What Tech Calls Thinking (Farrar Straus and Giroux, 2020) 

Mark Fisher, Fans, Vampires, Trolls, Masters (k-punk, 12 June 2009) 

Erving Goffman, Frame Analysis: An Essay on the Organization of Experience (Boston: Northeastern Press, 1974)

Andrew Iliadis, A New Individuation: Deleuze's Simondon Connection (MediaTropes Vol IV, No 1, 2013)

John Durham Peters, Speaking into the Air (Chicago University Press, 1999) 

John Durham Peters, U-Mad (Logic Issue 6, 1 January 2019)

Susan Sontag, Fascinating Fascism (New York Review of Books, 6 February 1975)

Related posts: Failure to Communicate (July 2004), Ethical Communication in a Digital Age (November 2018)

Wednesday, November 11, 2020

Dark Data and the US Election

The 2020 US Election wasn't the first time that many people misread a political situation, and I'm sure it won't be the last. What I want to look at in this post is the way that these misreadings were a consequence of missing data - commonly known as Dark Data.

The first misreading was the polls predicting a strong result for Joe Biden and the Democratic Party. Although Biden still had a majority of the popular vote, and also a majority in the Electoral College, the margin of victory was much smaller than most polls had predicted.

A plausible explanation for this error is that the data collection on which these estimates are based systematically excludes certain types of voter, and therefore underestimated the support for Donald Trump and the Republican party. Either because these voters are less easy than others to reach by traditional polling methods (telephone calls), or because these voters are less willing than others to reveal their true voting intentions (the so-called Shy Voter).

Following errors predicting the 2016 election, polling organizations thought they had worked out how to correct these errors. It seems that they were wrong about that.

The second potential misreading involved analysing the demographic breakdown of voters based on exit polls, resulting in statements such as Donald Trump increased his share of the XYZ-category vote from A% to B%.. But given that Trump voters were more likely to vote in person and Biden voters were more likely to vote by mail, surveys of people leaving polling stations would be highly skewed.

The third misreading was perhaps to take social media too seriously. Ed Pilkington argues that although Trump's strongest advantage going into the election was the economy, Trump proved incapable of keeping to the economic message.

Because his modus operandi is to stimulate a positive response from his fans, as well as outrage or scorn from his enemies. He then leverages the negative response from his enemies to reinforce the loyalty of his fans. (As Judith Butler argues, this effect is linked to shame.)

As I wrote in my earlier post 

What is special in Trump's case is that there are some feedback loops that strongly reinforce these particular behaviour patterns, because they have produced the desired outcomes in the recent past. Trump's worldview (Weltanschauung) causes him to pick up certain signals and ignore others.

Trump has a remarkable ability to create noise on social media, and he has often used this noise to his own advantage. He had a refined sense of what would play well to his core audience, and would rattle his core opponents, because those were the two categories that reacted on social media to his every move. But he appears to have had rather less sense of what mattered to those who didn't belong to either of these categories, and didn't broadcast their partisan views at every opportunity.


Judith Butler, Is the show finally over for Donald Trump? (The Guardian, 5 November 2020)

Ed Pilkington, Loser: Donald Trump derided defeat – now he must live with it (The Guardian, 11 November 2020)

Zack Stanton, People Are Going To Be Shocked: Return of the Shy Trump Voter? (Politico, 29 October 2020)

Related posts: Why this stupid behaviour? (October 2017), Dark Data (February 2020)

Saturday, November 07, 2020

Vain Repetition

At the time of writing this post, the US election is not quite over. The mainstream media (now including the Murdoch empire) are presenting the strong likelihood that the Biden-Harris ticket will turn out to have won, but President Trump and his loyal supporters appear optimistic of snatching victory from the jaws of defeat, with the aid of legal arguments, aggressive protest outside the vote counting stations, and of course prayer.

In this post, I want to talk about a prayer session on the Thursday after the election led by Paula White-Cain, Trump's controversial spiritual advisor.

A number of people have offered musical interpretations and mashups. @pjgrisar of @jdforward saw parallels with Steve Reich's 1965 composition It's Gonna Rain, which used a tape recording of an apocalyptic Pentecostal street preacher called Brother Walter.

