Join 3,433 readers in helping fund MetaFilter (Hide)


Binary Agreement Model
July 26, 2011 12:28 PM   Subscribe

Social consensus through the influence of committed minorities: We show how the prevailing majority opinion in a population can be rapidly reversed by a small fraction p of randomly distributed committed agents who consistently proselytize the opposing opinion and are immune to influence. Specifically, we show that when the committed fraction grows beyond a critical value pc ≈ 10%, there is a dramatic decrease in the time, Tc, taken for the entire population to adopt the committed opinion. [.pdf]

Here is a summary from Rensselaer Polytechnic Institute, home to the paper's publishing members of the Social Cognitive Networks Academic Research Center (SCNARC) at Rensselaer.
posted by troll (56 comments total) 36 users marked this as a favorite

 
...I feel stupid for asking this, but could someone rephrase the stuff above-the-fold for me, but pretend I'm only about eight?
posted by EmpressCallipygos at 12:30 PM on July 26, 2011 [7 favorites]


when a vocal minority reaches about 10% it becomes extremely effective at swaying public opinion
posted by p3on at 12:33 PM on July 26, 2011 [9 favorites]


That is a fascinating link! They are sorta kinda proposing a sort of "calculus of consensus" which has wide ranging ramifications , especially in the field of social manipulation via astroturfing.
posted by Poet_Lariat at 12:33 PM on July 26, 2011 [4 favorites]


Most people in a populationbelieve something. In this population, a minority believe something else. This minority believes it really strongly, so they spend all their time talking up what they believe and it is not possible to change their minds. If this minority interacts with everyone, not just among themselves, then if the minority is at least 10% of the population, they become very effective at changing the minds of the majority, very quickly.
posted by jeather at 12:34 PM on July 26, 2011 [8 favorites]


Suppose you have a few missionaries in a large population, trying to make converts. The more missionaries there are, the faster they can win over the unconverted. What they find is that the rate at which conversions occur goes up steadily as you add missionaries - just what you'd expect! - until the number of missionaries reaches a certain percentage of the population, at which the rate of conversions starts increasing much more rapidly. Additional missionaries introduced at this point raise the rate of conversion much more.
posted by Wolfdog at 12:37 PM on July 26, 2011 [1 favorite]


This is why I get annoyed by people who bring up politics at friendly gatherings. The personal is personal, and the political is political. Now let's get back to enjoying each others company.
posted by Afroblanco at 12:38 PM on July 26, 2011


I learned this in Age of Empires.
posted by odinsdream at 12:39 PM on July 26, 2011 [5 favorites]


I hope no Australian media magnate finds out about this. He could really screw things up for a lot of people.
posted by Daddy-O at 12:40 PM on July 26, 2011 [34 favorites]


I'm pretty sure I could get 10% of the populace to proselytize about how we'd all be better off if the majority of our CEOs and legislators were ground up for shark chum...
posted by Ryvar at 12:40 PM on July 26, 2011 [8 favorites]


...I feel stupid for asking this, but could someone rephrase the stuff above-the-fold for me, but pretend I'm only about eight?


Sure thing.

Basically, you have a room of 100 people and a bowl of cherries. 10 of the people in the room say "Wow, those apples look good." The other 90 people look at each other and say "Hmm... I'm pretty certain those are cherries, but, whatever..."

The fact is, the 10 people won't shut up about the damn apples. Every opportunity they get, they point out the apples. The other 90 are trying to ignore them, and sometimes to humor the 10, they'll say "sure, sure. Would you like another apple?" Eventually, through mechanisms I'm not really seeing in the articles, people start buying into the whole Apples movement, since the Cherries people considered it more than well apparent the bowl was filled with said fruit. The discussion of the content of the bowl was monopolized by the people who couldn't bear the idea that they were wrong, and the people who had half a brain in their head were talking about Curb Your Enthusiasm episodes, Neutral Milk Hotel, or something.

