Gambling with CoVid19

Bias warning, My wife and I can work from home, my kids kinda like homeschool (but really miss their friends) and I wiped down the groceries in the garage. 

It’s always helpful to ask, has someone faced my situation before? The answer is often yes. Rory Sutherland thrives at this.

On recent podcasts from Deep Dive (#249) and Wharton Moneyball (April 1, 2020) there were two very good steps to understanding anything with uncertainty.

Wharton Moneyball takes its name from Michael Lewis’s Moneyball. That book shed light on using advanced statistics to find other ways to win baseball games, that walking to first after a full count was actually better than hitting a single to first on the first pitched ball. Moneyball thinking has extended to new areas like basketball, movies, and Jeopardy.

On Wharton Moneyball, Adi Wyner spoke with Alan Salzberg who mentioned that he’s starting looking at CoVid19 deaths rather than cases. The former takes longer to materialize in number form but is better than the former which is mostly a product of testing. It’s trading a sampling error for a time lag.

“It was what we would generally call ‘garbage data’. A confirmed case  might me it was confirmed because someone came to the hospital and was already sick.” Alan Salzberg

Ok, good so far.

We need good data (walks instead of hits) but then Alan goes too far. The virus is mostly airborne and mostly won’t bother someone if it lands on a surface someone might touch and then finds a path into their body. That’s a lot of ifs. “Is that enough,” Salzberg wonders, “It stays for a little while, but in my mind I don’t think that should be a worry. I think you should wash your hands, and I’ve been doing that and I try not to touch my face a lot. But I think being ridiculously uptight about it is kind of crazy.”

Ok, that’s fine if we had better data.

But we don’t. Instead of six feet we might heed caution and stand at least twenty-seven feet apart. What’s the R0? How long is someone infected and asymptotic?

Ok, those are good questions.

There’s a lot of unknowns here and on Deep Dive, Matt (@PlusEVAnalytics) talked through what we can do when there are so many unknowns.

Think of Tom Brady’s 2020 over-under passing line of 4,256 passing yards, or 266 yards per game. His last four years totaled; 4057, 4355, 4577, and 3554. But with Tampa Bay he’s got better receivers. And he wants to prove to everyone that he’s still got it! And he wants to do it without Belichick!! Yeah!!!

But how much do those things count for? Like how much we know about CoVid19, we don’t know. Matt gives us a guide though. Do the things we don’t know make one outcome more likely? With age, ambiguity, competition, injury and so on, the unknown makes the under much more likely.

Matt credits much of this thinking to Taleb but the concept of sports and gambling make it clear. It seems like the unknown parts of the CoVid19 pandemic tilt the outcomes in favor of what’s much worse. Good data is a necessary start but ambiguity must be considered too.

Latest book: Idea Trails, 50 ideas from blogging the last four years.


Jeopardy: Paradox of Skill

Michael Mauboussin introduced the Paradox of Skill as a condition where as skills improve luck becomes more important to determine who succeeds. For example, combine many very talented basketball players and you get something like Kawhi’s shot.

The simplest way to think about the spectrum of luck to skill is to ask, ‘can I lose on purpose?’ If so, it’s more skill and the chance for the paradox.

So what about Jeopardy? Ken Jennings spoke with Nate Silver (of 538) about what his original run was like, what the GOAT challenge was like, and some of the logistics of playing.

Jennings explained that there were four aspects of Jeopardy that made him become the GOAT.

  1. Trivia knowledge.
  2. Buzzer skills.
  3. Game board strategy.
  4. Luck.

Let’s go through those and see where the Paradox of Skill exists and where it doesn’t. Given Mauboussin’s theory, we’d expect it to be present less often given Jennings long run of success.

Trivia knowledge. Ken Jennings is smart in Jeopardy question terms but so is everyone else in Jeopardy question terms. Everyone on the show has self-selected to be there and has passed a quiz to demonstrate their acumen.

“Most of the players know most of the answers most of the time,” Jennings told Silver.

Buzzer skills. The biggest advantage is winning because it means you’ve experienced the game. Unless there is a break to check an answer, Jeopardy is filmed in nearly real-time. Each game is followed by a break to change clothes and onboard new contestants before filming immediately begins again.

