Categories
Uncategorized

The pool of tears

A lesson from distance learning.

To keep up with my kids I’ve been taking Khan Academy classes and in one, founder Sal Khan noted that when Abraham Lincoln was in law school he used Euclid’s geometric proofs as a test for understanding. Recounted:

“In the course of my law-reading I constantly came upon the word demonstrate,” Lincoln said. “I thought, at first, that I understood its meaning, but soon became satisfied that I did not.” Resolving to understand it better, he went to his father’s house and “staid there till I could give any propositions in the six books of Euclid at sight.”

That’s ambitious, and demonstrates how much of learning is not linear.

In this way online learning excels. If we need time we take time. If we’re done early we make things. We act like Lincoln. Like Naval.

This is hard to do in school, scheduled to the year, week, day, hour, and even minute. Compounding and confounding is that we are relative creatures. I don’t get it compared to the kids that do. In the same way we are spending by neighbors but not saving, we see those who excel and calculateaccordingtothat.

Online learning isn’t great but it’s not all bad either and we’ve shed a few fewertears.

Categories
Uncategorized

Gambling with Votes

Thirty days of September and October PredictIt markets.

Like Gambling with Covid19, betting markets can demonstrate probabilistic thinking. In that post we considered an idea from Matt (+EV) about Tom Brady’s potential passing yards.

In April Brady’s over-under yardage was 4,256, nicely inline with previous years of 4057, 4355, 4577, and 3554. However, Matt noted, there’s a lot more room under 4,256 than over it. Brady could get injured, retire mid-year, have a worse system, lose teammates and so on.

On Wharton Moneyball Cade Massey noted that the same idea can apply to modeling voting and prediction markets. In the FiveThirtyEight simulation (40,000 runs), Joe Biden wins eighty-seven times out of one hundred.

What’s the gap between 87 and 66?

  • Potential polling errors. 538 is an aggregation. Put another way, the level of awareness while driving one hour twelve times is not the same as driving twelve hours one time.
  • The Brady effect. There’s just more room for ‘something to happen‘ in one direction instead of another.
  • Matt’s Twitter handle +EV gives an idea too. It could be that Donald Trump’s odds to wins are less than a coin flip just not as bad as a single number on a roll of the dice. That middle area is the market.
  • People like betting favorites, public teams, and for the safety.

A neighbor invited me to a watch party on November third. Another challenges himself to go as long as possible without finding out the news (in 2016 he made it three days). I follow things loosely but thinking about it this way does feel sharp.

As Howard Marks says, it’s not so much what you buy as what you pay. Brady, for those interested, is on pace to go over.

Categories
Uncategorized

Baseline data

Want a daily email with one idea? It’s quick, fun, and sometimes helpful. Sign up here.

One of the coronavirus problems, one of any system’s problems, is lack of good data. When data is precise and simple it’s just a math problem. This is why we have to gamble with coronavirus.

In mid-March I started to feel kinda ill. Did I have it? Everything pointed to yes.

I’d traveled through airports. I felt congested and achy. The news talked more about coronavirus than allergies. Wait. What? The noise of the news made me overlook the color of my car, which was a nicely tinged yellow thanks to an above average pollen count in central Florida. 

My problem was that the ‘fifth vital sign’ had overtaken all the others. Or put differently, the only data I was using was highly subjective. Instead of continuing my confoundedness I started counting. 

IMG_5490.jpeg

Regularly tracking my temperature showed nothing to worry about.

The other potential problem at the the time was toilet paper. 

Well before we were storming stores and short sheets I had stocked up. But watching the paper pandemonium I had no idea how long our stockpiles would last. So, I counted. Our  conservative count is two rolls per person per month. Prior to counting, I’d never have known.

Now do emergency funds

Good data is an objective tool to use alongside the subjective. If we kinda feel ill, we can take temperatures. If we see toilet paper rolling out of stores, we can use a rule of thumb. If we’re worried about finances, we can compare spending to savings. Good data is the base rate, our adjustments are the subjective. 

In any quantitative field three things matter: counts, computations, and communications.

Without accurate counts, we know nothing. 

Without accurate counts and computations, we infer nothing. 

Without accurate counts, computations, and communications, we do nothing. 

Sometimes we jump the gun. We build a model and share it to the world. #dataisbeautiful. Sometimes though we just need to start at the beginning and count. 

