Monthly Archives: March 2018

Articles I Like: The Law of Unintended Consequences: Shakespeare, Cobra Breeding, and a Tower in Pisa

The PolyBlog
March 4 2018

This is another article from Farnam Street, and I confess up until a few days ago, I’d never heard of them. Run by a guy named Shane Parrish, he’s based here in Ottawa. Some really fascinating stuff on there, with decent curation and a lot of links. This article highlights that:

Not all of our grand schemes turn out like we planned. In fact, sometimes things go horribly awry. In this article, we tackle unintended consequences and how to minimize them in our own decision making.

The Law of Unintended Consequences: Shakespeare, Cobra Breeding, and a Tower in Pisa | Farnam Street

You might think that the article is going to be about train wreck ideas or the butterfly effect causing tsunamis. Not really. In fact, I would say it is more about linear thinking from good intentions to good outcomes, without taking into account side effects. Some unknown, some unforeseeable, some just missed because they stopped thinking early. The article has a great quote from a book by William A. Sherden:

Sometimes unintended consequences are catastrophic, sometimes beneficial. Occasionally their impacts are imperceptible, at other times colossal. Large events frequently have a number of unintended consequences, but even small events can trigger them. There are numerous instances of purposeful deeds completely backfiring, causing the exact opposite of what was intended.

The conclusion is simple — systems thinking or second-order thinking is needed, but the article doesn’t pay much attention to the fact that often the culprit lies in defining the system too narrowly, when in fact the small system is part of a larger system, and it is the larger system that often has the other effects (like the examples of releasing a predator into a land to control one local population, not realizing that the predator will spread into the larger system). What I do like is the idea that sometimes the failure is in over-estimating the size of the system, assuming there are too many variables, and thus not trying at all to figure out ancillary effects.

Yet, if we know they exist (or in hindsight think we should have), the article explains some of the most common reasons:

Sociologist Robert K. Merton has identified five potential causes of consequences we failed to see:

Our ignorance of the precise manner in which systems work.

Analytical errors or a failure to use Bayesian thinking (not updating our beliefs in light of new information).

Focusing on short-term gain while forgetting long-term consequences.

The requirement for or prohibition of certain actions, despite the potential long-term results.

The creation of self-defeating prophecies (for example, due to worry about inflation, a central bank announces that it will take drastic action, thereby accidentally causing crippling deflation amidst the panic).

However, the article goes even further, adding in over-reliance on models and predictions (mistaking the map for the territory), survivorship bias, the compounding effect of consequences, denial, failure to account for base rates, curiosity, or the tendency to want to do something.

Of course, the article leads to the article I shared earlier (Articles I Like: Mental Models – The Best Way to Make Intelligent Decisions (113 Models Explained)), and the use of other mental models to help prevent a failure to consider other effects.

Cool stuff, love the site.

Articles I Like: Mental Models – The Best Way to Make Intelligent Decisions (113 Models Explained)

The PolyBlog
March 3 2018

If you’re interested in goals and theory the way I am, then an article about “cross-training for the mind” and different ways of thinking in various disciplines is like catnip. When I saw the article, and that it was going to work through 113 different mental models, I couldn’t NOT click on that bait. In fact, their goal in the article is based on the following:

The overarching goal is to build a powerful “tree” of the mind with strong and deep roots, a massive trunk, and lots of sturdy branches. We use this tree to hang the “leaves” of experience we acquire, directly and vicariously, throughout our lifetimes: the scenarios, decisions, problems, and solutions arising in any human life.

Mental Models: The Best Way to Make Intelligent Decisions (113 Models Explained)

The more mental models you have, the more roots and branches you have to build on. What types of models? How about:

General Thinking Concepts (11)

  1. Inversion
  2. Falsification / Confirmation Bias
  3. Circle of Competence
  4. The Principle of Parsimony (Occam’s Razor)
  5. Hanlon’s Razor
  6. Second-Order Thinking
  7. The Map Is Not the Territory
  8. Thought Experiments
  9. Mr. Market
  10. Probabilistic Thinking (See also: Numeracy/Bayesian Updating)
  11. Default Status

Numeracy (14)

  1. Permutations and Combinations
  2. Algebraic Equivalence
  3. Randomness
  4. Stochastic Processes (Poisson, Markov, Random Walk)
  5. Compounding
  6. Multiplying by Zero
  7. Churn
  8. Law of Large Numbers
  9. Bell Curve/Normal Distribution
  10. Power Laws
  11. Fat-Tailed Processes (Extremistan)
  12. Bayesian Updating
  13. Regression to the Mean
  14. Order of Magnitude

Systems (22)