Like a basketball player getting into a rhythm, Jennings synced his buzzer cadence with Alex Trebek’s voice. When the buzz-in window opened, Jennings was first in line. (This is also the reason he attributes to Watson’s success).

Game board strategy. It was James Holzhauer who took strategy to its ultimate mathematical end. Jeopardy James figured that if he had the most experience buzzing in, but his competition probably also knew the answer, his best chance was ‘shock and awe’. That meant going for large clues and finding Daily Doubles.

Which are not quite randomly hidden. Jennings said, “The Daily Doubles are not placed randomly. A human physically looks at the gameboard, reads through some clues, and sees what type of clue might be Daily Double friendly. They try to scatter them but pseudorandomness is not actually randomness.” This same thing occurred with Spotify, Ben Cohen told Barry Ritholtz (22:25).

Luck. Ken admits he got lucky. There were probably dozens of categories he could have gotten bounced on. But they didn’t come up or didn’t come up at a bad time.

Ok, so is Jeopardy a game of mostly luck or mostly skill?

Trivia knowledge is a wash with similar competition. Buzzing is a hard skill (once Jennings really got going new contestants were offered longer training sessions to find the rhythm). Strategy is a wash too, now at least. Luck, is, well, just luck.

When Jennings was asked if Jeopardy is a sport, he pauses. It doesn’t really seem like a sport, but it’s the coverage by sport writers that feels the most accurate to him.

Like Jeopardy? We’ve written more about James Holzhauer and Watson. There’s also my ebooks.

Edit, the first version of this post had goofed up formatting. Have fixed. 


Batting Cages for Life

Note, when drafting (and reading) don’t confuse ‘more effective’ with ‘better’.

As a kid I liked playing baseball. It was always organized baseball because baseball has a coordination problem because the game doesn’t scale.  Basketball scales from two to ten. Soccer scales. Street hockey scales. Baseball does not.

I wasn’t that good at summer little league but I did strive to be better and went to the batting cages a lot.

If you’ve never been, batting cages have a row of cages opposite a row of pitching machines, with ascending speeds. My plan was to work up to a certain speed, then finish one step down. Years later this would be taught in my masters program as the Zone of Proximal Learning—work up to the point of failure or confusion, then dial it back.


Batting cages are great approximations of baseball games. Life needs more batting cages.

In the 2018, Behavioral Household Finance paper, the authors highlight what has and what hasn’t worked in the field of personal behavioral finance. The short answer is that not a lot has made large changes.

The most effective interventions focused on the selfish, salient, and simple. Tell a man how to fish and he’ll forget. Sit with people about how to save into a tax advantage account and they’ll actually do it. For example: 

  • Soldiers who were taught about savings and offered help with their Thrift plan saved more. 
  • Farmers who played a drought game (to educate about insurance) bought more insurance. 
  • Soap Operas that include financial concepts teach the concepts better than high school classes. 

The researchers also looked at how changes in the tax code affected people’s savings. The thinking went that if people paid less in taxes to save more, they would save more. Nope. Tax-advantaged savings crowded out other forms of savings. It’s like the person on a diet who cuts out ice-cream for cake. 

Read through with the ideas of selfishness, salience, and simple in mind and you’ll be able to predict which reviewed study works and which won’t. 

Teaching personal finance, no matter how great the instructor, is like talking about baseball rather than visiting the batting cages. This might be why the cult of personality is so strong in personal finance. Lifestyle over lesson. Orman, MMM, Ramsey. All sell a way to live rather than something to learn.

Selfish, simple, and salient also works for hand washing


Made up start up: Showzam

When the iPhone 3g came out, one of the most impressive apps was Shazam. Through some techno-magic, the app could identify a nearby song. This, was a big deal for me because I’ve always been terrible at knowing names and understanding lyrics. My most egregious sin, according to my wife’s family was mistaking Elvira for ‘hell-fire-up’.

Besides this mistake, my wife has only one other problem with me, interrupting television shows to ask questions. With different chronotypes, she will watch one more episode which after two weeks means she’s a season ahead.

