Hacker News new | past | comments | ask | show | jobs | submit login
Research Finds Tipping Point for Large-Scale Social Change (upenn.edu)
114 points by nabla9 on June 9, 2018 | hide | past | favorite | 30 comments



There are a number of precedents for this using game theory and simpler models:

Salience Models: https://pmstudycircle.com/2015/09/salience-model-to-analyze-...

Predictioneers Game Model: http://harvardpolitics.com/books-arts/the-mathematical-proph...

Attacker Defender Games: https://imowensims.wordpress.com/2015/12/29/attacker-defende...

When you add Nassim Taleb's "intolerant minority," rule (https://medium.com/incerto/the-most-intolerant-wins-the-dict...) I would be willing to bet you could reconstruct much of that this research as described is proposing.

I like graphs for a lot of things, and I think they are going to supercharge our ability to learn things. However, while expressing this research as a graph sounds interesting, if you take the view that pretty much everything that can be expressed as objects and morphisms can therefore be expressed as a graph, and there is a bunch of game theory that already describes this, it's possible the graphyness of any proposition is illustrative, and not evidence.

I'm looking forward to the paper and am glad the author has been able to articulate this phenomenon in a popular way, but the press release has triggered a cynical intuition.


The cliff notes of the study are provided in the article:

In this study, “Experimental Evidence for Tipping Points in Social Convention,” coauthored by Joshua Becker, Ph.D., Devon Brackbill, Ph.D., and Andrea Baronchelli, Ph.D., 10 groups of 20 participants each were given a financial incentive to agree on a linguistic norm. Once a norm had been established, a group of confederates — a coalition of activists that varied in size — then pushed for a change to the norm.

When a minority group pushing change was below 25% of the total group, its efforts failed. But when the committed minority reached 25%, there was an abrupt change in the group dynamic, and very quickly the majority of the population adopted the new norm. In one trial, a single person accounted for the difference between success and failure.

Though I'm quite curious why they call this a study in "large scale social change" and not just peer pressure? I have no idea how to possibly design an experiment for "large scale social change" but if I was going to create an experiment for peer pressure this seems to be just about exactly what I would do. Though even there it might be a bit sketchy. Mercenaries don't tend to be particularly loyal to the country they're fighting for, and 'providing a financial incentive to agree on a linguistic norm' is creating linguistic mercenaries.

Another issue that I hope they are being forthright about is "majority." If 25% of the population is made up of confederates are they describing a majority as 50% of the total (confederate + subjects) population? If that's the case then it undermines its own conclusion as it only means that 1/3rd of the subjects actually changed their view. This is something that should be obvious, but just the phrasing makes me wonder and the study itself is paywalled so I can't check.


I’d be curious to know how this would apply to group dynamics in company meetings. Often it’s very hard for an outside idea to be considered seriously.


I agree. To my eye it more closely resembles small group dynamics research, including the findings.


An experiment demonstrating "rapid social change" wasn't competing with anything substantial - it was temporary belief sets vs temporary belief sets.

Social change that is against a dominant narrative, mainstream culture or religion will be much harder.


They acknowledge this limitation in the conclusion. Clearly, if the competing labels are "I identify as American" versus "I identify as Chinese", the empirical C=25% fraction would be nonsense.

A reasonable real world comparison might be, "What is the fraction of women we need in programming to create a phase change."


Though nearly all radical social change involves a slow moving of the Overton window. If you can move a concept from unthinkable to radical, eventually radical can become considered


It also had group sizes of only 20.


By the way, "opinion formation" is a pretty well-established area of research in theoretical computer science that studies stuff like this. You start with a graph where each node has some opinion that might change based on that of its neighbours. Then you either prove something about the steady state thanks to graph-teoretical stuff, or you run simulations and see what happens. The "committed" individuals (ie. those who never change their opinion) are known as zealots.


From Wikipedia:

"Surveys show that Americans without a religious affiliation (which include 'nothing particular', agnostic, atheist) range around 21%, 23%, 25%, 31%, 34% and 21% of the population, with 'nothing in particulars' making up the majority of this demographic. Since the early 1990s, independent polls have shown the rapid growth of those without a religious affiliation."