Meanwhile, many people took to social media to remind Mrs White of something Jesus said in the Sermon on the Mount: But when ye pray, use not vain repetitions, as the heathen do: for they think that they shall be heard for their much speaking. (Matthew 6:7).

Regular readers of this blog may not be surprised to learn that this verse has been subject to many different translations and interpretations, going back at least as far as Martin Luther. (The Greek version features the curious word battalogein, but Jesus's original words were probably in Aramaic.)

Theologians sometimes argue that there is no problem with repetition as such, the problem is with the vanity of the repetition. Steven Winiarski argues that repetition becomes vain when it is used with bad motives. Bad motives for repetition include any attempt to use music and repetition to elicit a purely emotional response, to gain a personal audience, or to manipulate God.

So what exactly was the purpose and intended effect of Mrs White's incantation?

P.J. Grisar, Paula White’s wild Trump sermon is begging for the Steve Reich treatment (Forward, 5 November 2020)

Tom McCarthy, Rupert Murdoch-owned US outlets turn on Trump, urging him to act with 'grace' (The Guardian, 7 November 2020)

Seren Morris, Paula White's Trump Prayers Go Viral on Twitter, Inspire Memes and Remixes (Newsweek, 5 November 2020)

Nicholas Till, Joy in Repetition: Critical genealogies of musical minimalism (Performance Research 20:5, 2015)

Steven Winiarski, Music, Culture, and Vain Repetition: Matthew 6 in its Context (Artistic Theologian 4, 5 April 2016)

Wikipedia: It's Gonna Rain, Language of Jesus, Matthew 6:7, Paula White

Related post: Worshipping the Golden Calf (October 2008)

Wednesday, October 28, 2020

The Purpose of Slogans

Some people may interpret the Black Lives Matter slogan as implying that racial injustice is greater than other forms of injustice. Those who experience other forms of injustice are encouraged to resent or even resist the Black Lives Matter slogan. Thus campaigners against different forms of injustice are divided rather than united.

No doubt this effect is most welcome to those who don't want any of these campaigns against injustice to succeed.

Ten years ago on this blog I discussed a talk by Amartya Sen on Reducing Global Injustice, in which Sen expressed his opposition to all kinds of injustice and refused to single out any one kind of injustice as greater than other kinds - whether gender or race or ethnocentrism or whatever. He argued that different forms of injustice were simultaneously incomparable and interconnected.

I was and remain convinced by this argument.

But I don't think it follows from this that campaigns against one form of injustice are necessarily wrong, simply because they ignore other forms of injustice. It may be tempting to counter "Black Lives Matter" by saying "But what about disadvantaged white lives" - but this whataboutery plays into the hands of those who would deny the existence of racial injustice altogether. 

I believe that those who suffer a particular form of injustice should be free to campaign against this form of injustice, without being obliged to simultaneously campaign against all other forms of injustice. Of course there may always be risks in such campaigns: risks of being misunderstood, risks of being deliberately misrepresented, risks of being divided from those who ought to be on the same side, risks of what Mark Fisher called snarky resentment, even risks of physical attack. But if the recommended alternative is to keep one's head down and suffer in silence, many will think that the fight against injustice justifies taking these risks. 


Mark Fisher, Exiting the Vampire Castle (OpenDemocracy, 24 November 2013)

Wikipedia: Whataboutism

Related posts: Global Injustice and Moral Challenge (July 2010)

Saturday, June 27, 2020

Discourse Wars

According to @kevinroose, one of the most popular posts on Facebook yesterday was a story about a Vietnam War memorial supposedly vandalized by BLM protesters. Except that the photo was from 2016, the vandals weren't anything to do with BLM, and the story had been debunked weeks ago.

Someone called Jeff responded on Twitter, claiming that Kevin and others were missing the point. Yes the picture is incorrect but doesn't subtract from all the other memorials the BLM movement did vandalize. And someone called James argued that BLM is still a group that doesn't give a damn about America and that is trying to destroy everything that it stands for. So it wouldn't have surprised me in the least if they had done that since it's in keeping with who they are and how they think.

Jeff and James don't seem bothered by the fact that the story is fabricated, because in their eyes it confirms some deeper truth. A truth in which one memorial is the same as another memorial, in which an act in 2016 described by locals as ignorant, having no sense of history is conflated with a series of acts motivated by an acute sense of historical injustice, and in which any attack on the symbols of this historical injustice counts as destroying everything America stands for.

Michel Foucault argued that each society had its regime of truth.
The possibility exists for fiction to function in truth, for a fictional discourse to induce effects of truth, and for bringing it about that a true discourse engenders or manufactures something that does not as yet exist. Foucault

Michel Foucault, “The History of Sexuality,” in Power/Knowledge: Selected Interviews and Other Writings 1972–77, ed. Colin Gordon, trans. Colin Gordon, Leo Marshall, JohnMepham, Kate Soper (New York: Pantheon, 1980), p. 193

Daniel Funke, Vietnam memorial was not vandalized by Black Lives Matter protesters (PolitiFact, 4 June 2020)

Daniele Lorenzini, What is a Regime of Truth? (Le Foucaldien, 1(1), 1. 2015)

Friday, June 28, 2019

The Habitual Vice of Epistemology

In his recent polemic on #ClimateChange, Bruno Latour writes

But these rational sorts are just as caught up as the others in the tangles of disinformation. They do not see that it is useless to be indignant that people believe in alternative facts, when they themselves live in an alternative world, a world in which climate mutation occurs, while it does not in the world of their opponents.

It is not a matter of learning how to repair cognitive deficiencies, but rather of how to live in the same world, share the same culture, face up to the same stakes, perceive a landscape that can be explored in concert. Here we find the habitual vice of epistemology, which consists in attributing to intellectual deficits something that is quite simply a deficit in shared practice.
Down to Earth, p25

Bernard Harcourt adds
A true fact cannot stand on its own, autonomously, independent of social relations, of who tells it, or finds it, or proves it—and where and how it is established. There may well be facticity, but in order for facts to stick, they have to properly form part of social life. And when our shared social life has been scarred by betrayal and exploitation, it will no longer be fertile ground for the trust necessary to maintain truths.

Bernard E. Harcourt, Bruno Latour on Truth and Praxis (Critique and Praxis 13/13, 16 December 2018)

Bruno Latour, Down to Earth: Politics in the New Climatic Regime (Polity Press, 2018).

Wednesday, June 26, 2019

False Sense of Security

According to the dictionary, a false sense of security is a feeling of being safer than one really is. Apparently that's a bad thing.

Peter Sandman is a strong believer in what he calls precaution advocacy - to arouse some healthy outrage and use it to mobilize people to take precautions or demand precautions. He has helped environmental groups arouse public concern about the need for recycling, the dangers of factory emissions, etc. In such contexts, his concern is that people are disregarding or underestimating some category of risk, and he is urging the introduction of appropriate precautions - whether individual or collective.

There are countless risk and security experts who take a similar position - for example, advocating greater diligence in corporate security, especially cybersecurity.

However, as Dr Sandman acknowledges, the notion of a false sense of security is often used rhetorically, suggesting that a given regulation or other precaution is not only unnecessary but even counter-productive, making people careless or complacent. This argument is sometimes based on the notion of risk homeostasis or risk compensation - that people adjust their behaviour to maintain a comfortable level of risk. The classic example is people with seatbelts and airbags driving faster and more recklessly.

Dr Sandman notes that the rhetoric can sometimes be deployed by both sides of an argument - for example "gun controls create a false sense of security" versus "guns create a false sense of security". What this suggests is that the rhetoric is often about other people - the implication is that We have a true sense of security, but They would be misled.

The notion of a false sense of security also arises in connection with security theatre - a performance that may have little real impact on security, but is intended to reassure people that Something Is Being Done. When Bruce Schneier introduced this term in his 2003 book, he regarded security theatre as fraudulent, and believed it was always a Bad Thing. However, he later came to acknowledge that security theatre, while still deceptive and potentially problematic, could sometimes be valuable. His example is security bracelets on newborn babies, which don't do much to protect against the actual but extremely small risk of abduction, but do a great deal to calm anxious parents. If Dr Sandman's precaution advocacy is targetted at situations of High Hazard, Low Outrage (in other words, people not worrying enough), then Security Theatre could be legitimately targetted at situations of Low Hazard, High Outrage (people worrying too much).