Here the pitch: Shazam but for television shows. A person opens the app and the service detects which show is on in the room. Then it pulls up a synopsis of the plot, who the characters are, and where they’ve been. Like the recap shows run. After a skim, the second person can join in and know what the hell is going on.

The potential monetization is large. Like Roku or Amazon, the app can earn affiliate fees as people pay to sign up for other services. Showzam can also earn advertising dollars with dynamic advertising and with data collection on searches, see what is popular and monetize that information in a variety of ways.

Whether Showzam would work like Shazam I have no idea and it seems like a much more difficult technology problem. However, there’s a need and JTBD .


Quarantine Education

Shane Parrish asked, “What are some of the second and subsequent order consequences of covid-19 that you foresee with 80 percent confidence?”, how things would be different from the quarantine for CoVid19. It’s a good question to ask, if students participate in home school what else will change. A running list.

  • Better chronotype matching. Morning people get to do school in the morning, night owls at night. My oldest daughter gets two extra hours of sleep and goes to math class in her pajamas.
  • Better resources. We’ve taken drawing class from Mo Willems and learned about animals from the Cincinnati Zoo and Botanical Garden staff. My kids had great teachers but online they have access to the best ones.
  • Teaching young people. Though I haven’t seen much of this yet, it’s coming. Many instructors comment that they didn’t really understand something until they taught it. This can be true for kids at home too.
  • Learning technology tools. My younger daughter dictates her homework rather than typing it which she could do whereas in school she would use a pencil. If tools shape our thinking she’s thinking in new ways.
  • Plato’s cave and school. That same younger daughter needed help with answering why we have a leap day. That led to a talk about why we have the Georgian Calendar and not one that uses a leap week. Which also applies to why we do school-school and not home-school.
  • Asynchronous communications. If the future of work requires some asynchronous skill then this quarantine has been good practice.
  • Intrinsic motivations. My kids follow a program put forth by their school but this is mostly finished before lunch and they can move onto more enjoyable things. My guess is that a long-term homeschool arrangement would break the link between learning and school and create a hub where learning is connected to school, but many other things as well.

One week down and we are doing well.


Framing the Replacement (WWDC)

Great creative ideas are rare but when they do occur they offer a chance for Alchemy

One way to be more creative is to think about the strengths and weaknesses of a current situation, the opposition, and the contingencies. The CoronaVirus offers an opportunity to see how certain businesses are doing just that. First is Apple’s statement, then John Gruber’s comment: 

“Now in its 31st year, WWDC 2020 will take on an entirely new online format packed with content for consumers, press and developers alike. The online event will be an opportunity for millions of creative and innovative developers to get early access to the future of iOS, iPadOS, macOS, watchOS and tvOS, and engage with Apple engineers as they work to build app experiences that enrich the lives of Apple customers around the globe.” 

“Very Apple way to put it — not as a cancellation of the in-person conference but as an all-new online format equally accessible to all developers.” 

So good. 

The advantages of an in-person conference are many. It’s an event, and Apple has always been great at events. There’s buzz with people there. There’s also a sunk-cost-ness to it. If a publication is going to send a journalist, there’s going to be something written about it. 

However, in-person events are exclusive. They’re also limited in scale and scope. There’s different coordinations between a stream and a session. 

The CoronaVirus situation forced Apple to get creative. They had to find the strengths in “an entirely new online format.” 

Napoleon was doomed when he had to fight a defensive campaign. Clayton Christensen warns that capabilities become disabilities when disruption is afoot. Chuck Akre said that he doesn’t compete much with Wall Street because “Wall Street has a different business model than we do.” 

Without effort we all default to ‘the way we’ve always done things.’ Often this works fine. Sometimes outside forces force us to be creative. One path is to consider how strengths might be weaknesses and how weaknesses might be strengths. 


Hand Washing Design

John Gruber posted at Daring Fireball that when he washes his hands, he turns the water off and feels less rushed and more likely to wash for the CDC suggested twenty seconds. “It’s very clear to me after just two days that doing so makes it far more natural to spend more time actually sudsing your hands up. When you leave the water running, it subconsciously puts you in a bit of a rush, because you know you’re wasting water.” 

