COVID-19: Ioannidis vs. Taleb (International Institute of Forecasters)

The International Institute of Forecasters facilitated a very interesting discussion in the form of two scholarly blog posts – COVID-19: Ioannidis vs. Taleb, Learning from the positions of Nassim N. Taleb and John P. Ioannidis in the COVID-19 debate. The idea of this match up of differing views from such profound thinkers is literally brilliant. The setting is what may be the closest thing we have seen of an old fashioned discourse. Pardon me for being old fashioned but actual, respectful dialogue and discourse is what is desperately needed in our present time and has been missing for decades. How much more do we need such a thing in an age where many of our national and international challenges have taken on a complexity due to the natural order of progress, technology and globalization. From the introduction page:

“Two…voices have been highly visible in the public debate, with seemingly diverging opinions. I am thinking here of John P. Ioannidis and Nassim N. Taleb (arguably two of the greatest living thinkers) holding opposing views about how to deal with the present pandemic and its potentially destructive consequences.

…Nassim N. Taleb believes that all efforts and resources should be directed to halt its spread and reduce the number of infected and deaths without any concern about forecasting its future course as the uncertainty of doing so cannot be measured and the risks involved are highly asymmetric. John P. Ioannidis, on the other hand, claims that more reliable information is needed to make multiple billion-dollar decisions and that forecasting has failed us by being too pessimistic about the future growth of the pandemic and by exaggerating its negative effects.

…This debate will not only allow us to better understand the points of view of the two great thinkers but be also left as a guide for how to deal with future pandemics.”

I would only take issue with the comment, billion-dollar decisions, since we are already well north of multi-trillion dollar decisions. Of the two positions, Taleb remains consistent with his central theme for more than a decade, namely, the profound consequences of risk versus upside benefits, and this is only amplified in pandemics, where he states they “represent existential risk.” In other words, survive now:

“Science is a procedure to update knowledge; it can be wrong provided it produces interesting discussions that lead to more discoveries…for matters that have systemic effects and/or entail survival, the asymmetry is even more pronounced.”

Ioannidis, on the other hand, has been one of the few critics of our nation and the world being put on “horror-alert,” with models that he states have:

“Failed when they used more speculation and theoretical assumptions and tried to predict long-term outcomes, e.g. using early SIR-based models to predict what would happen in the entire season.”

One note that I frequently see as both a criticism and straw man argument against Ioannidis that is worth mentioning: he is clearly not saying to do nothing, or that COVID-19 is trivial. He opens his post with the following, “COVID-19 is a major acute crisis with unpredictable consequences.” Later he and his colleagues state, “a doomsday forecast may come handy to protect civilization, when and if calamity hits. However, even then, we have little evidence that aggressive measures which focus only on few dimensions of impact actually reduce death toll and do more good than harm.” See introductory post here, Taleb here, Ioannidis here.

Epidemic Forecasting Challenge: Human Behavior

It is interesting to search and review older articles (looking for pre-COVID perspective) and realize how long we have been discussing these topics that have monopolized our current discussions in the last six months. In The Journal of Infectious Diseases, a study from 2016, Epidemic Forecasting is Messier Than Weather Forecasting: The Role of Human Behavior and Internet Data Streams in Epidemic Forecast, the authors explore the challenges of forecasting the spread of epidemics and compare and contrast with the progress of weather modeling.

The authors note in the abstract:

In the past 3 decades, the weather forecasting community has made significant advances in data collection, assimilating heterogeneous data steams into models and communicating the uncertainty of their predictions to the general public. Epidemic modelers are struggling with these same issues in forecasting the spread of emerging diseases.

Sound familiar? Here’s why, “epidemic models rely on human interactions, multiple data sources such as clinical surveillance and Internet data, and environmental or biological factors that can change the pathogen dynamics.” This reminds me of the hasty rush to a positive affirmation of test and trace (which I would not argue against) but could prove fraught with problems associated with the many and variegated variables of human behavior, requisite cooperation, and potentially unknown interactions (such as if asymptomatic superspreaders prove as accurate as is being posited). On this side point, I am not arguing against the activity of test and trace as it may be the best option we have, but the declaration of its success, given the myriad of unknowns.

The authors make a very interesting observation given what we, the entire world have experienced in the last six months, with some profound policy decisions that have been ostensibly driven by data:

Epidemic forecasting is still in its infancy and is a growing field with great potential. The challenges for accurate epidemic forecasting include data availability and emergent changes in human behavior and pathogens.

They conclude with this hopeful note which could prove prescient from four years ago, as phones and data have only increased, “there will be a parallel world, similar to that for weather forecasting, where billions of sensors will be uploading real-time information to obtain personalized disease forecasts.” Full study found here.

FRB Atlanta GDPNow: Q2 Throttled Down Slightly

From the FRB Atlanta nowcast:

The GDPNow model forecast for real GDP growth (seasonally adjusted annual rate) in the second quarter of 2016 is 2.5 percent on May 17, down from 2.8 percent on May 13. The second-quarter forecast for real residential investment growth declined from 5.3 to 2.5 percent after this morning’s housing starts release from the U.S. Census Bureau, the forecast for real consumer spending growth ticked down from 3.7 percent to 3.6 percent after this morning’s Consumer Price Index release from the U.S. Bureau of Labor Statistics, and the forecast for the contribution of inventory investment to second-quarter growth declined from -0.24 percentage points to -0.39 percentage points after this morning’s industrial production release from the Federal Reserve. The latter decline was concentrated in motor vehicle and parts dealers’ inventories.

NOWCast 2016-05-17Get the full dataset here and report here.