  1. Scale
  2. Law of Diminishing Returns
  3. Pareto Principle
  4. Feedback Loops (and Homeostasis)
  5. Chaos Dynamics (Sensitivity to Initial Conditions)
  6. Preferential Attachment (Cumulative Advantage)
  7. Emergence
  8. Irreducibility
  9. Tragedy of the Commons
  10. Gresham’s Law
  11. Algorithms
  12. Fragility – Robustness – Antifragility
  13. Backup Systems/Redundancy
  14. Margin of Safety
  15. Criticality
  16. Network Effects
  17. Black Swan
  18. Via Negativa – Omission/Removal/Avoidance of Harm
  19. The Lindy Effect
  20. Renormalization Group
  21. Spring-loading
  22. Complex Adaptive Systems

Physical World (9)

  1. Laws of Thermodynamics
  2. Reciprocity
  3. Velocity
  4. Relativity
  5. Activation Energy
  6. Catalysts
  7. Leverage
  8. Inertia
  9. Alloying

The Biological World (15)

  1. Incentives
  2. Cooperation (Including Symbiosis)
  3. Tendency to Minimize Energy Output (Mental & Physical)
  4. Adaptation
  5. Evolution by Natural Selection
  6. The Red Queen Effect (Co-evolutionary Arms Race)
  7. Replication
  8. Hierarchical and Other Organizing Instincts
  9. Self-Preservation Instincts
  10. Simple Physiological Reward-Seeking
  11. Exaptation
  12. Extinction
  13. Ecosystems
  14. Niches
  15. Dunbar’s Number

Human Nature & Judgment (23)

  1. Trust
  2. Bias from Incentives
  3. Pavlovian Mere Association
  4. Tendency to Feel Envy & Jealousy
  5. Tendency to Distort Due to Liking/Loving or Disliking/Hating
  6. Denial 
  7. Availability Heuristic
  8. Representativeness Heuristic
    1. Failure to Account for Base Rates
    2. Tendency to Stereotype 
    3. Failure to See False Conjunctions
  9. Social Proof (Safety in Numbers)
  10. Narrative Instinct
  11. Curiosity Instinct
  12. Language Instinct
  13. First-Conclusion Bias
  14. Tendency to Overgeneralize from Small Samples
  15. Relative Satisfaction/Misery Tendencies
  16. Commitment & Consistency Bias
  17. Hindsight Bias
  18. Sensitivity to Fairness
  19. Tendency to Overestimate Consistency of Behavior (Fundamental Attribution Error)
  20. Influence of Authority
  21. Influence of Stress (Including Breaking Points)
  22. Survivorship Bias
  23. Tendency to Want to Do Something (Fight/Flight, Intervention, Demonstration of Value, etc.)

Microeconomics & Strategy (14)

  1. Opportunity Costs
  2. Creative Destruction
  3. Comparative Advantage
  4. Specialization (Pin Factory)
  5. Seizing the Middle
  6. Trademarks, Patents, and Copyrights
  7. Double-Entry Bookkeeping
  8. Utility (Marginal, Diminishing, Increasing)
  9. Bottlenecks
  10. Prisoner’s Dilemma
  11. Bribery
  12. Arbitrage
  13. Supply and Demand
  14. Scarcity

Military & War (5)

  1. Seeing the Front
  2. Asymmetric Warfare
  3. Two-Front War
  4. Counterinsurgency
  5. Mutually Assured Destruction

The article has lots of links to the models to explain them. It’s like a treasure-trove of mental improvement rabbit-holes. And perhaps the grounds for 113 new blog posts by me as I work through each of them! Mind-blowing.

Reading “Big Box Reuse” by Julia Christensen – Chapter 3

The PolyBlog
March 2 2018

Chapter 3 is an interesting chapter on Wisconsin Rapids. The town profile is basically that of an old mill town, with a huge philanthropy base from two key families in town who owned the mills in days gone by, and a mostly summer tourism influx. Other than that, it is has a strong aging population and huge summer crowd, with upwards of 20% seniors in the general population.

The big box in question is again a former Walmart, and as with the examples that belied my original expectation of “out of business” big box stores, this is another one where the initial store was successful and eventually moved to a bigger store in other location, leaving the previous one sitting empty. What made this a bit unique in my view though is that the Walmart is relatively “downtown”.

So here’s the basic skinny…Walmart left, and a shell remained. It had a leaky roof, but the rest of the place was structurally sound. A community group was trying to build a seniors centre where three large service providers could co-locate to serve mostly shared clientele across their base. Yet their first instinct was not to occupy the old box space but to build something entirely new. When they couldn’t secure funding for that, they looked at the Walmart space and found ways to reconfigure it to attract funding.