After a while, the contents of the bowl is considered to be Apples, and the people who know darn well there is no such thing in that room are in the corner, avoiding the topic, and talking about the pizza delivery service in Snow Crash.
posted by Bathtub Bobsled at 12:41 PM on July 26, 2011 [34 favorites]


If you are a revolutionary/freedom fighter/terrorist, don't worry about your ideas being perceived as extreme. Instead, get your message out and make sure you control that 10% of the population, the rest will follow.
posted by Foci for Analysis at 12:41 PM on July 26, 2011 [1 favorite]


could someone rephrase the stuff above-the-fold for me

Go to any public planning meeting in San Francisco. Witness a vocally persistent minority redirect and monopolize the conversation toward their own pet issue. You will soon understand the tactic.

I realized this at a neighborhood meeting where a talkative young man from the bicycle coalition shared many passionate opinions about how traffic in my neighborhood should be arranged. He brought many friends to the meeting to help sway opinion. When asked where he lived he replied, "Oakland".
posted by quadog at 12:45 PM on July 26, 2011 [4 favorites]


This is a physics paper that models social interactions using tools designed to describe particles bumping into each other, and comes up with various conclusions that may or may not be at all robust.
posted by grobstein at 12:46 PM on July 26, 2011 [8 favorites]


Points to p3on for paraphrasing perfectly.

Bathtub, I'm not sure what you're getting at, but I get what the FPP is about, so I'm calling it good.
posted by EmpressCallipygos at 12:49 PM on July 26, 2011


could someone rephrase the stuff above-the-fold for me

If a committed group of proselytes tell the Big Lie often enough, the rest of population will begin to believe it. The rate at which new believers are minted increases once one in ten people is on your side.
posted by darkstar at 12:51 PM on July 26, 2011 [1 favorite]


It's really important that the vocal minority talk to people in the majority, and the minorities should be interacting with different groups, so they reach everyone in the majority. This doesn't work if the minority all spend their time agreeing with each other, or surrounding a single other person.
posted by jeather at 12:52 PM on July 26, 2011


... thus Fox News
posted by zomg at 12:53 PM on July 26, 2011 [2 favorites]


Interesting model, but real world politics would always have many different committed individuals who're extremely unlikely to change their mind, which changes your 10% figure considerably.

It nevertheless suggests why 'the stupid' has made soo much progress in U.S. after the religious right became involved in politics. You should however note that the default explanation for 'the rise of the stupid' has usually been 'elites cared little once communism was defeated'.
posted by jeffburdges at 12:54 PM on July 26, 2011


Now that it's been reduced to elementary-school level in ten different ways, can we move on to discussing the content of the paper?

What do people think of the differences between the binary agreement model and epidemiological ones, and in whether it's practical to test these findings experimentally on an existing social graph. I'd be particularly interested in hearing from anyone who knows a good way to scrape Disqus comments.
posted by anigbrowl at 12:56 PM on July 26, 2011


You know, I tend to question things like the "Overton window," and claims to the ability to make people do or think something that didn't already want to do or believe (funny, given that I work in advertising).

However, after the terrible events in Norway, and the number of "voices" I've found ostensibly disagreeing with the killer's methods, but sympathizing with his stated aims and vehemently arguing that "multiculturalism is in crisis" in Europe, I'm starting to wonder. Are we creating a new political culture of crazy? Are there interested parties actually manipulating public opinion through trollish posting in comments sections?

Oh, here's a good example.

Is this just noise, or are there that many people walking around in some sort of paranoid fantasy that just happens to occupy the same physical space as evidence-based reality?
posted by TheWhiteSkull at 12:58 PM on July 26, 2011 [2 favorites]


Slides, for those who don't want to read the full paper.
posted by anigbrowl at 1:01 PM on July 26, 2011


This is interesting, especially given the well know Crazification Factor at play in American politics.
posted by Freen at 1:08 PM on July 26, 2011 [1 favorite]


> agents who consistently proselytize the opposing
> opinion and are immune to influence

So, sending out Turing robots that sound convincing is the key to controlling public opinion.

I hope nobody figures this out. It would certainly ruin Usenet.
posted by hank at 1:09 PM on July 26, 2011 [2 favorites]


Thank for the post. It's been awhile since I've read an academic paper, so the pdf is gonna take me some time to puzzle through.

Is it accurate to say that this work is based entirely or almost entirely on models and assumptions about human behavior, rather than data on human behavior?
posted by ibmcginty at 1:11 PM on July 26, 2011


OK, I read the paper. It's a model. There are actually no data in here. In a situation like this, the model may describe the behavior of the system very well in a qualitative sense, but the exact numbers you get out of it (like this 10% number) may not reflect reality. So yes, it's possible that human opinion "dynamics" will have a tipping point just as the authors describe, but I wouldn't put money on the critical value actually being 10% of the population.
posted by zomg at 1:11 PM on July 26, 2011 [2 favorites]


This is a physics paper that models social interactions using tools designed to describe particles bumping into each other, and comes up with various conclusions that may or may not be at all robust.

I think this is how their model works:

(1) You do not believe that you are Napoleon.
(2) Someone tells you that you are Napoleon.
(3) "I am Napoleon" goes onto your list of possible opinions.
(4) If someone else then tells you that you are Napoleon, you agree and believe that you are Napoleon.

Their model assumes away most of what we've learned about public opinion and its formation since 1950.

Is it accurate to say that this work is based entirely or almost entirely on models and assumptions about human behavior, rather than data on human behavior?

Their paper presents no data whatosever. The graphs you see are simulations based entirely on the assumptions of their model.
posted by ROU_Xenophobe at 1:22 PM on July 26, 2011 [5 favorites]


> agents who consistently proselytize the opposing
> opinion and are immune to influence

What they are showing is the effect of thinktanks. You create a narravite, and you get a figurehead to spout the narrative. Said narrative is repeated by true believers and ends up being sent out across social networks (Twitter, email lists, Facebook, etc, etc, etc). There was a recent one that hit facebook about how Florida has passed a law that will require people on "welfare" to get drug tested to receive public aid, and the end of it asks "if you agree with this, repost this". These things get tracked to see how far and wide they go. It is a known phenomenon that Snopes.com half-heartedly tracks, but this institue has gotten money from the government to formally study it and come up with equations on how it works.

Don't kid yourself. This isn't new. This is basically studying the "grapevine" of gossip and how it spreads through a population, but focused specifically on internet social networks. The part that would be more interesting, I think, is to look at how much the information is simply polluted along the way. I've overheard conversations in public about politics where someone misattributes where they heard something, but repeats the talking point pretty much point for point, even though they say before hand "I normally vote Democrat, but I heard Obama is a secret Muslim and I can't vote for him because...".

It can be incredible, too. Once someone "learns" the talking point, their thought processes stop. That's all they know, is the soundbite. And that influences their decisions. They don't know that the lie originated in a think tank whose sole purpose is to come up with ways to spread misinformation and lies.

I'm going to shut up now because yetserday I already made myself a winner for arguing this point on the internet and found myself sadly disappointed by people unable to get past their own cognative biases to evaluate what they "believe".

posted by daq at 1:23 PM on July 26, 2011 [3 favorites]


What O'Sever being of course the famous murderer of 1870s Dublin, who would chop his victims' limbs off with a katana.
posted by ROU_Xenophobe at 1:23 PM on July 26, 2011 [3 favorites]


At some point I was going to make an FPP about the interesting developing literature on mathematical models of consensus and cooperation. Just as a taste, here's Lee Worden and Ben Golub. Golub's paper "Network Structure and the Speed of Learning" is especially relevant to this thread.
posted by escabeche at 1:25 PM on July 26, 2011 [6 favorites]


Thanks for your reply, ROU_Xenophobe. I remain ambivalent as to whether or not I am actually Napoleon; do you have any thoughts?

Their model assumes away most of what we've learned about public opinion and its formation since 1950.

Do you have right at hand a source, or source of sources, that might allow a neophyte to get into this kind of thing? The one class I took on public opinion was about 100% models and 0% empiricism, so I'd be glad to get a bit more of an idea about what the data says...
posted by ibmcginty at 1:41 PM on July 26, 2011


could someone rephrase the stuff above-the-fold for me

The Tea Party is going to fuck America.
posted by twoleftfeet at 1:42 PM on July 26, 2011 [4 favorites]


ctrl F Atheist. No. Really? Cool.
posted by herbplarfegan at 1:50 PM on July 26, 2011


Legalize it.
posted by furiousxgeorge at 2:05 PM on July 26, 2011 [4 favorites]


It's a model.

It is not empirically verified.