We're almost there on religion.

[1] https://en.wikipedia.org/wiki/Irreligion_in_the_United_State...


This might imply that democrats should move to very republican towns, marry into republican families, and join republican churches.


That could be the modernized plot for a remake of 1956’s “Invasion of the Body Snatchers”. An America so polarized that, when neighbors in a small town are revealed to be Democrats, they could as well be aliens.


The study does indeed explain the reverse, namely why Republicans, despite fielding a smaller coalition demographically, still dominate American politics. They simply make up for it in zeal and refusal to switch beliefs. That enables them to punch well above their weight, whereas diverse democratic coalitions can struggle even if they win in numbers.


> " despite fielding a smaller coalition demographically"

I don't follow, the ratio is about 50:50.


As of October 2017, Gallup polling found that 31% of Americans identified as Democrat, 24% identified as Republican, and 42% as Independent.[3] Additionally, polling showed that 46% are either "Democrats or Democratic leaners" and 39% are either "Republicans or Republican leaners" when Independents are asked "do you lean more to the Democratic Party or the Republican Party?"

https://en.wikipedia.org/wiki/Political_party_strength_in_U....


It’s actually not. The majority of Americans are liberal, but through a combination of apathy, voter suppression, gerrymandering, and unequal representation, the ratio appears closer to 50:50. For what it’s worth, a republican president has not won the majority popular vote in the last 3 decades.


For what it’s worth, a republican president has not won the majority popular vote in the last 3 decades.

What about 2004?


you are very devoid of facts. I'll start with the simplest one: bush kerry Popular vote 62,040,610 59,028,444


The fact pertains to "nonincumbent president". I think the OP just left that out.


Electoral college, that last vestige of slavery...


I'd be interested to see how that would go. It's my theory that one of the biggest influences on political opinions is the population density you live in. I wouldn't be surprised to see the tipping point cause a conversion, only to see it erode completely shortly afterwards as people realize how many of the things they relied on in the city seem like a waste of time and money in the countryside.


I'd glady do that if it didn't involve living somewhere awful, marrying into a family a despise, and/or attending church :-)


This is the probably why such a polarization exists in current politics. When you consider other's opinions so unreasonable that they should be eradicated from within, instead of engaging in a civilized discourse.


I agree, they are more likely to become conservative if they follow that route.


Meanwhile, republicans should do the opposite, reversing the country's democrat-republican split. Since you want to achieve a one-party state, an easier and guaranteed effective way would be to move to Russia or China.


Actually, my personal view is that in general more diverse views are better. My assumption is the places I mentioned could benefit from more diverse views. Maybe partly why cosmopolitan cities tend to be more liberal.


I also think people living in liberal bubbles could benefit from understanding conservative views and appreciate some of their rationale.


Some questions I had while reading this article (have not read the paper)

1. Is the number actual number of people in a population or percieved number of people? That is, can someone generate fake social media accounts, inject a belief, and make it seem as if something reaches the tipping point?

2. Does this tipping point vary by culture? For example, when I read about American techies working in Japan, they speak about the excessive meetings, and consensus making that happens in companies. Do social changes require more people? Or is it that, that tipping point remains about the same, yet fewer people are likely to voice something in the first place?

3. What happens when you have multiple activist groups who are each above the tipping point, each advocating for some change?


I wonder how bad faith affects this. I'm thinking of the following areas:

* Sybil attacks/sock puppets. If I notice momentum gaining against my position, I either split my team into 2 avatars per team member, or pay astoturfers.

* assymetric warfare. If I notice momentum gaining against my position, I use my superior knowledge of the medium. For example, I use tools so that my sock puppets always get the first post, and other techniques shown to have a big influence on participants.


The actual paper is paywalled, so meh, I don't really know what the paper says.

I only know what the author of the article says the paper says.

So yeah, unless some of you have access to the PDF and want to share it, we're only discussing about an interpretation of a paper, not the paper itself.

(yeah, I know sci-hub exists)




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: