by Avtar Sehra
This is an extract from this paper, providing an in-depth analysis on the structural and dynamic economic properties of Initial Coin Offerings.
Wisdom of Crowds
One of the key driving forces behind crowdfunding type financing models, such as those underlying current Initial Coin Offerings (ICOs), is their reliance on the “Wisdom of Crowds”. This is where proponents highlight that the collective decision making capability of a crowd outweighs the capability of any one person. Therefore, such concepts are used to support why crowdfunded ICOs have the capability of selecting winners much more efficiently and effectively than traditional funding models e.g. through VCs. However, this misses a very crucial point and lacks an adequate understanding of the deeply statistical foundations underpinning the concept of Wisdom of Crowds.
Aggregating Diverse Views
Wisdom of Crowds is a framework for aggregating a diverse set of views from entities that are independent and decentralised, in order to arrive at a view that is closer to the truth than that of any one entity. This is based on the statistical concept that if you have a sufficiently large number of ‘informed’ inputs all with differing points of view, you can take the statistical average to cancel out the noise and arrive at the truth.
Using language from experimental science, this is known as the process of reducing “random error” that is always present in complex systems. However, there is another error that can be more elusive and challenging to manage. This is known as the “systematic error” or statistical bias. Such errors can skew all inputs by the same amount, so while the overall aggregated result reduces the random error of the inputs through the process of “averaging”, all the inputs can be skewed by the same amount and thus the average will also be skewed.
The ideal state would be to minimise both random and systematic error in order to ensure our results are both precise and accurate (see diagram below). However, while being less precise (uncontrolled random error) can show us something is clearly wrong, being less accurate (uncontrolled systematic error) can lead us into a false sense of security. This is why systematic errors (bias) are the bane of any good experimental scientist!
Conditions for an Effective Crowd
In the context of wisdom of crowds, we have four key conditions that need to be satisfied for effective crowds, where we can ensure we can average random errors (be more precise) and minimise systematic errors (be more accurate) to arrive as close as possible to the truth through the wisdom of the crowd. These conditions are :
- Diversity: entities should have access to private information that they use to interpret known facts related to the truth that is being deduced.
- Independence: entities should not be influenced by leaders or susceptible to herding factors such as social pressure and FOMO.
- Decentralisation: entities should be separated to ensure both diversity and independence are maximised.
- Aggregation: there needs to be a mechanism for collating views from all entities to provide a statistical average.
Social Influence and Crowds
We can argue that underlying efficient markets are social networks, however it can also be shown that not all social networks enable efficient markets . This comes from the view that while social networks can provide efficiencies in unimpeded transmission of information, establishing trust etc, badly incentivized social influencers who dominate the dissemination of biased/skewed information, can undermine the wisdom of crowds.
The experimental impacts of this are discussed in the work “How Social Influence Can Undermine the Wisdom of Crowd Effect” . This work highlights that informed “social groups can be remarkably smart and knowledgeable when the averaged judgments are compared with the judgments of individuals.”
However, it is shown that social influence effects can diminish the diversity and independence of the crowd, thus skewing the statistical aggregate and deteriorating the resulting collective wisdom of the crowd. In this way crowds can be considered as acting as a collective (or a mob), where group thinking ensues rather than diverse, knowledgeable and informed individuals acting on personal points of view.
Why This Matters for ICOs
What does this mean from an ICO perspective? When there are high profile ICO advisors with large followings on easily accessible social networks, who are incentivized inappropriately to market products irrespective of quality, such a situation can result in crowds becoming less effective.
This social influence in markets becomes even more damaging when there is information asymmetry, and a lack of a market mechanism to ensure transparency and minimize predatory and outright criminal behaviour. So, when you have a market where most of the information is controlled by one side of the market, as in the case of issuers of ICOs, and the market has strong social mechanisms for information dissemination, we can end-up with a case of “adverse selection”, that can lead to market instability. This phenomenon may be afflicting the current ICO market, where people start off by chasing dreams of getting rich quick but just end up holding lemons in the longer term!
Examples of Insanity
An extreme example of skewing market perceptions through nothing but marketing comes from the “Great Beanie Baby Bubble” of the late 90s , when rare versions of the plush toys, created by the Ty Warner company, started selling for up to $5,000, simply due to phenomenal marketing based around their manufactured scarcity! However, by mid 1999 announcements of retiring product lines ceased to have price rise affects the Beanie Baby market was accustomed to. As a result the “short term market makers” started exiting in anticipation of a potential decline, thus resulting in an all out collapse! Those holding for the long term and the late adopters were left holding “lemons”. In one extreme case an “investor”, saving for his children’s college tuitions, ended up losing up to $100,000 when the bubble burst 
Tulips or Beanie Babies?
Many people like to use tulip mania  to put current ICO market in context, but maybe the Beanie Baby mania is a more appropriate mirror. Beanie Babies and ICOs/tokens have potential “utility” for a small number of people, both have imposed artificial scarcity and are implicitly marketed as instruments of speculation in a market driven by asymmetric information. However, the key difference between the two is that Beanie Babies are clearly not securities, not unless Ty Warner was pooling funds from investors with the expectation of delivering beanie babies that would go up in value from their initial “investment” price.
I’ll end this post with an article from Slate, that puts the insanity of crowds (or systematic errors) into context : Beanie Babies: Bubble Economics and Psychology of a Plush Toy Investment :
“From this distance, it’s easy to laugh at Beanie Baby fever, to mock it as just another pointless fad in a chintzy, hollow decade. But in the latter part of the 1990s, Beanie Babies were so much more than a fad: They were a mania, an obsession that ensnared not just gullible children but also otherwise responsible adults who lost all sense of perspective over these plush playthings. People sold — and bought — some rare Beanie Babies for $5,000 each and expected others to skyrocket in value within a decade. (Collectors were careful to keep each toy’s tag attached and protected by a plastic case; a Beanie Baby’s worth was said to fall by 50 percent once the tag was removed.) Looking back, it’s clear that the Beanie Baby craze was an economic bubble, fueled by frenzied speculation and blatantly baseless optimism. Bubbles are quite common, but bubbles over toys are not. Why did America lose its mind over stuffed animals?”
Thanks to Vic Arulchandran for his contributions to the refinement of this and upcoming work.