Ignore it until they at least try and do some real world science rather than just math and stats.
posted by srboisvert at 2:13 PM on July 26, 2011


I looked at the paper and was confused until I realized that this was a result about the dynamics of "opinions" in graphs, and only metaphorically about human behavior. Unfortunately the press release has the authors expounding on psychology rather than their fields of expertise.
posted by parudox at 2:16 PM on July 26, 2011


Doesn't anyone here find this encouraging?
posted by TwelveTwo at 2:41 PM on July 26, 2011 [2 favorites]


If a committed group of proselytes tell the Big Lie often enough,

It's funny because when I read the above-the-fold, my thoughts jumped immediately to the issue of same-sex marriage (which, you know, I think is a good thing). This isn't a...technique, or whatever the proper term would be that applies only to things we believe are bad.
posted by rtha at 2:46 PM on July 26, 2011 [5 favorites]


I have seen the evolution of defenses to this strategy, i.e. serious social ostracism for bringing up the subject at all. I rarely discuss politics in certain groups now, as it's completely toxic to all further discussion.
posted by effugas at 2:50 PM on July 26, 2011


Would the effect a minority has on the general opinion be more to do with the loudness of the minority, rather than the size of that group? Or have I already been proved wrong?
posted by Solomon at 3:04 PM on July 26, 2011


Do you have right at hand a source, or source of sources, that might allow a neophyte to get into this kind of thing?

I don't do behavior, so I'm a bit out of the loop, but you won't go too far wrong if you hit Zaller's The Nature and Origins of Mass Opinion, which is a pretty easy read for academic stuff, or just search for "RAS model" (for "receive-accept-sample").

FWIW, the core thing I was thinking of that their model omits is the "accept" part of the RAS model -- you're constantly being bombarded with new considerations, but you're only likely to accept the ones that (you think) agree with your predispositions.
posted by ROU_Xenophobe at 3:08 PM on July 26, 2011 [1 favorite]


From my skimmed reading of it, it seems that the paper only addresses the effect of agents with a specific opinion on/in a society where the remainder do not possess that opinion. That is, the opinion state is 0 or 1. It does not appear to account for hostile entrenched opinions, ie agents that possess opinions of -1 and are also running around converting others, unless I've missed something.

Also, commitment to the opinion should be an analogue rather than binary state: on any given issue there are individuals more and less committed than ourselves, and we find their conversation tolerable or not according to both their distance from our own commitment, and a fudge factor to model their and our personalities.

Promotion is different from commitment - it is entirely possible to be very committed to an opinion, but these opinions are "promoted" to others in ways that do not fit with the spread model: "you must make up your own mind", "I am in love with Freddie" ("and you should be too!" would be deprecated although "and you should like Freddie" would not), "this matter should not be discussed with the unenlightened", or a variety of other edge cases.
posted by aeschenkarnos at 3:13 PM on July 26, 2011 [1 favorite]


Do you think we can use this to get people to start using Google+ instead of facebook?
posted by ChipT at 3:25 PM on July 26, 2011


I learned this in Age of Empires.

Wololo
posted by East Manitoba Regional Junior Kabaddi Champion '94 at 4:03 PM on July 26, 2011 [3 favorites]


You know what? We'd all be better off if the majority of our CEOs and legislators were ground up for shark chum.
posted by benzenedream at 4:25 PM on July 26, 2011 [1 favorite]


TwelveTwo that was my first reaction. i decided to proselytize more about the good stuff
posted by paradise at 4:29 PM on July 26, 2011



This is why I get annoyed by people who bring up politics at friendly gatherings. The personal is personal, and the political is political. Now let's get back to enjoying each others company.

Except I rarely enjoy the company of rabid right wingers whether they are prosthelytizing or not.
posted by notreally at 4:35 PM on July 26, 2011


The paper describes how their model behaves, but their model is totally disconnected from the way people actually behave, and there's no check against data.

This tells us nothing about human society.
posted by kprincehouse at 5:17 PM on July 26, 2011


This is why I get annoyed by people who bring up politics at friendly gatherings. The personal is personal, and the political is political. Now let's get back to enjoying each others company.