Thanks for reading. 

Categories
Uncategorized

Hand Washing Update

bathroom bottle clean container
Photo by Pixabay on Pexels.com

We looked at hand washing design research because conditions matter. People are influenced by their environment, often more than they realize. In that first post we highlighted to:

  • Turn off the water, to feel less rushed.
  • Make bosses (attending physicians) clean their hands.
  • Use incentives to reward (or penalize).
  • Put the hand-cleaning area adjacent to the need-hands-clean area.
  • Create a social expectation.

That research maps well to the EAST framework. To change behavior make things Easy, Attractive, Social, and Timely.

There are two updates since then.

First, The Behavioral Insights team researched which infographics communicated the best. Comparing seven ‘how to’ posters from around the world on 2,500 UK adults they found that “bright infographics with the step-by-step procedure prominently displayed without too much accompanying text” worked best to communicate good hand washing steps.

However, this was a ‘what I say’ question on a ‘what I do topic.’ Instead of hand washing it could have been a personal savings infographic about spending too much on a car. Sure, people will confirm they know the information but what would they do? It’s an encouraging start but more needs done.

Second, Google Search Trends for ‘hand wash’ negatively correlates with coronavirus cases. A few years ago, Google Trends predicted the flu rates ahead of the CDC but in following years erred enormously. Researchers suggested it was because people aren’t great at diagnosing the flu. How many times have you gone to WebMD AND had the thing. This bodes well for  the hand washing research, which stepped over that obstacle of unfamiliarity.

This focus on hand washing is timely but it’s also generalizable. It’s any verb. Investing. Driving. Loving. Parenting. All of these things are affected by the conditions they exist in.

Thank you for reading and supporting.

Categories
Uncategorized

Parlay Maths

A gambling parlay is a bet where two or more things have to happen. Will you have coffee and eggs for breakfast is less likely—thus longer odds and higher payout–than just betting on one or the other.

And people love betting parlays. The most popular Super Bowl bet is the coin toss, and Americans bet seven billion dollars (legally) on the game. 

And casinos love people betting parlays. According to UNLV, sports books earn five percent on bets, except for parlays. On those bets casinos take 30%.

Why do bettors do so poorly? It’s a little too much psychology and a little too little numeracy. Bettors, said Rufus Peabody, love to bet for things to happen. It’s easier to imagine one outcome than all outcomes. It’s why the ‘no safety’ bet almost always has positive EV. 

Bettors also don’t consider the numbers in the right light. Two independent seventy percent events only both occur half the time. Let’s run with that.

According to smart air filters, a t-shirt-mask will stop 70% of an airborne bacteria which is smaller than the coronavirus. That’s good. But what if we parlay masks?

If I wear a mask a t-shirt-mask and you wear a t-shirt mask we’ve reduced the viral load ten-fold. Thirty-percent of thirty-percent is .09. 

The same math that makes parlays good for Vegas and bad for gamblers is what makes masks good for all of us.

I wore mine to the store for the first time. It felt kinda foolish. But then I did the math.

UNLV explains the casino win percentage as “Win percentage, or win as a percentage of drop, AKA hold percentage, the percentage of money wagered that the casino kept.”

Peabody also tweeted about this: 

Categories
Uncategorized

Framing Employment

Framing is so important because it’s a way to get ‘free value’. Things well framed are perceived as well done—and perceived value is all there is.

I ran this one question poll on Mechanical Turk, Amazon’s data collection service, and Twitter to see how people perceive the same news. Each option describes the US labor market from mid-March to mid-April.

The question was, which one of these is the best (or least bad) way to describe what’s happened.

Over the last month..

The most important data isn’t that one section of the pie is larger than another but that there are sections of the pie. If  “135 million people remained employed” took the king’s share there would be nothing to figure out. In the many slices though we get many ideas.

Jason Zweig demonstrated this magnificently in a recent WSJ column. Imagine you’re of a certain age with a certain income and a certain promise of social security. It’s likely more than you realize. Zweig wrote:

The $2,000 a month that you and your spouse will each receive in the future has a present value of $772,235, according to OpenSocialSecurity.com.

That’s roughly what it would cost an insurance company to provide each of you with a guaranteed, inflation-adjusted $2,000 monthly payment for the rest of your lives (assuming you file for retirement benefits at age 70 and your spouse at 62).

