Rancor Wins Arguments, Undermines Science

A recent study of on-line discourse came to the daunting conclusion that opinions expressed with rancor are more likely to fix opinions and create polarity than are views expressed with civility.  Most disturbing of all, inclusion of language denigrating opposing views makes people less open-minded about the topic, less able to approach the issue in a scientific, rational manner.

Social scientists have long studied how and whether argumentative, obnoxious talk may influence peoples’ perceptions. A growing body of research suggests that cantankerous rhetoric pushes some deep primal buttons that may override the more reasonable, conscious parts of our brains. One study demonstrated this phenomenon by experimentally manipulating the tone of an imaginary blogger, “Curt,” who opined about a climate change policy story. Though Curt’s reasoning was consistent, experimenters altered his language to make one post civil and the other rude, denigrating those who didn’t agree. Readers of insulting Curt came away from his blog less open-minded about the policy than readers of polite Curt.

So, if you are a propagandist who wants to polarize and shut down discourse, be rude.  You will enjoy the convenient side effect of undermining in a general way the ability of evidence-based reasoning to counter ideas being pushed only for their value to some for economic or political reasons, ideas such as denying climate change.

If you are an evidence-based afficionado of civil discourse, you are facing the prisoners dilemma.  The extensive study of bower bird behavior illustrates the problem.  In the species, males build intricate bowers to attract females, competing with other males.  Males face the decision of how much energy to exert, respectively, toward building one's own nest and toward destroying the nests of neighbors.  As one study put it,

...marauding and/or stealing  are always the best individual strategies, but neighboring males that do not disrupt each other always do better than neighboring males that disrupt each other.

Liberals who seek a society based on civility, mutual respect, and reliance on evidence-based reasoning face this same dilemma in on-line discourse, especially in the face of those invested (quite literally) in preventing solidarity of the many and evidence-based decision making, decision making which might lead to such things as using less fossil fuel.  It is a simple matter for opponents of science and equality to intentionally sow discord and hard feeling, succeeding not only in shutting down open discussion of the topic at hand but also in weakening general societal ability to rely on science and facts for steering a wise course.

Internet trolls, it seems, negatively frame the science-based debates we see online. Their rancor turns what ought to be open-minded considerations of the facts into ad hominem shouting matches among antisocial dwellers beneath bridges.

Reputation and Win Stay/Lose Shift

Considering these matters, which were already on my mind, I remembered a terrific book whose conclusions I had intended to apply to on-line behavior.  Better late than never, I hope--here is my stab at it.  The most sophisticated study of these issues to date was reported in Martin Nowak's stimulating Supercooperators:  The Mathematics of Evolution, Altruism and Human Behavior (Or, Why We Need Each Other to Succeed).   In the preface, Nowak shows understanding of the liberal raison d'etre

Many problems that challenge us today can be traced back to a profound tension between what is good and desirable for society as a whole and what is good and desirable for an individual. That conflict can be found in global problems such as climate change, pollution, resource depletion, poverty, hunger, and overpopulation. The biggest issues of all - saving the planet and maximizing the collective lifetime of the species Homo sapiens - cannot be solved by technology alone. They require novel ways for us to work in harmony. If we are to continue to thrive, we have but one option. We now have to manage the planet as a whole. If we are to win the struggle for existence, and avoid a precipitous fall, there's no choice but to harness this extraordinary creative force. We now have to refine and to extend our ability to cooperate. We must become familiar with the science of cooperation. Now, more than ever, the world needs SuperCooperators. - Martin Nowak

Power of reputation emerges from Nowak's studies as a feature of game playing which can bring some respite from the gloomy success of exploitation in previous results.  Another conclusion is that multiple smaller groups afford protection against society dominated by exploiters.  Perhaps most interesting for everyday on-line discourse, Nowak's studies revealed the optimal solution to the Prisoner's Dilemma.

First, here is the classic Prisoner's Dilemma as explained on Wikipedia

Two members of a criminal gang are arrested and imprisoned. Each prisoner is in solitary confinement with no means of speaking to or exchanging messages with the other. The police admit they don't have enough evidence to convict the pair on the principal charge. They plan to sentence both to a year in prison on a lesser charge. Simultaneously, the police offer each prisoner a Faustian Bargain. If he testifies against his partner, he will go free while the partner will get three years in prison on the main charge. Oh, yes, there is a catch ... If both prisoners testify against each other, both will be sentenced to two years in jail.

Early unsophisticated studies of this dilemma concluded frustratingly that, as with bower birds, the best individual policy of each prisoner is to defect rather than cooperate even though the best result for each would be if both cooperated with one another by maintaining silence.  (Real world studies showed that the tendency to cooperate is much higher than that predicted by studies of humans as purely selfish actors.)  Nowak added the element of reputation to the game--each prisoner will know how the other has behaved in similar situations.  Adding this element refines decision-making such that an intelligent prisoner who knows his partner has a history of not singing like a canary will himself wise up and keep his trap shut.  Now we're getting closer to actual human behavior.

Reputation refers to the ability of game players to know the cooperating and defecting history of each player.  One can see both unconscious and calculating appreciation of this fact while observing the energy put in on line to make and break reputations.  We are not likely to see an example of Obama exploiting on behalf of the 1% (defecting from his constituency) become part of the consensus view without ferocious resistance.  If we think Obama has a history of cooperation, we are more likely to choose to cooperate with him..  Alternatively, those who fight for the liberal ideal of cooperation most effectively are the most likely to face ruthless attacks on their character.  The absurd label "the professional left" is a case in point--the implication is that these opinion-makers are exploiters only in it for the money.  If the public believes you are an exploiter, they are less likely to cooperate.  Brilliantly insidious, this label, to the extent that it stuck, attacked the cornerstone of liberal attitudes--commitment to a culture of cooperation.  We also see the fiercely fought battles of reputation on the micro level in blog discussions in which participants are familiar with one another.  Although such attacks are ad hominem, and thus unscientific and illogical by nature, much on-line debate consists of attempts to make and break reputations.

Referring back to the more rudimentary analysis of the dilemma, Nowak proved an optimal strategy:  win stay, lose shift.  Before explaining this result, I would like to propose an imperfect analysis of internet debate along the lines of the prisoners dilemma.  It goes like this:  Cooperation means being civil and thus keeping minds open to determing the truth of the matter as well as keeping the playing field level in terms of persuasive power.  Defection means deploying uncivil language in order to thwart evidence-based reasoning and decrease the likelihood of minds changing.  If one commenter cooperates while the other commenter defects, the cooperator loses because his appeals to reason will be ineffective for two reasons: 1) the uncivil commenter increases the emotion level, shutting down ability to rationally weigh evidence presented neutrally, and 2) all bystanders will be less likely to be persuaded to change their views.  If both commenters are civil, then we have the optimal solution for liberals--mutual decision-making on the basis of reason, evidence, and open debate.  Unfortuately for liberals, if both commenters defect, then liberals lose for, I hope, obvious reasons.  What's a socially responsible bower bird to do?

Here is what I think Nowak's win stay/lose shift strategy would tell us.  The rule in simplified form is, whenever we have both used the same strategy, then I will cooperate; whenever we have used different strategies, I will shift.  Applying this rule would mean

1) When both my interlocutor and I cooperate--appealing to evidence without attacking the other--then I will cooperate in the next round.

2) When both my interlocutor and I exploit--each denigrating the views of the other--then I will cooperate in the next round.

3)  When I have cooperated and my interlocutor has defected, I will shift to defecting.

4) When I have defected and my interlocutor has cooperated, I will shift to cooperation.

 In summary, this strategy means that if I have done well, I will repeat my behavior and if I have done badly, I will change it.  Since doing well only means both cooperating, the only time my strategy remains consistent is when my interlocutor is also cooperating consistently.

Cut to the Chase: Applying Game Theory to On-Line Discourse

Of most interest to us blog commenters is what this strategy would advise in the face of the kind of consistently defecting behavior with which we are all painfully familiar.  If my interlocutor consistently defects, them my win stay/lose shift choice will be to alternate responses, a cooperative response followed by a defection response.  The way I think of this is that, knowing that in a society of ruthless exploiters, liberals lose, the liberal optimizes his postion by "keeping hope alive", cooperating every other interaction.  At the same time, we do not simply lie down for our beating, leaving our interlocutors assured of always carrying the day.  No, every other interaction we give them a dose of their own medicine.  I will look at practical questions concerning this strategy below, but first, let's look at two other results which may apply to maximizing utilization of the internet in a cooperative fashion.

One, methods must be developed among cooperators to support and share evidence-based evaluation of the reputation of major players as either exploiters or cooperators.  Another related strategy is to develop more and smaller networks of cooperators with many interconnections.  In this way, when one network becomes dominated by defectors, cooperators can move to networks featuring cooperation as a dominant, mutual strategy.  Like, just to take a for instance, leaving dailykos for Voices on the Square.  This result is the most sophisticated yet of game-based theory.

What is a Troll and Other Challenges

The study that began this discussion calls all who use inflammatory language and denigration of their opponent trolls.  My understanding of trolls is that they lack integrity.  But cynical calculation is not the only reason people behave irrationally and angrily; trauma interferes with logical functioning, even the trauma of previous on-line attacks.  If I, a liberal, am interacting with someone who is defecting consistently because they have been emotionally damaged, do I feel comfortable using defection strategies in every other interaction I have with them?  My frustrating, painful years on the internet lead me to answer, in theory, "Yes", but only when the stakes are important.  It strikes me that this attitude is problematically similar to attitudes of the USG, which apparently is willing to inflict unlimited "collateral damage" in order to win the endless war games they play.  This is a question for difficult discussion.

