#information cascades

[readyplayr18]

Bubbles can be created even when everyone acts rationally. This phenomenon is called an information cascade.

Tomasz Tunguz

The obligation to dissent is the only way to combat information cascades.

Here’s how it works. Suppose we have two opaque mason jars. One jar contains 2/3 red balls, 1/3 gray balls (majority-red). The other contains 2/3 gray balls and 1/3 red balls (majority-gray). I pick one of the two jars and place it before the a line of four people.

The first person steps up to draw a ball, and I instruct them,“Look at the color of the ball, but reveal the color to no one, and replace the ball. Given the color of the ball in your hand, share a guess of which jar is in front of you. Either majority-red or majority-gray.”

Suppose the first person pulls out a red ball. The first person should declare “majority-red,” since the odds are better that a red ball originate from the majority-red jar (66% vs 33%)

Suppose the second person also pulls out a red ball. She should also say “majority-red.”

Irrespective of the ball color the third person pulls, the third person will always say “majority-red.” If it’s a red ball, three red balls imply the majority-red jar. If it’s a gray ball, then it’s 2 reds and 1 gray, also imply a majority-red jar.

Every subsequent student is going to say majority-red, even if they are holding gray, because everyone before them will have said majority-red. This is the permanent and inevitable information cascade that can result when the first two balls are of the same color and everyone acts rationally. Read more about this theory and the math supporting it here.

[honorepouesi]

[batamericanboy09stuff-blog]

[femfreq]

On Twitter, Conspiracy Theories, and Information Cascades

On February 9th, Twitter announced the formation of a council made up of over 40 organizations, of which Feminist Frequency is one. These organizations are not involved in micromanaging Twitter on a daily basis or making decisions with regards to action taken against accounts or tweets, but rather, have been assembled to consult with Twitter generally about how to best navigate the challenges of allowing freedom of expression while also fighting abuse. In the days since the announcement, a wild conspiracy has formed that presents me as an Orwellian villain with the power to control what others can and can’t say on Twitter, despite the fact that, again, Feminist Frequency is just one of more than 40 organizations on a council that also includes groups like GLAAD and the National Network to End Domestic Violence. Blog posts have been written, video rants have been filmed, and hundreds of tweets have been posted as this ludicrous conspiracy has spread. According to this conspiracy, any opinions I don’t agree with are being quietly silenced, while the people expressing them are being ominously “shadowbanned.” As the conspiracy spread, I went from just being a member of the council, to being in charge of the council. A few tweets even characterized me as being in charge of Twitter entirely! Here is a very small sampling of tweets. Messages like these have been spreading for weeks: Of course there isn’t a shred of truth to this perception, but for the people spreading it, the truth is irrelevant. The tweets, blogs and videos are designed to incite rage and enlist more people to join the crusade against me. They are designed to make me out as a powerful enemy of free speech who is determined to crush the expression of any idea she doesn’t agree with, and they conveniently support the perception some have created that I am not a feminist pop culture critic but rather some kind of diabolical supervillain, hell-bent on creating an oppressive society. This conspiracy and others like it are themselves a manifestation of misogyny, borne out of a deep distrust and hatred of women. They’re designed to foster fear and serve as a warning to other women about what awaits them if they challenge the status quo. Rather than replacing the pre-existing conspiracies, this new one fits neatly in with the others that have been and continue to be proliferated. It’s as if I’m a folk demon and this is yet another horror story people whisper to each other about me and, by extension, about what effects feminism may have on our culture if this imaginary menace is allowed to spread. These absurd characterizations, unquestioningly accepted as true, then serve as the justification for more extreme forms of harassment. I know how this works because I’ve been down this road many times before. In fact, I talked about exactly this same phenomenon at XOXO in 2014, in my talk on information cascades and conspiracy theories. If you watch the talk now, you’ll see that while the story being spread about me may be different, the tactics haven’t changed at all.

[greatkingmick-blog]

[thirstygargoyle]

The number of persons holding an idea is no warrant for its objective character, else the many never could be wrong ; for uniformity of education, or the sympathy kindled by enthusiasm, may carry many minds into one state, in which belief in certain ideas, and the mistake of formulae or usages for external truths, will be natural or necessary. Such are popular superstitions ; or the law of honour, as professed by men of the world; or the heated notions created by mob oratory ; all of which are as baseless and untrue as they are influential. Again, a whole train of investigation or inference may depend on the original admission of some one proposition which is false ; and the consequent unanimity with which separate minds regard and treat the same matters may be unfairly taken as a concurrent evidence of the truth of the conclusions at which they arrive.

John Henry Newman, An Essay on the Development of Christian Doctrine, 94.

[justin-singer]

Social Proof and Information Cascades

In the venture world, there's a lot of talk about social proof. As someone who's spent a good chunk of his life studying cognition, I know that social proof is one of mankind's most powerful tools in decision-making. At the same time, we at IA Ventures believe that blindly following social proof is a poor investment strategy (if you can consider it a strategy at all). But how can it be that social proof is both powerful and useless? I think it's because there are (at least) two conflicting models of "social proof."

"Independent" Social Proof

I tend to think that when most people see social proof, they assume independent decision-making has taken place. What I mean by "independent" is that each member of the group has reached an independent conclusion about the concept through first-order research. When four different people you respect reach the same conclusion about a concept, that rightly sends a strong validation signal. Because I love diagrams, here's how this looks:

This is what people seem to believe has happened when they see a list of social proof. However, I think social proof more often conforms to a different model...

Information Cascades

I believe what's often called social proof is in fact an information cascade. In this schema, one person makes an independent decision about a concept, and others make independent decisions about that first person's decision. The result is a "cascade" of decisions, each influenced more by the previous decision-maker than by the concept at issue. Here's the snazzy diagram:

It's important to note that this behavior is not de facto irrational. Leveraging the judgment and experience of others is one of the most powerful cognitive tools we have. Unfortunately, as time expands and linkages grow, what begins as rational decision-making based on cascaded information can morph into irrational herd behavior.

Identifying Information Cascades

What's especially frightening about this conflation is how difficult it is to recognize and protect against a transition from rational following to irrational herding. The major complexities I see are:

Information asymmetry. It's impossible to divine the basis of an investor's decision from his or her presence on a list of social proof. Maybe they invested after taking a hard look at the company and drawing an informed decision. Maybe they invested to support another angel. And maybe they invested because they saw the herd and wanted a piece of the slaughter. Who knows?

Temporal uncertainty. The sequential nature of cascaded information implies a time lag between the first investor's examination of the business and the second investor's examination of the first investor's judgment. Seed stage businesses often evolve three times a day, and twice more at night. In other words, sequential investment decisions are almost necessarily based on the stale judgments of others. 

Delayed feedback. The most meaningful feedback loop in venture investing (returns) is extremely delayed, often on the order of 7-10 years. In contrast, it only takes a short time to build up a reputation as someone who can be relied on for social proof. Thus, someone whose investment looks like a positive proof point now may in fact be the exact opposite.

There are, of course, many more factors at play in such a boundlessly complex system. These are but a representative sample meant only to illustrate that social proof does not exist in a vacuum.

Conclusion

This whole post of course takes place in the context of the latest AngelList kerfuffle, but I choose not to wade into that particular debate. Rather, I wanted to lay out a few models for thinking in a more nuanced way about social proof and its value as an investment criteria. First-degree social proof is often a rational and intelligent decision, but if you want to understand your risks, it's essential to understand exactly what type of "proof" you're relying on.