Rationally whether the water runs or not shouldn’t matter. The most important thing (mid-March 2020) is to kill the harmful viruses and bacteria people pick up during their (limited) social exposures. Though the chances are small, the consequences are the largest. However we aren’t rational and we don’t always wash our hands. 

At one teaching hospital, the best predictor of hand-washing was attending physicians. If they washed, the medical students followed. Multiple meta-analysis (meta-meta-analysis?) suggest the best option might be “multifaceted” nudges, educational materials, and bedside hand sanitizers. Another showed that performance reviews (personal wealth) and access to hand sanitizer (ease) had the strongest though-not-super-duper-strong effects. Incentives (personal health) also kept hand-washing levels high after the 2003 SARS outbreak.

What’s so interesting is that even though one path is clearly better, people need help following it. Hygiene is like diet or investments

This randomized control trial in India found a way to increase hand-washing 30X, even twelve months after the intervention. 

A study of 802 Kenyan households offers the model that makes the most sense to me for why people do anything. Those, “significant predictors of observed hand-washing behaviour: having the habit of hand-washing at particular junctures during the day, the motivated need for personal or household cleanliness, and a lack of cognitive concern about the cost of soap use.” 

Like finches, people are influenced by their environment. If we want to encourage actions like hand washing, social distancing, and factfullness we should design conditions that make those thing easy.


Disagreeing in a Crisis

Recently on Twitter there’s been a trend of “it’s not that bad” tweets gong around. One said that half of Italy’s CoVid19 fatalities were people with three or more existing illness while people with no other illnesses existed in less than one-percent of deaths. Among the ‘maybe it’s not that bad’ list are Elon Musk, Phil Hellmuth, and Bill Gurley.

No one is saying doing nothing, but many are saying to look at the costs. Many are saying to think like economists. 

With hindsight we’ll see that someone had the right model from day one. It likely won’t be you or me. However we get to sharpen our thinking (skill) rather than be right (luck).

So, what might account for these experts in one domain to be right in this one too? 

  • Data. It could be that there’s so little good data that we face an elephant problem. The Italy statistics look like this. The China statistics look like this. One country sees a pandemic, one an outbreak. 
  • Uncertainty. Maybe I’m too confident in my projections of outcome distributions. It could be way better or way worse than I expect right now. 
  • Salience. It could be we’re all caught up against a ‘common enemy’ with nonstop news fanning the flames. 
  • Opportunity costs neglect. People tend to overemphasize the importance of what comes to mind and dismiss what else they could spend money or time on.
  • Stock data. The stock market thinks that immediate future earnings will be significantly less. Could this be a bad proxy? 
  • Outcome severity. Maybe there will be many more with ‘zero effects’ than ‘death/ruin’. If that’s the case then CoVid19 edges more towards “driving across the country” and away from “contracting Ebola.”
  • Existing immunity. The virus has already spread through many people and those that have survived are resistant to antibodies. The influence like illness data that’s coming out might suggest this. 

It could be that Musk, Gurley, and Hellmuth were wrong in their consideration of all the details. However the process of considering why is right. Our Phantom Tyler Cowen suggests we write out why the opposing side is correct. 

There’s a lot here about arguing well and the critics of that idea say that doing is so much more difficult than discussing it. In this crisis is an opportunity.

(I use luck in the Mauboussin sense of anything out of one’s control. For example, if this were a physics problem like ‘where will planet X by at time t we would have the answer for the CoVid19 pandemic). 


Three differences between hurricanes and viruses

Worth noting, this is only based on one hurricane season (2019) and one viral pandemic (CoVid19). 

There has been a huge contrast in preparation for CoVid19 and for hurricane Dorian. Though both had the potential to do serious economic, physical, and life-threatening damage, people reacted in different ways. Why? I think there are three aspects. 


Hurricanes generate a lot of data that’s not too difficult to collect and we’ve been collecting it since the 1870’s by Catholic missionaries in Cuba. With time and tech we’ve become more precise and practiced. Hurricane forecasts include when the winds will arrive. It’s also easier to share hurricane information with the people who need it most. 

