When Space Shuttle Challenger burst into flames quarter century back, it took 21 minutes for the stock market to dump shares of all the suppliers to the shuttle and Morton Thiokol stock took the biggest hit by the end of the day. All the investors – amateur and pros have dumped Morton Thiokol stock within that trading day but it took months for the investigators to conclude the same. Crowd knew it before the experts could figure out – way ahead.
Enter Wisdom of Crowds!
Wisdom of Crowd is one of the most counter intuitive concepts out there. I revisited this problem when Leigh Drogen (@LDrogen) pitched me his super awesome startup - Estimize – to disrupt sell side research for publicly traded firms.
Just because Morton Thiokol stock took a hit doesn’t mean that all the individuals who avoided to buy its shares are smart or well informed. Chances are many of the traders might have “followed the leader” or in some cases the trend. After all that is the wisdom of the crowd!
We start to question the wisdom of crowd when we pay close attention to each of the individuals in a crowd. The argument goes like – if majority of participants in a sample are not smart enough to make the right decision, even with few smart well informed individuals the average outcome of the crowd decision can be skewed by the “not so smart” lot. This argument assumes that average members of the group do not get influenced by their surrounding in making their decisions such as why is everyone selling this stock, why is there is a trading halt on the stock, etc. Such arguments are based on the individual merits and do not take into account individual decision making acumen regardless of their domain knowledge or Intellectual Quotient (IQ).
For example, members in a crowd making decisions based on their friends, other experts, trend/sentiment, could actually tilt the average towards the right decision. So, one could generate more accurate results if members of the sample know what the previous members have chosen and their reasoning behind their decision.
What’s more important? The individual member’s IQ/domain knowledge or their individual motivations?
I would argue that their motivations are as important as individual’s domain expertise/IQ in a crowd assessment.
Of course the above argument assumes that individuals in a group always make rational decisions and all the true individual motivations are apparent to the rest of the group members. That is a very big assumption but in a very controlled experiment or system, one could turn such assumptions (at least some) into reality.
For example, if we take a sample of traders in stock market – their individual motivations to estimate future stock price is based on following -
(note am only talking about motivations)
- Do they already have a position in that particular stock?
- Is that long or short position?
- If neither, what is their incentive to make an accurate assessment?
- What information do they have and how they can benefit from that information versus other members in the group, and so forth.
* If motivations are aligned, group’s decisions have lowest variance. That doesn’t necessarily mean that it would lead to most accurate decisions.
Each of the group members have their own individual motivations. Aligning the group’s motivations would get you results that are much closer – and with right motivations, average estimate could be accurate or close to accuracy. For example, traders in Morton Thiokol stock all had the similar motivations – to make money and avoid loss. Traders who own the stock want to reduce their losses and the traders who don’t own the stock want to put their money to work somewhere else. Of course some of the traders got into the stock assuming it is a bargain and hoped to make a quick buck the very next day.
* If individual motivations are apparent to the rest of the group members, rest of the group might make rational decisions.
If individual motivations are apparent to other members of the group, individuals can make the most rational decision – but depending on the individual motivations, someone of the members might not make their true drivers apparent with the rest of the group. Such behavior is the pulse of Stock market – since a trader needs another trader to sell off the losing stock and sellers need to conceal their true motivators to buyers of the stock.
For example if traders who are already in Morton Thiokol stock are trying to reduce their losses by selling, other traders who might be aware of such motivations decide not be on the other side of such trade.
Individual motivations dictate their individual actions which will ultimately impact the crowd sentiment about the assessment.
What are the types of motivations within the members of a group?
At a broad level there are group motivations as well as individual/more dispersed motivations.
Individual Motivators -
Within individual motivations, there are at least five types -
Vanity (intangible benefits) -
An individual could be motivated by vanity of being accurate and correct in crowd experiments. Such intangible motivations could be provided by the system in which such experiments are conducted such as leaderboards, badges, points and so forth. Even if the system doesn’t provide intangible benefits, building a consistent and successful pedigree would also result in future intangible perks and that could become a motivator for some of the individuals.
Tangible Benefits -
Group experiments and crowd sourcing when tied with tangible benefits could become major motivators for individuals. For example, in case of Jellybean Jar, if the instructor promises to give the entire Jellybean jar to the individual or individuals with most accurate estimate, the quality of estimates will be much higher. Of course, it would help to know the demographic of the individuals to assess the impact of such motivators – for example – the older you are, the less you are interested in winning a Jellybean jar; you might not care for it if you are diabetic or don’t have children; etc – in short – knowing what such tangible motivators mean to the individuals within the group might help rest of the group members in making informed decisions.
Could careless -
As I was saying in the previous point, motivators should mean sometime to the members of the group to have act on them. For example if the motivators are not attractive enough, the members in the group might not respond to them. So it is as important to know what motivates individuals but also the impact of motivators on individuals to understand the crowd and its decisions.
Pain versus Profit -
Not all motivators are equal. For example, if a system offers rewards for most accurate estimates the impact of such motivators is far different than when a system punishes the individuals for inaccurate estimates. Also, individuals respond differently in systems that reward accuracy and punishes inaccuracy versus systems that only reward accuracy and doesn’t punish inaccuracies.
Good denizen -
While hard to find, there are always members in a group who are just passionate to make rational decisions without any external motivators.
In case of stock market and traders, knowing about individual motivators and the impact of those motivators on individuals will help the system assess the crowd estimates and hence present the data in a most informed manner to rest of the users of the system.
Group Motivators -
Now if we move on to group motivations, if groups of individuals are pitted against other groups, and the group’s actions are tied to tangible benefits, “co-operative” and “collaborative” thinking will produce accurate assessments from the group.
Ideally any system or platform that taps into wisdom of crowd should make sure that individual motivators should not counter group motivators lest all motivators will cancel out the group actions.
PS: I limited this blog post to motivators within wisdom of crowd. I will try to address other aspects of wisdom of crowd in future blogs.
Opinions expressed are my own ….
Ps 2: I have not read book Wisdom of Crowds by James Surowiecki – so, this blog is not a reflection of that book.
Ps 3: I have not done any scientific analysis or research in any of the thoughts – open to debate and suggestions on these thoughts.