So perhaps sometimes giving people a false sense of security is ethically justified?

Peter Glaskowsky, Bruce Schneier's New View on Security Theater (CNET, 9 April 2008)

Peter Sandman, False Sense of Security (25 May 2018), Precaution Advocacy (undated)

Bruce Schneier, Beyond Fear (2003), In Praise of Security Theatre (Wired, 25 January 2007)

Gerald Wilde, Risk homeostasis theory: an overview (Injury Prevention Vol 4 No 2, 1998)

Wikipedia: Risk Compensation, Security Theatre

Related posts: Surveillance and its Effects (May 2005), Technical Security and Context (September 2005), Hard Cases Make Bad Law (September 2009), The Illusion of Architecture (September 2012), Anxiety as a Cost (January 2013), Listening for Trouble (June 2019)

Updated 28 June 2019. Thanks to Peter Sandman for comments.

Saturday, February 09, 2019

Insurance and the Veil of Ignorance

Put simply, the purpose of insurance is to shift risk from the individual to the collective. When an individual cannot afford to bear a given risk, the individual purchases some risk cover from an organization - typically an insurance company or mutual - which spreads the risk over many individuals and is supposedly better able to bear these risks.

Individuals are sometimes obliged to purchase insurance - for example, car insurance before driving on the public roads, or house insurance before getting a mortgage. In some countries, there may be legal requirements to have some form of health insurance.

Insurance companies typically charge different premiums to different individuals depending on the perceived risk and the available statistics. For example, if young inexperienced drivers and very elderly drivers have more accidents, it would seem fair for these drivers to pay a higher premium.

Insurance companies therefore try to obtain as much information about the individual as possible, in order to calculate the correct premium, or even to decide whether to offer cover at all. But this is problematic for two reasons.

The first problem is about fairness, as these calculations may embed various forms of deliberate or inadvertent discrimination. As Joi Ito explains,
The original idea of risk spreading and the principle of solidarity was based on the notion that sharing risk bound people together, encouraging a spirit of mutual aid and interdependence. By the final decades of the 20th century, however, this vision had given way to the so-called actuarial fairness promoted by insurance companies to justify discrimination.
The second problem is about knowledge and what Foucault calls biopower. Just suppose your insurance company is monitoring your driving habits through sensors in the vehicle or cameras in the street, knows how much red meat you are eating, knows your drinking habits through the motion and location sensors on your phone, is inferring your psychological state from your Facebook profile, and has complete access to your fitness tracker and your DNA. If the insurance company now has so much data about you that it can accurately predict car accidents, ill-health and death, the amount of risk actually taken by the insurance company is minimized, and the risk is thrown back onto the individual who is perceived (fairly or unfairly) as a high-risk.

In her latest book, Shoshana Zuboff describes how insurance companies are using the latest technologies, including the Internet of Things, not only to monitor drivers but also to control them.
Telematics are not intended merely to know but also to do (economics of action). They are hammers; they are muscular; they enforce. Behavioral underwriting promises to reduce risk through machine processes designed to modify behavior in the direction of maximum profitability. Behavioral surplus is used to trigger punishments, such as real-time rate hikes, financial penalties, curfews, and engine lockdowns, or rewards, such as rate discounts, coupons, and gold stars to redeem for future benefits. The consultancy firm AT Kearney anticipates 'IoT enriched relationships' to connect 'more holistically' with customers 'to influence their behaviors'. (p215)

So much for risk sharing then. Surely this undermines the whole point of insurance?

Sami Coll, Consumption as Biopower: Governing Bodies with Loyalty Cards, (Journal of Consumer Culture 13(3) 2013) pp 210-220

Caley Horan, Actuarial age: insurance and the emergence of neoliberalism in the postwar United States (PhD Thesis 2011)

Joi Ito, Supposedly ‘Fair’ Algorithms Can Perpetuate Discrimination (Wired Magazine, 5 February 2019) HT @WolfieChristl @zeynep

AT Kearney, The Internet of Things: Opportunity for Insurers (2014)

Cathy O'Neil, How algorithms rule our working lives (The Guardian, 1 September 2016)

Jathan Sadowski, Alarmed by Admiral's data grab? Wait until insurers can see the contents of your fridge (The Guardian, 2 November 2016)

Shoshana Zuboff, The Age of Surveillance Capitalism (Profile Books 2019) esp pages 212-218

Stanford Encyclopedia of Philosophy: Foucault

Related posts: The Transparency of Algorithms (October 2016) Pay as you Share (November 2016), Shoshana Zuboff on Surveillance Capitalism (Book Review, February 2019)

Saturday, November 10, 2018


Some of my posts recently have mentioned the work of @zeynep and others on the polarizing effects of social media platforms, especially YouTube.