For example, one of the regional groups was more likely to fund them if they were revitalizing an existing space than building new. Public support would be key to all the funding options, and a huge effort was made by media, government, advocacy groups to get everyone on board. Many were opposed to it being “Walmart-quality” and the optics, but once the designs were in place, it seemed viable. In the end, they had pretty creative financing.

Reading the chapter, it is obvious that “something” was going to happen, the question really was “where”. In the end, the Walmart space was way more costly than building new, but the redevelopment aspects attracted different sources of money. But for me, I found three really interesting factors to be:

a. The importance of the aesthetic redesign so people would stop seeing it as the Walmart space;

b. The renting out of space to non-retail renters which allow the three core service providers to basically generate some income to cover usage costs for other parts of the building (i.e. sustainability); and,

c. The creative idea to tear down PART of the building so that the remaining space would all be used, and there wouldn’t be the appearance of “empty unused space” as the facility was actually bigger than they needed.

This example comes closest to what I hoped for in the initial premise of the book — examples where an existing SPACE gets repurposed and flourishes (as opposed to simply repurposing land or putting in a different retailer).

Onward…

Articles I Like: The Lottery Hackers – The Huffington Post

The PolyBlog
March 2 2018

Usually when you see an article or some link about “beating the lottery”, you know it’s going to be a scam site, so there’s no point in clicking on it. You can’t beat the lottery, you just can’t. Right?

That’s certainly the popular wisdom, and the attached article (I used to hate the Huffington Post, but it’s become a little less sleazy and irresponsible of late, I find) walks through some of the history of how it is just a tax on the poor, the underrepresented, etc. And if someone found a way to “beat” the system, it would be illegal, right?

Well, apparently not. Based on some articles that ran in the Boston Globe, a down-home, blue collar guy in a white collar job (working for a cereal company on packaging), was fond of puzzles. And he loved math, so he would read, take courses, constantly learning new things. And then, one day, he was reading an ad for a new state lottery when he noticed something odd. It was your standard “pick-six” numbers out of a possible 49 for up to a $5M pay-off. Nothing unusual there. But they were adding a feature — if the big prize wasn’t won, it would roll-over, and roll-over (similar to most lotteries) BUT with one key difference — after a few weeks, if the big prize wasn’t won, they would do something called a “roll-down”. They would take all that big prize money and pro-rate it across all the smaller prizes. So, for example, the smallest prize of $5 could win $50 that week, if nobody won the big prize.

While that seems like no big deal, it drastically alters the math for your “return on investment”. So if you think of your chance of winning $10 in a game that only has ten tickets that cost a $1 each, then your expected return for your dollar “bet” is only $1. How does that work:

Expecting winnings = the prize money x the likelihood of winning = $10 x 1 in 10 = $1

So, statistically speaking, if you played over time, you would come out even. You would pay a $1 and expect to get a dollar back. Even Steven. And lotteries take that into account when they design the games. The math is NEVER in your favour. For example, your normal “return” calculation would look like this:

Expected winnings = HUGE prize x low odds of winning = $50M x 0 = $0

It isn’t zero, admittedly, but it is so low, it doesn’t change your payout calculation other than to say your return would be somewhere around one-thousandth of a penny. Over time, you would be guaranteed to lose money. Lotteries are rigged so the house always wins and suckers can’t game the system.

However, the rolldown would change that calculation, if for example, every sixth game, the payout was $20:

Expected winnings = $20 x 1 in 10 = $2

Or, put differently, if you could buy all ten tickets for $1 each for $10, your guaranteed payout would be the $20 and you’d be up $10. The math works because the winnings that week are NOT based on your normal return, they add in winnings from previous weeks that went uncollected. This means the state isn’t losing money — they already got their take. This is more like previous people didn’t win, so you can win THEIR money as well as the money from this week.

The problem though, in a state lottery, or any lottery where there are millions of combinations of tickets and millions of players, you can’t buy all the tickets, of course, but you also can’t buy enough tickets to even out random chance. So in the above example of $20, if you only buy 1 ticket, then your odds of winning don’t change, and it could take ten times before you “hit” — on average. But it could be 20 times or 2 times…if it is only twice, you’re way ahead. If it is 20, you come out down $10. The example in the article is with coin tosses, but the basic idea is that you need enough tickets to offset the random fluctuations of chance so that your investment matches the statistics (i.e. you need enough coin tosses for statistics to prevail).

How many tickets? The more you buy, the more it evens out the fluctuations. The main guy in the article starts at $3500 on $1 tickets. And he lost $150 or something. Next time, he went larger, $10K, $15K, etc., and evened it out. So he was making almost 50% return. Then he took on investors and jacked it up to $100K and more.

But the time investment was huge — he had to stand at a terminal all day long printing tickets. And only for “roll-down” pots. It wsa the only time the payouts were in your favour. And over time, the lottery officials would notice and kill it.