So your expected Social Security payments are like a giant phantom annuity—a bundle of inflation-adjusted bonds you don’t own but whose income you have the right to receive. The same is true—usually without the ability to keep pace with inflation—if you are fortunate enough to have a defined-benefit pension plan.

Much of poker is played by the number. Professionals fold many more hands than they play because the numbers tell them that. But people don’t like to feel like automatons. They want action. When Annie Duke started teaching clients how to play poker she had to reframe how they saw the game.

Duke’s insight was to get her clients to choose to play by the numbers. She appealed to their meta side. They had to see the analytical next to the emotional and make a choice from those two options.

People are relative thinkers and many many decisions come down to framing one thing against another. It works for news, marketing, home purchases, dinner options, and dates. It works for everything.

 

Categories
Uncategorized

Colossal Comprehension

IMG_5475.jpeg

This is the earth.

Part of our quarantine education was to get outside and make some scale drawings of our solar system.

We made our earth one roadway wide, about twenty feet in diameter and paced off two hundred yards and drew the moon. It was five feet wide. The ISS was seven inches from the earth’s surface.

It’s always challenging to consider the scale of the universe. It’s huge. It’s so huge that Mars was sixty miles away in our little universe.

Part-of-the-reason Einstein marveled about compound interest is because scale is really hard to understand. Once things scale up or down past the human perspective we just don’t quite get it. This came up on two recent podcasts.

First, Peter Attia spoke with his daughter about the coronavirus. It was an excellent, simple, good-for-kids episode. So how big (or little) is the virus?

“If were to cut one of your hairs, and you can barely see the edge when it’s cut, how many coronaviruses do you think we could line up on the tip of your hair when it’s cut?” Attia asked

A thousand viruses. That’s beyond the human scale of understanding.

One the other end of the spectrum, and closer to the solar system situation was Cade Massey’s longhorn lament.

“One of the things that frustrated me most when I to talk with people was them saying ‘Well, you’re not going to get this if you’re young.’ We knew the probabilities are steeply related to age but there’s still a probability for every age group. Throw millions of people at a small probability and you’ve got some sick people. We just aren’t good psychologically with these kinds of probabilities.” Cade Massey

The percentage for infection, hospitalization, and ventilation are remarkably small.

New York City houses eight million people and the metro area is home to twenty-one million. Projections note that only .27% will need beds, and only .063% will need ventilators.

Right now my sixth grade daughter is learning percentages as parts of the whole. She answers questions like; “If sixty percent of a class of twenty-four are boys, how many children are in the class?”

That’s good sixth grade math but it gets hard with large numbers. One-fourth of a percent is really small but eight million is really large. How does someone make sense of that? We probably just need to think slow, not fast.

 

Categories
Uncategorized

Numbing Numbers

On Epidemic, Ronald Klain talked about how long a shutdown may last.

“I’m asked this question when I’m on TV all the time, what’s the date, what’s the date? But this discussion about the date is the wrong discussion, the question is, what are the preconditions that we need to have in place before we can reopen large swaths of economic activity?”

That’s a harder question.

The CoVid19 situation is like a Sudoku board with very few numbers filled in. If that’s a nine this might be a four which makes that a two—shit that can’t work. There are so many interchangeable parts it’s easier to ask, ‘what’s the date?’

To get away from ‘what’s the date’ questions we can add one more small step, asking why.

‘Why’ gets us to answer.

For example, why is social distancing six feet? Is this a case like a power law where the bulk of the results come from one source? For example, when researchers looked at what size particles passed through what size fabrics, “0.02 micron Bacteriophage MS2 particles (5 times smaller than the coronavirus)“, a surgical mask stopped 89% of the particles, a vacuum bag 86%, and a cotton blend t-shirt stopped 70%. Not bad.

But when they doubled up, masks improved to stopping 89% and shirts to 71%. Small relative increases.

Is social distancing like that? Six feet is like wearing a mask made from a cotton shirt? Maybe not. The gas cloud research rather than aerosol or droplet research—the six feet origin work was done in the 1930’s—hints that viruses could travel twenty-seven feet in the air.

It’s hard to not recommend something other than ‘when we hear numbers we should ask why‘ but there’s so much ambiguity that’s all we can say with confidence. As for dealing with the here and now, here’s how to gamble with the coronavirus.

Categories
Uncategorized

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.

Categories
Uncategorized

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.