Rarely do economists and virologists have these conditions. On the Bloomberg Odd Lots podcast, Claudia Sahm bemoaned that in this case they had good data, “This started overseas and it’s different because we don’t need the unemployment rate to tell us that something bad is working its way through the global economy.” 

With ample tests, this is a different kind of problem. But no tests, no data and no data, no action. 


Floridians have a clear understanding of what hurricanes can do. The twenty-one million people who live in the state know someone or have themselves, lived through a storm. If culture is “what people do when you don’t tell them what to do,” then Florida has a pretty good preparation appreciation.

Part-of-the-reason for this culture is the cause-and-effect relationship. Storm comes through, storm destroys lives, storm leaves. Post hoc ergo propter hoc. This is harder to see with someone’s health. People look fine until they aren’t.


It’s clear what needs to be done for hurricanes. Bring things inside. Have food and water for some number of days for some number of people. Close shutters. Collect medicines. Pump gas. Pack a bug-out-bag. There’s even a tax-free weekend of shopping, where the state government encourages people to prepare. Planning for a hurricane is socially, economically, and timely easily done.

It’s hard to do anything for the virus besides take zinc. Sometimes the best action is to do nothing but that never really feels like enough, does it? Investors often say to ‘don’t just do something, sit there’ and that might be the best idea yet. 

So what? 

More tests solves the data problem. Experience affects the culture. Here we’ll focus on the actions and use the EAST framework to make choices easy, attractive, social, and timely. How would you get people to self-quarantine and practice social distancing? 

In my local area, Retirementville, Central Florida, residents should have been told how deadly this disease is to their age cohort (fifteen-percent for those seventy and older). Newspapers and radio could have emphasized the even elevated rates for people with conditions health conditions. In China it was men who smoked, but in the states it will be anyone with high blood pressure and cardiovascular conditions. 

Along with this public service announcement, we should have appealed to the patriotism of this particular part of Florida. We’d come together to defeat this microbial adversary. We could pass out stickers. It’s not Rosie the Riveter poster material, but it’s still a common foe.

Further, there’s enough technology to have virtual meetups. Let card games be on computers. Let people Facetime friends. With the right framing this would have been fun. People already have hurricane parties.

Had this been shared at the right time, things would be different.

Postscript, there’s probably something here too about distributions of outcomes. For the worst storms of the past thirty years, the median normalized damage is $26B and the average damage is $33B. How that data fits all hurricanes and compares to viruses is TBD.


Quarantine Psychology (HPTP)

One of our Quarantine Challenge items was to play every game we own at least once. Starting near the top of the this-will-take-a-while pile was Harry Potter Trivial Pursuit.

It’s typical for my wife and daughters to gang up on me. Mostly because I like the games so much, but in HPTP it’s them who have an advantage. Never-the-less, they pick on me and use a classic psychological study finding to do it. 

There’s a riddle that goes like this, a “A bat and a ball cost $1.10 in total. The bat costs $1.00 more than the ball. How much does the ball cost?”

Most people answer that the ball costs ten cents. People are thinking fast and thinking wrong. 

However, when the riddle is printed in a small or 𝔬𝒹𝚍 font most people answer correctly. When the riddle uses automotive brands, rather than a bat and ball, most people answer correctly. What gives? 

“A bat and ball…” leads most people to quick thinking that serves us well in many parts of our life but doesn’t do so well when we need to think slow. In HPTP this manifests in the way a person can ask the question. When my wife asked, “Which horcrux does Harry destroy first?” she emphasized the word “first.” She verbally bolded word. My twelve-year-old daughter slowed down and gave the right answer. 

The girls do this all the time.

But I have a few tricks too.

One that I use a lot is Gigerenzer’s recognition heuristic. Usually the first answer that comes to mind is the right one. But not always (see Austin or Fort Worth?)

I asked my wife, “In the first film, the library scenes were filmed at one of which famous English university’s library?” 

I only know two, Oxford and Cambridge. She only knows one, ‘Where the prince went’. Using the recognition heuristic, I’d have gotten it correct. 

HPTP is in the perfect zone of don’t-quite-know. For some answers I’m better off slowing down. For some answers I’m better off following recognition.