So he started playing another similar game in another state, and the newspaper article profiles other “investor” groups who noticed the same design flaw. However, to be clear, they weren’t cheating. They were just doubling down their bets when they knew the odds and payouts were more favourable. They weren’t rigging the game (although one group did that a bit, albeit not illegally). They weren’t cheating. They also weren’t anonymous — the lottery knew what they were doing and wasn’t stopping them. Because they were playing like everyone else — press the button, buy some tickets, take your chances. They were just doing it on a MASSIVE scale. Which the lottery officials didn’t mind because 40 cents of every bet was going into revenue for the state. The tax part of the winnings.

Overall, a really cool article. Even if the HP is mostly piggy-backing on stories written elsewhere, it’s decent reading.

Articles I Like: 10 breakthrough technologies this year

The PolyBlog
March 1 2018

Technology Review has released their list for “10 Breakthrough Technologies” for 2018. It’s hard to argue with the list having some important developments in it:

  • 3-D printing with metal — this could drastically disrupt manufacturing and give rise to lighter, stronger parts;
  • Artificial embryos — not exactly coming to a lab near you, but basically creating an embryo from another cell without an egg or sperm…great for research, but the ethical issues haven’t been worked out;
  • Smart-design for urban settings — using sensing technology and integrating tech into high-end design has always been part of the “future” in various sci-fi movies, but Quayside in Toronto will make some of it a reality;
  • Dueling neural networks — computer AI’s are bad at “creating”, but new techniques teaching them to learn off each other is creating a pseudo creativity with amazing applications for modelling, virtual entertainment, design, etc.;
  • Babelfish earbuds — auto translation in an earbud is great in theory, but I’m not convinced it will move out of the tourist zone as rapidly as some claim, particularly as early designs by no less than Google have been pretty clunky;
  • Zero-carbon natural gas — obviously, it’s still a non-renewable fuel, but having a clean version with no GHG emissions would be amazing, even if “Net Power’s technology won’t solve all the problems with natural gas, particularly on the extraction side. But as long as we’re using natural gas, we might as well use it as cleanly as possible.”;
  • Perfect Online Privacy through zero-knowledge proof — the idea is that you can provide “proof” of something (age, financial balance) without actually providing access…not quite a simple “proxy”, more like a cryptographic tool that says “You want to check if that record over there shows the person is over 18? Let me ask it”, and rather than performing the check itself, the cryptography gets the yes/no without seeing the original data…kind of like PayPal on steroids, but that doesn’t solve all the privacy issues online, it just makes the anonymous transparency of blockchains a bit more practical;
  • Genetic Fortune-Telling — the ethical issues of using DNA to predict health issues or even IQ are ridiculously bad, and based on the discrepancies in DNA testing for geneology, it can make economics look like a pure science; and,
  • Quantum leaps — building quantum computers is one thing, figuring out what to do with one is another…but modelling of molecules for design seems like a great first use.

However, for me, the one “breakthrough” that I think will affect us the most is the one the magazine dubs “AI for Everybody”:

Artificial intelligence has so far been mainly the plaything of big tech companies like Amazon, Baidu, Google, and Microsoft, as well as some startups. For many other companies and parts of the economy, AI systems are too expensive and too difficult to implement fully.

Machine-learning tools based in the cloud are bringing AI to a far broader audience. So far, Amazon dominates cloud AI with its AWS subsidiary. Google is challenging that with TensorFlow, an open-source AI library that can be used to build other machine-learning software. Recently Google announced Cloud AutoML, a suite of pre-trained systems that could make AI simpler to use.

Microsoft, which has its own AI-powered cloud platform, Azure, is teaming up with Amazon to offer Gluon, an open-source deep-learning library. Gluon is supposed to make building neural nets—a key technology in AI that crudely mimics how the human brain learns—as easy as building a smartphone app.

…

Currently AI is used mostly in the tech industry, where it has created efficiencies and produced new products and services. But many other businesses and industries have struggled to take advantage of the advances in artificial intelligence. Sectors such as medicine, manufacturing, and energy could also be transformed if they were able to implement the technology more fully, with a huge boost to economic productivity.

Most companies, though, still don’t have enough people who know how to use cloud AI. So Amazon and Google are also setting up consultancy services. Once the cloud puts the technology within the reach of almost everyone, the real AI revolution can begin.

You’ll want to keep an eye on these 10 breakthrough technologies this year | Technology Review

My only disagreement with the last one is the timing. They argue it’s available now, partly based on things like Siri and Alexa invading homes. Combined with the dueling neural networks, there are great things to be accomplished. I just don’t think they’re as close as they optimistically project they are already.