Another challenge to liberals involves going against nature.  I doubt I am alone among liberals in having believed for many years that, if I want to live in a civil, scientific culture, then my duty is to remain as civil and logical as possible no matter what others do.  In real life, choosing to behave trollishly on the basis of some game study brings up questions of ethics and integrity, for example, can a person's integrity remain intact while playing calculated games of persuasion on the internet.  Again, a question for discussion.

Another question not yet answered by research is the effectiveness of punishment.  Nowak argues forcefully against punishment as a means for achieving cooperation.  He describes the depressing experiments which have shown that players are willing to lose themselves just to see consistent defectors lose.  In my opinion, this means that in on-line discourse, liberals faced with consistent defection may reach a point of abandoning their commitment to open-minded, civil, evidence-based discourse, coming eventually to prefer to inflict what they feel is punishment.  It is disturbing to contemplate the very real possiblity of a spiral down of civil, evidence-based discourse, begun by intentional and cynical strategies of stubborn defection from a relative few, to the point that even liberals have begun to value punishment of others over their desire to live in a liberal culture.  This strikes me as highly problematic.  My top-of-the-head solution is Never Engage Strategies of Punishment.   They are the opposite of the liberal ideal and they never push others toward cooperation.  (The latter claim is a debated point.)

Identity Politics: One Final Point

Gossip. Banter. Chat. Let's talk. Let's organize a colloquium. Even better, let's have a party! Language allows people to work together, to exchange their ideas, their thoughts, and their dreams. In this way language is intimately linked with cooperation. For the mechanism of indirect reciprocity [tit for tat based on reputation] it needs gossip, from names to deeds and times and places, too. Indirect reciprocity is the midwife of language and of our big, powerful brain.  - Martin Nowak

It has been convincingly argued that the human brain developed to its size as a result of evolutionary pressures to be masters of social interaction, with language being central to this process.  Many of us attempt not to make things personal on the internet; it may be helpful to face how challenging this is.  Interaction on the internet is pure language, and use of language is inextricably entwined with assessing whether others are cooperators or exploiters.  It is my guess that, unconsciously and consciously, internet particpants are constantly vying for social position while seeking to identify allies, friends, and neutrals. (I believe this also explains the success of identity politics.)  It might be useful to accept that interaction on the internet is always and unavoidably personalized; armed with this awareness, we can better minimize personalizing of arguments.


The liberal ideal of society in which all the people participate in decision-making on the basis of evidence, logic, and respectful discourse is under attack, with the internet as a primary battleground.  Like bower birds, liberals face the challenge of finding a skillful response to an environment of exploitation.  We have seen that in on-line discussion, the win stay/lose shift strategy offers the optimal response to strategies of exploitation.  We have also seen that multiple smaller groups stand a better chance of resisting dominance by exploiters.  We have seen the special importance to cooperation of having reliable means of determining reputation--who usually cooperates and who usually defects.  We have learned that punishment is not an effective means of developing cooperation.  And finally, we have learned that identity politics and on-line discourse focused on identity are built-in features of the human brain, requiring awareness and effort to subordinate to reason and evidence.




Your rating: None Average: 3 (5 votes)


Great essay, Geo

Thanks for writing it. I'm planning on picking up the kindle edition of Nowak's book via your helpful Amazon link.

Re this interesting question you pose:

In real life, choosing to behave trollishly on the basis of some game
study brings up questions of ethics and integrity, for example, can a
person's integrity remain intact while playing calculated games of
persuasion on the internet.

It seems to me that questions about integrity always, sooner or later, go to questions of intent. If the intent of adopting strategies like this is to promote cooperation instead of "defection", then there's no problem with adopting them. As soon as one begins adopting the strategy out of desire to punish, that intent has changed, and integrity is lost. Hence your conclusion about never engaging with punishment seems even more critical than it might be at first blush. I'd personally rather switch networks than pursue the strategy when I'm really looking to punish.

Your rating: None Average: 3 (1 vote)

Thank you for that insight, willibro

geomoo's picture

Very helpful in a practical way.  Of course, our ability to lie to ourselves is strong, and catching ourselves in self-deception is, I'm afraid, going the way of the dinosaur in a culture with precious little time spent in self-reflection.  I'm going to have to spend some time with the notion of integrity going to intent.  It strikes me immediately as being the heart of the matter.

Thanks for reading the essay.  You'll love that book.

Your rating: None Average: 3 (1 vote)