Are outliers the new black?

How the ease of information flow is distorting perspectives

Note:  I write as therapy.  It helps me think, it helps me orient my thoughts, and it helps me challenge myself.  It also helps my wife not have to listen to my ramblings, and as they say, happy wife happy life.  That said, if you find this entertaining, interesting, or valuable, and think others will as well, please share it.  I don’t have social media and I’m uncomfortable with self-promotion, so I don’t push this on anyone or through any mediums.  It is entirely up to anyone reading this to decide if anyone else would enjoy reading it as well.  And I do appreciate when people invest the time and energy in reading through my thoughts because feedback, especially from those who disagree, only helps further challenge me to be better, and I’m always interested in getting better. 

The most important thing you can say is “I love you.”  Second is “I’m sorry.”  Third is “I forgive you.”  And fourth is “but the data says…”

I originally published this piece on Medium on May 10th, 2020.  This was prior to my having content removed from Medium (my piece on my personal decision not to take the COVID shot) and making the move over to Substack.

I wanted to republish this today because it feels like, more than 2 years after publishing this piece, the problem of over analyzing outliers is getting worse, not better.  Whether it’s the “ghost of Kiev,” or screaming about the world ending due to climate change every time there’s a weather event or taking two Kathy Barnette tweets out of context to bury her in a senate primary we all should wish she’d have won, isolating information, presenting it outside of context and without any base rate information, and then normalizing it has become how media is done these days. 

Bari Weiss’ Common Sense Substack just did a great piece on a data scientist (it’s always us poor statisticians 😊) who was removed from Reuters for presenting data to his colleagues showing that police officers killing black people not only isn’t some massive existential crisis, it’s an outlier -

I Criticized BLM. Then I Was Fired.

And for doing so, for presenting correct and accurate data that represented the actual base-rate, the rate against which all data should be analyzed and all decisions should be founded, he was fired.  If that doesn’t represent the era of the Outlier, I don’t know what does.

And there’s only one way that era ends.  We all get much better at accepting 99% of what the media presents to us as fact, are, in fact, outliers.  And while outliers should be noted, smart decisions are made based on trends.

our baby girl ~30 days after birth

How many of you know a 30-something that was hospitalized by COVID-19? The number is likely very high. Judging by media reports and social media anecdotes I’d venture that almost all of you “know” a 30-something that was hospitalized by COVID-19. The number of individuals hospitalized with (not BY, but WITH, since that is how the data is tracked today) COVID-19 between the ages of 18 and 49 is 2 out of every 100,000 individuals. That is .002% of the population. That’s an outlier¹. So why can almost all of us recall someone that fits this definition?

First, and perhaps most importantly, though ignored for the rest of this piece, is that people lie. In an era defined by “if you’re not first, your last” media and “celebrity at any cost” social media people are more incentivized than ever to craft reality to their liking and ignore the most basic of validation steps when publishing or republishing stories. Most of the people you “know” who were hospitalized, are lying.

Second, and the focus of this piece, is that perhaps the largest societal change of the last 20 years is that you now “know” a LOT of people.

My daughter was born at 26 weeks and just over 1 pound. I kept a web journal throughout the experience, and despite the fact that I probably only spoke with, or even texted, a couple dozen people, thousands of individuals, connected through our social circle, our family’s social circles, etc… read a story, or followed our whole story, on caringbridge.com. We were one family, with truthfully a relatively small social circle, that now a thousand people “knew” in a very intimate way.

I remember a conversation I had with Betsy very soon after Lyla was born about how I hated hearing all these premature birth success stories². We have a relatively small social circle, but that circle extends almost indefinitely (as everyone’s does these days), and from every corner of that circle someone knew someone who gave birth to a prematurely born baby and were sharing their success stories. Premature babies aren’t especially rare. About 1/10 babies are born premature, so statistically rare, yes, but not an outlier. But babies born below 2 pounds are extremely rare — .3% of babies are born prior to 28 weeks — and yet, it felt like everyone had a success story. And that makes sense, if every premature baby is “known” by a thousand people, then almost inevitably every one of us “knows” a premature baby success story.