I wish it were that simple. It certainly hasn't been in my reality.
posted by rtha at 5:36 PM on July 26, 2011 [1 favorite]


Wow, that's really interesting. Here's my attempt at simplification:

In this model you've got two sorts of people: normal people and cranks. Normal people can change their minds, but cranks are irrevocably fixed in their opinions. Normal people form their opinions by talking to their neighbours. They will agree that an opinion is possible when they first hear someone say it, and they will decide that an opinion is certain to be correct when they hear it a second time.

Imagine you have a room full of normal people, most of whom believe in science while the rest believe in astrology. What happens when they start talking to each other? Well, it can go either way - but usually the largest group will convert the smaller group. This happens even if there are a few cranks - the sheer number of science believers will rapidly convert anyone persuaded by the cranks.

This all changes once you have enough cranks. In that case you have a core of people who can never be persuaded, and eventually they will start persuading their neighbours, who will persuade other people and so forth. It turns out that the critical percentage *in this model* is around 10%. If you have 10% cranks then they will almost always change the minds of the group.

So, given enough time, if 10% of the population believes firmly and irrevocably in astrology, they will persuade everyone else - if this model accurately reflects human behavior, which it probably doesn't. But it's still interesting, because it shows a very sharp and sudden change in results at a particular level (10%) - so even if it isn't an accurate reflection of the real world, it might help explain why we do see sudden shifts in popular opinion: it's being formed by a core of people strongly wedded to a particular view who are able to persuade their uncommitted neighbours.
posted by Joe in Australia at 6:07 PM on July 26, 2011 [1 favorite]


Related concept: Diffusion of innovations
posted by anifinder at 7:53 PM on July 26, 2011


To put this into a political context, this is exactly the modus operandi of the then nearly moribund conservative movement beginning about 40 years ago and continuing to this day. Conservative pundits are very vocal and sure of their positions, are not amenable to discussion of the weight of evidence for or against their positions, will repeat only the facts that support their position and ignore any facts not consistent with it, and will not shut up even after their positions have been factually dismissed. I believe it is an intentional tactic that was used to gain and maintain ascendancy against a much more widely accepted generally liberal outlook in the US and other Western countries.

Note that I'm not saying these characteristics are unique to conservatives, but that the frequency of this behavior seems to be very much higher in that group. I say this as part of a family and a social circle that has both committed conservatives and committed liberals.
posted by Mental Wimp at 8:43 PM on July 26, 2011 [1 favorite]


This is very funny. Someone comes up with a mathematical model that has little or no empirical connection to actual human behavior....and the last time we saw this approach used in broad application was in the valuation of collateralized mortgage securities that caused the debt crisis.
posted by storybored at 8:44 PM on July 26, 2011 [1 favorite]


kinda like osmosis ideas flow to where the they aren't...
extraordinary popular delusions and the madness of crowds
posted by zachhouston at 11:55 PM on July 26, 2011


The weird thing about models with laughably bad assumptions is the way that they can sometimes lead to totally implausible results. Oh, wait, that's not weird.

The problem with "agents who consistently proselytize the opposing opinion and are immune to influence" is that there's nobody like that. Seriously: I know you have groups you don't like very much, but it's impossible for those groups to exist without their members being influenced to join. And it's impossible for a group to proselytize effectively without exposing themselves to alternate ideas and losing their conviction. Haven't you ever met a lapsed member of the group you hate?

This is an "ideas as untreatable diseases" model. I think somebody's been reading Snowcrash for something other than the pizza delivery methods.
posted by anotherpanacea at 4:23 AM on July 27, 2011


However, after the terrible events in Norway, and the number of "voices" I've found ostensibly disagreeing with the killer's methods, but sympathizing with his stated aims and vehemently arguing that "multiculturalism is in crisis" in Europe, I'm starting to wonder.

Of course, following this model they would argue that you yourself have long been co-opted a wrong headed lefty minority opinion that must be countered by an established majority.

And the wheel goes round and round.

Anyway, the scientific symbols are most impressive to us liberal artists, but as noted passim above, isn't this sort of a well known phenom?
posted by IndigoJones at 5:51 PM on July 27, 2011


the Social Cognitive Networks Academic Research Center (SCNARC) at Rensselaer.

I hope that they pronounce this 'snark'.
posted by polymath at 10:09 PM on July 29, 2011


« Older After creating legislation that requires voter ID ...  |  Stereogum's Stroked... Newer »


This thread has been archived and is closed to new comments