Tag: education

Monte Carlo Mistakes

MIKELeave a comment

“I’m going to work until my next birthday,” a near retiree friend told me. We were having dinner and she’d already punted on retiring her previous birthday. But this next one would be it – unless she wanted to travel more. 

Alaska was last year, Galapagos is this year, Africa the year after.  “I met with a financial advisor who ran a Monte Carlo simulation,” she said, “if I save a little more and take vacations from my income rather than savings, it’s more likely I’ll be better off.” 

Monte Carlo simulations, I thought, that sounds like a task for an LLM. After entering some approximate figures and going back and forth, Gemini ended with this: The Median Outcome: Even with this safer spending, the median outcome is ending up with $10M+ at age 83. This highlights the “Saver’s Dilemma”—to guarantee you don’t run out of money in the worst 10% of scenarios, you usually end up with a massive surplus in the other 90%.

Humans are feeling creatures, and it was a good feeling to know the spending “floor”.  (See also: Maxims for Thinking Analytically, extremes). In almost all situations $X is your spending floor. 

Decisions are a constant tradeoff of: what kind of mistake are you willing to live with? 

When I returned to teaching the cheating whack-a-mole was a whole new level. Twenty years ago laptops were a minority, now Chromebooks are ubiquitous. 

Imagine you’re running a marathon against one other runner. You’re neck and neck through the first five miles. You pull ahead. But then you notice something. There’s another runner ahead. You misunderstood. You’re running a marathon against a relay team. 

That’s the whack-a-mole challenge. Pop one area, a fresh one arises. 

Teachers have the same choice as retirees running Monte Carlo simulations (and all of us!). What side do I error toward? There’s no bullseye because there’s no right answer, just blunders one way or another. 

In the classroom, I give grace. Punishing a kid who worked hard and had questionable (LLM?) answers is worse than catching the number of cheaters +1. Retirees aim for grace too, only from the market. 

It’s like an alarm system. Do you want it to go off at false alarms or miss real events? 

This is the same spirit at the heart of 4000 Weeks – mistakes are inherent, are we choosing which ones to make?

Texts from School

MIKELeave a comment

My daughter’s high school (where I teach) has a new software program this year called Parent Square. It’s an app/service that allows school messages to be delivered more consistently, more immediately, and (unfortunately) more often.

During our training meeting where the administration sung the app’s praises I couldn’t help but think like an economist: oh this is too easy.

We all want to be informed parents. Or do we? Do parents want to know everything going on their children’s lives?

Regardless of if we want it (I don’t think we should) – we can’t!

Around the same time as my frustration with so many notifications, Kris Abdelmessih was asked about how to be a trader. He gives helpful advice. He’s a great writer, a good dad, super smart, and kind online. But part of kindness is honesty, he wrote:

“It’s gonna sound maybe harsh, but I tend to think that if you’re gonna figure it out, you just kind of are. You’re gonna find what to read; you’re gonna find the right things. And it’s like, if you’re unable to do that meta work, you’re just not cut out for it.”

Don’t bring information to a design fight. Want to change behaviors? Make it easier to people to take actions. Want to not change behaviors? Give people (more) information about the world. The ones who want it will get it.

Zero Cost, Pure Upside

MIKELeave a comment

One of the (few) calculations for my Business and Entrepreneurship class is customer acquisition cost. I’m a fan.

Rewriting the text for the upcoming year forced me to reemphasize and reiterate the value of a low CAC. Especially in the world of social media marketing and influencer marketing where CAC is sometimes zero. It’s an incredibly powerful idea.

But anything zero or low coast can be. Especially when part of an equation that divides (zero by many) or one that multiples many by a low cost (such as almost zero).

Talking with Chicago Fed Chair, Tyler Cowen asked about this very thing.

COWEN: What would be an example of something with a marginal cost of zero?

GOOLSBEE: [laughs] Well, I don’t want to reveal anything about our operations and get myself in trouble about the Federal Reserve operations. If you look at marginal cost of zero things, opening meetings to include others and having folks work together, sharing of information can often have very low cost — if not literally zero — and strong benefits.

Maybe this is the engine of alchemy. Find something that costs nothing (or very very very little) and do more of that. If it’s valuable (or very very very valuable) do a lot more of that.

Related: Free Things.

They’re just Puppies

MIKELeave a comment

All dogs wanted to be good dogs, no matter how unpromising they seemed. You just had to help them find a way. – The Old Man

There were frustrations when I returned to the classroom. Pesky persistent complainers. Foot draggers. Sleepers. Phone concealers. Make up concealers. C’mon, we’re in this together.

But then, after threatened and assigned detentions. After exhausted days. After hallway gripes. I realized, they’re just puppies.

People are just like puppies. Both have basic states of operation given their genetics, rules of physics, body chemistry and so on. Then, they both learn rules.

Puppies are taught not to pee in the house. Puppies are taught to walk on a leash. Puppies are taught to socialize. We don’t get mad at puppies for being puppies – we once had a puppy that only chewed up the right shoe. ¯\_(ツ)_/¯ 

Everything my students did was learned – successfully!

Complaining got results. Procrastination worked. Sleep was fine.

Life got a lot better when I thought about students as puppies.

Don’t Shoot the Dog! It takes time, persistence, and work to train puppies. Oh, and it’s fun. Puppies are happy to learn things and it feels good to teach them. Just like students. Just like all of us.

They’re just puppies.

Teaching in Verbs rather than Nouns

MIKE2 Comments

There are two ways to teach chess, a master explained.

The first, is to teach chess principles that may apply to life. And the second, is to teach life principles that definitely apply to chess. 

“Could you give me an example of one such principle? Because I love in biology teaching not names, not using nouns, but instead teaching verbs. Because ultimately, if you want to understand, for instance, how the nervous system works, or the immune system, you teach the verb actions of molecules.

And the names of the molecules are important if you decide to go into that field professionally. But otherwise, the principles and verbs are what’s most important. So what’s an example of a principle of chess or a mode of action on the board that you think transfers?”

Yet we teach with nouns and employ nouns because nouns have a lower cost, even though they may not be the right tool for the job.

–

From Huberman Lab: Josh Waitzkin: The Art of Learning & Living Life, Jan 27, 2025
https://podcasts.apple.com/us/podcast/josh-waitzkin-the-art-of-learning-living-life/id1545953110?i=1000685629066&r=5936
This material may be protected by copyright.

Outgrind?

MIKE1 Comment

The big idea from Outlive is that most health advice centers around avoidance (smoke in the lungs, sun on the skin, cells overloaded with sugar).

The next chunk of advice centers around basic things: move some, cultivate relationships, eat nutritious foods. It’s two 80/20 systems stacked on each other.

But we think about the last mile because we like to feel like we’re doing something. We like to feel that action is progress.

Pretty good is pretty good.

On his great blog, Justin Skycak writes about this in terms of “grind”.

Think of grinding on a project 50% of the time. If you double that to 100% that’s a 2x increase. What might take two years now takes one. That’s a sizeable change.

But “If you’re pushing 80% of the time, then the multiplier drops to 1.25x. You’re getting fairly close to max capitalization.”

Pretty good is pretty good.

This might be a more optimal situation too because of another idea: you have to be consistent before you can be heroic. I often said this to myself during a lot of Zone 2 training during 2024 (ultimately this worked out for me, I set a PR in the half marathon of 1:32:30).

There’s not a lot of return for grinding away at something all the time relative to most of the time. Like with Outlive’s advice, pretty good is pretty good. And, having the extra wiggle room allows for other things like serendipity and consistency.

How much coding do you *really* need?

MIKE

There are some ideas where I feel more or less confident about being true, this is one where I am confident it is a thing, but I am not so confident I understand it.

How much should someone code? Is “=CORREL” enough? What about HTML? I use RSS and Markdown and emojis all the time but I don’t code any of those things. Yet, there’s a lot of talk about needing to code. You need to code to get ahead. You need to code to keep up. We need to code the code’s code! But coding is situational. Replace “code” with “cook” and its like, oh yeah, that’s pretty varied.

What we really mean when we say, “people should know how to code” is that people should be able to use tools to deliver value. Code is a tool that does a job. The appeal of code, says Richard Feynman, is that it’s so dumb it’s fast. Code can also be copied. Code is a tool that delivers value unrestricted to time or place.

There’s a lot of ways for code to be a tool, it depends on the level. Sometimes =CORREL in a spreadsheet cell is the right amount of code. A spreadsheet is code too. There’s a bunch of math behind =CORREL that someone doesn’t need to know. That function is a tool within a larger function, the spreadsheet. All these functions within functions are the User Interface (UI).

“Consciousness is more or less the UI for how your brain works. Much like a UI, it wouldn’t help if Microsoft Word made me code in the structure of a document. There are programs, like LaTeX, that allow you to do that, but they are a pain in the ass to deal with because you have to specify the underlying details. The whole point of a UI is that it is an abstraction that sits on top of all the stuff underneath. For the most part our conscious is the UI for what is happening in our minds.” – Kyle Thomas, Stoned Studies August 2021

This nesting-doll nature of code is to be expected according to Brian Arthur. Yes, he tells Jim Rutt, that it has gotten more difficult to repair the things we use but this is because, like code, we create sub-systems for convenience of use.

“One of the things that happens is that if some sub-system is used often, it might have say 54 parts to it. If it’s used again and again and again, over several models and years it becomes modularized, becomes its own thing and it’s separately manufactured and it may have a cover on it. It may not be accessible to amateur mechanics, and it may only be accessible if you’re trained by Mercedes or Audi or whoever it is. And you’re properly trained. You have the proper tools. So as the lesson here is that as inner parts are used again and again, in the same configuration, the tendency is that they become modules.” – Brian Arthur, The Jim Rutt Show, August 2021

What is code? What does it mean “to code”? It depends!

This is good news. Rather than “learn to code” we should focus on “learn to solve problems”. Many of those problems will require tools. Some of those tools will be code. It is these kinds of problems: needs repeating, needs scaling, fits-with-existing-modules where code will be the tool for the job. Sometimes that code will be deep in the nesting dolls. Sometimes that code will be a simple spreadsheet cell. How many layers of UI depends on the problem to solve, the job to be done.

–
Arthur is famous for his comments on the Increasing Returns Economy.

Also, this. It gets good around 7:00.

Search Tricks

MIKE1 Comment

One effect of all the great content creation is the long-tail effect. Most of what’s created, from business breakdowns to that seventies show, will only be consumed by a small number of people. The long-tail idea is also true for an individual. Any given day my consumption is family news, then local and regional, then a national service or two, my favorite feeds (related: The Three Ways to Spend Your Day) and then the long tail stuff.

I used to feel bad when good episodes appeared in my feed and I skipped them. That’s fine, it’s just a query away. Which brings us to today’s point: a few of my favorite internet tricks.

Twitter search is not great, but with a few search operators it gets better. Mostly this is from:@mikedariano “jobs”, which returns tweets mostly about jobs-to-be-done. This is especially helpful to do before tweeting at someone to see if it’s been addressed already.

https://twitter.com/davidklaing/status/1404314465403752450?s=20

Wikipedia. Google (IMO) has suffered due to the incentives. It’s not a big deal, but rather than having a higher trust threshold I now go right to Wikipedia for Wikipedia-style searches.

Reddit. In 1994 I was twelve and one of the best feelings was visiting a video rental store. There were super-interesting sections I could plumb all day, there were areas I had no interest (at that time), and a restricted section I did not investigate for fear of what was behind the beaded curtain and whether or not I could unsee what I saw. That’s Reddit. The best Reddit communities might be the best places on the internet.

