On Feedback Loops and the Selective Amplification of Random Healthcare Data

Here, within six minutes of one another, are two tweets on the subject of Amazon’s fourth quarter results, one from CNN and one from the WSJ:


The New York Times explained the apparent contradiction:  Sales were up 22% but profit fell behind investor expectations and investors freaked out. Disproportionately.

The Times notes:

the mood swings just reflected an enthusiasm that got out of hand.

Over a day, some $24B was wiped from Amazon’s value.  Not because there was anything fundamentally wrong with the company, but because of a media-fueled feedback loop.
I came across the idea of these feedback loops on a Linked In blog entry written by Tim Price, a Director at an investment firm, Price Value Partners.  Tim’s piece was on the price of oil, and the effect on markets of post-hoc speculation and rationalization.  His piece led me to a brilliant article written by Thomas Schuster, a professor of communications at Leipzig University.

Schuster writes that investors don’t make rational decisions based on an unemotional interpretation of available data.  Instead, they react, as a herd, to the selective amplification of random data and media inferences.  This mass movement of opinion destabilizes markets.

A widely held belief in financial economics suggests that stock prices always adequately reflect all available information. Price movements away from fundamentals are assumed to occur only infrequently, if at all… By conditioning trend-following behavior and fostering coordination among large numbers of investors, the media can help bring about such destabilizing moves. Media attention can induce positive feedback by increasing the level of excess noise in the market while decreasing the number of perceived behavioral options. Meta-communication thus generated is a prime source of instability in financial markets.

The mass media’s selective processing and amplification of information is the main problem.

The media produce explanations by establishing logical links and causal relations; these interpretations, though, are only more or less adequate to reality. The media enrich information by adding new elements such as “emotion” or “suspense”; through this process, however, the character of the information is altered. The media can even create their own events where nothing would happen otherwise – or they can encourage others to do so. In short: The media select, they interpret, they emotionalize and they create facts.

Schuster suggests that the reason this is done is clear: the mass media is less a public interest than a business that must be sustained by generating readership.

As generators of attention, the media are prone to condition selective awareness: The media not only reduce reality by lowering information density. They focus reality by accumulating information where “actually” none exists. The goal behind this is to win public attention, to control it and to keep it as long as possible. Ideally (from the point of view of the media), feedback loops are generated in which selected events increase attention, which, on the other hand, serves as a proof for the importance of the news, whose visibility then is increased even further.

This explains the Amazon story, which at the end of the day is a non-story.

Here, as context, is Amazon’s share price over 5 years:  The 14% “calamity” is on the far right side of the graph.

Screenshot 2016-01-31 10.30.20


3389513720_b87157656c_oOf course, feedback loops fueling themselves are seen in healthcare news as well.  The media currently seems saturated with hysterical articles about Zika virus, this year’s epidemic de jour.

In 2008, Larisa J. Bomlitz and Mayer Brezis published a piece in the Journal of Public Health describing the relationship between the intensity of media coverage of health events and their actual importance (as measured by the number of deaths).

Main Findings: Recent new health hazards were found to have been over-reported by mass media as comparison with common threats to public health. The intensity of media coverage inversely correlated with the actual number of deaths for the health risks evaluated. The current results indicate a bias toward over-reporting emerging health hazards, in comparison to their actual impact on public health. The pattern observed suggests that the more commonplace the cause of death, the less likely it is to be covered by the mass media.

Here was a graph describing the relationship.  Today it’s easy to imagine that the concerns of 2008- SARS and bioterrorism- have been replaced by breathless articles on Zika and Ebola.

Screenshot 2016-01-31 11.29.15

Here’s my take: It’s too easy to blame this derivative, aimless and distorted coverage on a mass media that has been reduced to producing clickbait to survive.  It’s also too easy to point to a generally unsophisticated and undiscriminating general public that accepts and amplifies the nonsense news they’re fed.

If information density and legitimacy are indeed the problems, the blame- I’d argue-  lies largely not with the media or the public, but with the medical establishment.  What a miserable job we’ve done pruning, prioritizing and explaining ideas and making them relevant to the average human being.

Last week I spent a few hours reading both the NEJM and JAMA, and was struck by a couple of intelligent ant important articles on healthcare policy- particularly two important articles on the subjects of long term opiates and physician malpractice.  They are two stories that should be shared with patients.  They aren’t because 1) they are dry and written in leaded, outdated academic jargon and 2) are, like most of the medical literature, hidden behind a paywall.  In order to get the malpractice article in the NEJM, for example, I’d either need to pay $20 to access the article or $35 to access the journal archives for 24 hours.

So, who can blame the public for responding to nonsense when the real, meaningful healthcare information is obscured and restricted to a guild of insiders?  Instead (and with no disrespect for my colleagues in Ohio intended.. they are at least trying to engage…) this is what often passes for healthcare communication between healthcare providers and patients.

Screenshot 2016-01-31 16.37.26

Here’s my takeaway: if we want to have a meaningful dialog around big healthcare, we need to get a lot better about simplifying and sharing sometimes complex ideas with the lay public.  The alternative is watching as media outlets create self-perpetuating information loops based, very very loosely, on some kernel of medical data.


Image: vaXzine Flikr via Creative Commons.  The ouroboros (image of a serpent eating its tail) is a symbol of cyclicality, e.g. something constantly re-creating itself.