One of the largest societal changes of the last 20 years is our access to information. The internet, but more importantly social media as a media platform, has enabled us to touch and feel information from every corner of the globe in almost real time. This is well known and well documented, but it also preys on our most detrimental cognitive biases, replacing data with anecdote and information with hyperbole. The definition of an outlier hasn’t changed, but what has changed drastically and detrimentally is the ease with which we can recall an outlier — and therefore our cognitive assumption that it can’t possibly be an outlier at all.

I’ve struggled since day 1 to understand the reaction to COVID-19, as has been well documented (but happy to share more of my hysterical emails with those interested 😊), given the massive divide between what the data says and how the public is reacting. If all you had to go off of was the below chart, which includes COVID-19 in the 2019–2020 (Pneumonia and influenza) P&I report, could you justify locking down the entire country?

CDC Hospitalization data for sample Pneumonia and Influenza, including COVID-19, seasons

But the above chart³ isn’t all we have to go off — in fact, I’d bet for most, you’ve never even seen this chart before. What you have seen are thousands of media stories and millions of social media posts, all of which immediately fill our cognitive bias toward “availability,” with horror stories of how COVID-19 is destroying lives and incapacitating “young healthy people” across the globe. There is no data to support this, again, less than 2 out of every 100,000 people in their 30’s were hospitalized from COVID-19 — about the same fraction of the population as currently plays in the NFL — and a TINY fraction of those people lost their life due to the virus, and yet our availability bias is triggered constantly by horror stories of those impacted. How is this possible?

Because outlier is the new black.

“Contagious — why things catch on” is a tremendous book, written by Jonah Berger, a marketing professor at the Wharton School of Business. The book spends more than 200 pages breaking down why certain ideas catch on and become “contagious,” spreading through the population like, ironically, an infectious disease. It’s a marketing book written to help marketers understand how to build a product, and more importantly a product message, that is attractive enough to become “contagious.” If “Tipping Point,” the Malcolm Gladwell book, is the economist’s analysis of how and why ideas break through, “Contagious” is the marketing case-study of how to build a message that has the elements needed to break through.

Most likely all of you have heard the phrase “if it bleeds, it leads” at some point in your life. The premise is simple — people are abnormally drawn to the negative, and the media is singularly focused on the things that attract eyeballs. Put the two together and you end up with a scheduling premise built on putting the things that will get the most eyeballs in “the lead” — and unfortunately, for many humans, that means things that bleed.

“If it bleeds, it leads” is the media version of a “contagious” story. And they have absolutely perfected this messaging strategy. Nothing draws eyes, especially in the era of “trending,” like a good bleed.

COVID-19 has become the “if it bleeds, it leads” media story to end all media stories. The problem with COVID-19, of course, is that it doesn’t actually affect that many people. It doesn’t “bleed,” if you will. Hospitalizations, prior to an outrageous lock down that I’ll cover in my next piece, were, as you can see above, trending well above the 2018 flu season, but below 2017. Even assuming all deaths being attributed to COVID-19 were caused by COVID-19 (they weren’t, even (the now disgraced??) Neil Ferguson, one of the scientists responsible for leading the UK lock down, stated in his analysis that anywhere from 1/2 to 2/3’s of those who had died had a life expectancy of less than 6 months), and assuming we actually zip up to the 150,000 projection (which we now will, because anything can be coded as a COVID death, so there’s no slowing down at this point) that still makes up less than 5% of the total deaths in the US this year. At that point it will no longer be an outlier, and at this point that might actually be a good thing. If outliers are the new black, then perhaps once it surpasses media exceptionalism, we’ll return to actually reporting facts? A man can dream, right?

We don’t yet have global statistics on mean or median age of death for COVID-19 patients, but we do have a number of states and countries that are providing open and accurate reporting. One is Pennsylvania, whose largest hospital network (I’ll also cover in a future piece that our only hope for a return to sanity are the hospital networks themselves, which are being driven to bankruptcy by the fraud that is COVID-19, and SEC Football) is now asking for an end to the lockdowns, both because they’re ridiculous, and because they’re bankrupting the very hospitals that we depend on to protect us from all illnesses, real and overblown⁴. The median age of death in PA is 84 years old — or 2 years OLDER than average life expectancy in the US. Now, we could argue that might be an outlier, but Minnesota is reporting 83 years old. Massachusetts is reporting 82 years old. Sweden has indicated its median age is over 80. And we’ve already covered that the UK, PRIOR to lockdown, indicated an average life expectancy of less than 6 months for as much as 2/3rds of those impacted.

