In what may be a politically biased article which bemoans flaws in US military intelligence organizations, Julian Sanchez argues that information cascades are a potential reason these flaws occur. He references an article on information cascades which mentions an experiment performed by economists Angela Hung and Charles Plott:

“Subjects were told that they would be picking marbles from one of two urns. Each urn contained a mix of dark and light colored marbles, but urn A had many more light than dark marbles, while urn B had many more dark than light ones. Each subject would pluck a marble from the urn, showing it to nobody, and then, in sequence, they made guesses as to which urn they were picking from, without showing the other group members what marble they’d seen. Each member of the group stood to win a few bucks if she guessed correctly.” - Sanchez

Assuming that Sanchez also meant to include that subjects would announce their guess, this seems very similar to the 2nd problem of our 5th homework. Hung and Plott’s experiment indicated that an information cascade would likely form in this situation, which was coincidentally also the answer to our homework problem. Sanchez asks us to think of the existence of a similar situation in an organization. Envisioning this, we could give a semi-formal structure to this problem.

If a certain decision, A or B, is to be made by a manager and a number of analysts. The analysts, each having an initial 50/50 chance of an A or B signal, voice their choice on the decision in a sequential manner. After they are done, the majority choice is chosen as a decision by the boss. A cascade could very well develop here if the first people who voice their opinions choose A because they think the boss wants them to choose A (say for political reasons).  Sanchez also mentions a potential reason why it is in the individual’s interest to follow the cascade in an organization, which seems intuitive:

“If you’re wrong when the majority gets it wrong, you’re unlikely to get singled out, but if you dissent from an accurate consensus, the mistake is much more likely to get noticed.”  - Sanchez

Now we can consider what happens if we add hierarchies to the situation, and the boss is himself/herself an analyst reporting to some other boss which reports to some other boss and so on. What we are likely to have is an example of  the SNAFU principle, which Sanchez quotes from Robert Anton Wilson:

“SNAFU Principle: Because subordinates tend to tell superiors what they want to hear, the higher up any hierarchical ladder you go, the more distorted the picture becomes. The person with the most authority in the system will likely be the most ignorant…”  - Sanchez

I would like to change this to SNAFU’ which starts with “If subordinates tell superiors…” This decision-making structure can be envisioned as a tree where a boss with analysts corresponds to a node with children. A decision-making process would start with each analyst in a set of analysts in the lowest layer of the tree choosing some decision. When a set is done, their results are reported to their boss which takes the majority decision as a signal and announces the boss’s choice. Some bosses will announce earlier than others depending on how many analysts they have, which would make announcement of choices by bosses more or less sequential

If the politically motivated analysts choose faster than their more objective colleagues, sets of analysts with more politically motivated individuals will choose faster, and consequently their boss will announce his choice faster. This implies that the politically motivated cascade would likely continue to bubble up the tree until it reaches the head node, which makes the final decision.

If this problem of politically motivated analysts does exist, Hung and Plott give advice on how to avoid a politically motivated cascade. In their experiment, they found that if all individuals announce their choices at the same time and without collaborating with each other, it increased the chance of the overall group being right. Paying the group as a whole if the majority chose the right answer also increased the chance of the overall group being right. Incorporating these suggestions into an organization would likely have the same effect although the second may be more feasible than the first. This is one example of how the information cascade concept (and research) can be used to model (and solve) real life problems.