But this phenomenon is not restricted to the Internet: traditional mass media is subject to similar effects. Following an extraordinary confrontation between CNN and the White House, Michael Massing reviews CNN's political coverage and finds it to be extremely one-sided. It appears that Trump and CNN each benefits from constantly attacking the other. Massing calls this codependency, but I believe a more accurate term would be symmetrical schismogenesis. This concept, originally developed by Bateson and elaborated by some of his followers including Jackson and Watzlawick, refers to the situation where two parties mirror each other, the behaviour of each serving to reinforce the behaviour of the other.

Who benefits from this polarization? The media platforms (YouTube, CNN) are essentially selling eyeballs to companies that want to advertise stuff. This is not just about the number of eyeballs but the number of eyeballs in relevant demographic categories. Thus for example gender or socioeconomic polarization may be helpful to this mission if it helps produce an audience that is particularly receptive to whatever is being advertised. However, polarization can also produce effects that are unwelcome to risk-averse advertisers - for example, associating their brands with controversial content, or even exposing them to the risk of consumer boycotts.

Writing in 2013, Markus Prior notes the correlation between cable news consumption and political polarization, but also notes the way that increasing choice on cable networks allows non-partisan viewers to avoid watching cable news altogether. Thus the apparent polarization would appear to be a consequence of a self-selecting audience.

Massing regards CNN's coverage of Trump as "seeming uninformative, repetitive, and nakedly partisan". This echoes a more widespread complaint about 24 hour rolling news: that it fills the airwaves with endless chatter (which Heidegger called Gerede and the Lacanians call Empty Speech.)

On cable news, there are two feedback loops that reinforce this phenomenon. Firstly, the partisanship alienates non-partisan viewers, thus further concentrating the audience. Secondly, people with genuine knowledge and insight quickly discover that the platform doesn't give them a fair opportunity to communicate to an open-minded audience, and therefore abandon the platform in favour of those who are happy to spout dogma on a variety of topics.

On YouTube, these two feedback loops are less in evidence. There is a wealth of good content on YouTube if you know where to look, including Zeynep Tufekci herself talking about this very phenomenon. (But just compare the numbers of views of selected videos on different channels.)
(view numbers as shown on 10 November 2018)

Update (March 2019)

@charlesarthur observes that even politicians aren't always immune to the polarizing effects of social media. He suggests that the closed WhatsApp groups now favoured by all political factions are radicalising their members "so they egg each other on to take more and more extreme positions", and notes that this kind of effect has been understood for a long time. He references Cass Sunstein's 1999 paper on the Law of Group Polarization.

Incidentally, Sunstein is also known for his work on Nudge Theory, which is usually described in terms of nudging people in a beneficial direction. But the psychological mechanisms of the nudge would appear to work in any direction.

Charles Arthur, Social media polarises and radicalises – and MPs aren’t immune to its effects (Guardian 11 March 2019)

Gregory Bateson, Steps to an Ecology of Mind (1972)

Michael Massing, Trump and CNN: Case History of an Unhealthy Codependency (NYR Daily, 9 November 2018)

Markus Prior, Media and Political Polarization (Annu. Rev. Polit. Sci. 2013. 16:101–27)

Jeff Sorensen, 24 Hour News Killed Journalism (HuffPost 20 August 2012)

Cass Sunstein, The Law of Group Polarization (John M. Olin Program in Law and Economics Working Paper No. 91, 1999)

Paul Watzlawick, Janet Beavin Bavelas and Don D. Jackson, Pragmatics of Human Communication

Wikipedia: Nudge Theory, Schismogenesis

Additional references in the following posts

Ethical Communication in a Digital Age (November 2018)
YouTube Growth Hacking (November 2018)