Listen Notes. Nowhere is the long-tail evident more than Listen Notes, a podcast search engine. Recent deep dives into Sears, DTC, MTV, and behavioral science all yielded results I could not have Googled. After creating an account, add your query results to the Listen Later playlist and add that RSS to your podcast app. If that sounds complicated it was a bad explanation rather than a difficult process.

Crudely the future of work will be some dichotomy of I give computers instructions or Computers give me instructions. During the Sears research (via Listen Notes) I found out that their first mail-order system was terribly bad. One customer wrote to Sears asking for the sewing machine she’d ordered, she’d received four wrong ones. It was only when Sears centralized their operation in Chicago that the mail order businesses succeeded. In 2021 there are companies like Locus Robotics.

In the future Cal-Newport-Style-Work will be doing things computers don’t do. Computers solve predefined problems really well.

But computers aren’t creative. Computers can’t handle a bunch of conditionals. Computers can’t frame things. Computers don’t in Bob Pittman’s words, understand when this is another one of those. Using the internet well is using computers to do non-computer work.

There is a 70’s show, The Long Seventies Podcast, that’s pretty darn good. I listened to the oil crisis and MPAA episodes.

Second Order Parenting

MIKELeave a comment

“Perhaps the most important thing is supporting a kid’s sense of autonomy.” – William Stixrud

Peter Attia wants to be an Olympian. When he’s 100. When Attia first explained the idea, he worked backward. What should someone do now if they want to do something else later? If you want to be able to go to Disney World with your grandkids you better be able to walk eight miles now. Or some such thing.

Working backwards is a nice tool for solving problems. I want to get to n so first I need to get to n-1.

A similar approach is advocated in the book, The Self Driven Child. In that book William Stixrud and Neil Johnson ask parents to consider their future eighteen-year-old. Then, like Attia, work backwards and consider what a child needs to do at twelve so they are successful later.

Stixrud’s and Johnson’s ideas come down to four pieces of advice for parents:

  1. Offer help, not force 
  2. Offer advice, it’s their choice  
  3. Encourage children to make their own decisions
  4. Have kids solve their own problems, as much as possible 

It’s not so much about ice-cream for dinner, but it’s about setting some (wide, but safe) boundaries for children to operate within.

This has been hard because parenting is a bit of a wicked problem. There’s a lot of showy things a parent can do but might not necessarily be that helpful. Once a child is mentally and physically safe, what’s the next clear thing?

In the ERE book, Jacob Fisker shares a reverse fishbone diagram to show ‘net’ effects. The aim is to end up above the horizontal line. Eating a candy is a positive first order effect, it tastes good. But a negative second order effect (low nutrients, high sugar), and maybe even more (bad habits).

Early Retirement Extreme --- A philosophical and practical guide to financial independence -- Contents

Easy decisions, hard life. Hard decisions, easy life.

There are many things with negative first order effects (difficult conversations, certain exercises, working late/early) but which have positive n+1 effects and so they are worth doing. Personal finance follows this diagram for example. Parenting according to Stixrud and Johnson follows it too. It might not feel easy and it might violate the actions=progress maxim, but it’s difficult non-actions that help kids the most.

One personal instance is schoolwork. We’ve been distance learning and I’ve played a large role from checking work to answering questions. And, sometimes just telling my daughters the answers. Like junk food it’s easy at first but it violates both the fishbone approach and the self-driven child advice.

How much to change, I don’t know. But this isn’t baking. Like over-price or over-rated or over-indexed, the direction for gains is clear: my kids have to learn to live their own lives and it’s my job to support them.

Gaming Work or Working Games?

MIKELeave a comment

“We have every reason to believe,” said Jane McGonigal, “the future of work will be more like Fortnite than the kinds of office jobs that we have today.”

2020 results

I would wager that future response balances will have a COVID-19 influence to them. School for example, seems to be tilting more digital as both administration wants to appear technological, tools become cheaper to implement students, and external forces require it.