These stats have been true all along, which is exactly why you aren’t reading about them in the media. Because they aren’t an outlier, and who wants actual facts when the outliers tell such a compelling story⁵.

But stories of 50 children being impacted with Kawasaki disease (a disease that has been around for decades mind you), and blaming it on COVID-19, with no evidence for such a connection, those stories SELL. They’re the most beautiful kind of outlier — they tug at the heartstrings of everyone, as even the most jaded individual (at this point that likely includes me) will be sucked in by stories of sick children. Ironically, of course, this actually ISN’T an outlier. 50 children in the population of New York is perfectly normal for Kawasaki disease, an extremely rare infection, but one well known none the less. BUT, if we blame COVID-19 for it, now it’s exceptional, now it’s an outlier (because, you know, COVID-19 doesn’t impact children). And now it sells. Now the fact that it impacts less than .001% of the population actually makes it a story!

And if the only way to keep that sales gravy train running is to sacrifice truth, data, and real journalism to the altar of the outlier? And if it leads to hundreds of thousands of lives lost, millions of children struggling to find food⁶, and tens of millions of careers lost?

Well, so be it, because outliers are the new black.

For our sake, let’s pray the new color of the summer season is something pastel.

1. For the sake of this article I’m using 4 standard deviations off the mean, assuming a normal distribution of the population. This is anything outside of ~99.8% of a given population. If you’re a statistician you realize this isn’t a perfect definition of an outlier — “rare” is probably a more accurate term, but “rare” has zero statistical relevance, outlier here at least holds us to a standard, if not perfectly defined.

2. I feel terrible about this now — people were being amazing and doing everything they could to support us and try to cheer us up — but all I kept thinking was ~10% of babies born before 28 weeks don’t survive, so the more success stories I heard the more likely it was that Lyla would fall into that 10%. But, and this is important, that reaction was RIDICULOUS. People were sending us success stories on purpose, if they knew of a baby that hadn’t survived they obviously WEREN’T going to share that — we were looking for validation, not statistics — it took me a while to realize that.

3. https://www.cdc.gov/flu/weekly/index.htm

4. https://www.wtae.com/article/upmc-chief-medical-officer-what-we-cannot-do-is-extended-social-isolation/32393261#

5. Many of you likely took a debate class, legal class, or drama class at some point in your life (not our kids, our kids aren’t allowed to go to school, so sadly this may be lost on them…). One thing that really stands out to me was the concept of debate as an active engagement activity. That is, in order to be a great debater, you need to fully understand both sides of the argument. We, as a people, rarely do this anymore. We pick our side, we let confirmation bias run wild (we only interact with things that support our side of the story) and we dig our feet in ever deeper and deeper as we pile on the support our side of the debate deserves.

But if modern media has taken us anywhere, it’s to a position where we should question everything. Default to truth, as covered in my last piece, no longer protects us. We’re simply surrounded by too much access to too many half-truths, politically slanted “fact pieces,” anecdotal “evidence,” and out and out lies (especially on social media) to trust defaulting to truth on anything of serious consequence (and I hope we agree that 30+ million people out of work, and based on real research almost 1 in 5 American children not having enough food to eat should be considered serious consequences) - https://www.usnews.com/news/national-news/articles/2020-05-06/1-in-5-young-children-dont-have-enough-to-eat-during-the-coronavirus-pandemic.

In an era where outliers are the new black, or cognitive biases act even more powerfully against us, it’s our job to manage how we interact with facts and data, because the media simply can’t be trusted, on either side of the conservative/lefty wall, to present even the unbiased truth, much less the full and complete truth.

6. https://www.usnews.com/news/national-news/articles/2020-05-06/1-in-5-young-children-dont-have-enough-to-eat-during-the-coronavirus-pandemic