Category Archives: Logic

Storytelling for Children and Executives

The Wall Street Journal tells how the CEO of Procter and Gamble is more interested in the storyteller than in the powerpoint slides. Therefore presentations should have powerful stories.

Mr. Atkinson suggests organizing your story into three acts and starting by establishing context. You want to let your audience know who the main characters are, what the background of the story is, and what you’d like to accomplish by telling it, he says. You might open, for example, by describing a department that’s consistently failed to meet sales goals.

Move on to how your main character—you or the company—fights to resolve the conflicts that create tension in the story, Mr. Atkinson says. Success may require the main character to make additional capital investments or take on new training. Provide real-world examples and detail that can anchor the narrative, he advises.

The ending should inspire a call to action, since you are allowing the audience to draw their own conclusions about your story versus just telling them what to do. Don’t be afraid to use your own failures in support of your main points, says Mr. Smith.

Whatever you do, don’t preface your story with an apology or ask permission to tell it. Be confident that your story has enough relevance to be told and just launch into it, says Mr. Smith. Confidence and authority, he says, help to sell the idea to your audience.

The idea here is not new. Humans are more receptive to stories than to data. The powerful message that was omitted, however, was that while people are more receptive to anecdotes than boring powerpoint presentations, good decisions are made based on information. A story backed up by data combines the power of human psychology with the power of knowledge.


Income Inequality in NYC Extremely Depressing

New York Times reports the gap between rich and poor in NYC is widening further. Median income for the bottom fifth was $8,844. Median income for the top fifth was at a staggering $223,285.

What can be done? Sam Roberts of the New York Times interviewed Jilly Stephens, executive director of City Harvest, which helps get emergency food to hungry New Yorkers. jilly Stephens is in a unique position, because she runs a charity for the poorest in NYC, while she takes a salary for herself that places her in the richest fifth, earning $294,528 in total compensation in 2010. Her take on the problem? “The statistics demonstrate quite clearly that our most vulnerable neighbors are far from a recovery.”

Yes, she should know.

Book review: “The Laws of Simplicity” by John Maeda

Simplicity is a good thing, says John Maede, a professor at the MIT Media Lab.

“The Laws of Simplicity” presents ten rules and three axioms on how to achieve simplicity:

1. REDUCE everything when possible; use the process of SHE (Shrink, Hide, Embody)

  • Shrink – have more functionality in smaller form-factors
  • Hide – subsume secondary functionality in larger categories (e.g., navigation menus )
  • Embody – imbue item with a feeling of value and quality despite simplicity (e.g., a Bang & Olufsen remote control is very simple but also very heavy)

2. ORGANIZE things into categories makes more thins appear to be fewer – this coincides with the “Hide” above

3. TIME – saving time makes things simpler. Why? because there’s less happening

4. LEARN – the more you know about something, the simpler it appears.

5. DIFFERENCES – Contrast makes things look simpler

6. CONTEXT – focus isn’t always a good thing; carefully consider what might appear peripheral to see how it can create CONTRAST

7. EMOTION – sometimes emotion dictates adding more (versus REDUCE). Ornamental components can be a good thing.

8. TRUST – Trust leads to simplicity

9. FAILURE – some things will defy all attempts at simplification


Some of the “Laws” aren’t laws that lead to simplicity; some laws are contradictory; others are redundant. Here’s how I would reformulate them:

1. ORGANIZE AND SUBORDINATE – structured information is easier to digest than unstructured data; secondary functionality should be hidden behind larger categories

2. REMOVE unnecessary or nonessential elements. Once you’ve done that, wait a day and then REMOVE some more

3. SHRINK AND CLARIFY remaining elements until they are intuitive; summarize; take out distracting details

4. LIMIT DEPTH – remove elements that are too “deep” into hierarchies/categories, since no one will ever find them and they will just add confusion

Friendship Causes Obesity

In a groundbreaking study, researchers at Harvard Medical School used social networking techniques to track the spread of obesity. They found that even though obesity is a non-communicable disease, risks for becoming obese could nearly triple in some instances, solely based on relationships one has with obese people. Having an obese brother or spouse makes you 37% more likely to become obese in the next 2-4 years; having an obese friend can make you up to 171% more likely for you to become obese yourself.

The Harvard researchers believe this is a causal relationship:  that the obese friend causes you to become obese as well. If this is true, then those politicians in favor of “family values” or “friendship” may soon be in a conundrum – since such tendencies are likely to increase the spread of obesity.

Of course, other scientists note that correlation is not causation: it may be more likely that there are other environmental factors at play that merely correlated with social networks. If friends like McDonalds – then should we blame the friends for eating with us there or should we blame McDonalds for serving fattening food?

Prediction…or Wild Guess? (Decision Analysis Part 2)

In the recent bestselling book The Black Swan, Nassim Nicholas Taleb argues that traditional models focus on predicting events that stay within a “normal” range – that is, outliers and extremely rare events are excluded from the analysis and therefore are not predictable.ans, he argues, alter history with great frequency but we tend not to recognize their importance and rationalize them away post-facto. His hedge fund takes advantage of the rare event of a stock “exploding” – much the way a venture capital firm bets on many startups with the hope that one will become a Google.

The Economist reports on a study showing our inability to predict recidivism in criminals – even within a group of criminals who are very likely to become repeat offenders, there is wide variation in any individual criminal’s likelihood. What this means is that although we may be able to classify someone into a high-risk group, we still don’t know how likely he/she is to commit a Black Swan event .

In an article in Scientific American (“Shaping the Future,” April 2005), researchers at RAND Coporation and decision anlaysis firm Evolving Logic describe a second way of making predictions…

Rather than building a model based on one (or a few) scenarios and optimizing to find the result with the highest expected value, what if we optimize to choose the result that is most robust? I.e., when dealing with policy affecting global environment and international economics, the best policy will be one that can *never* result in widespread destruction. We can’t build a portfolio on worlds the way a venture capitalist can, so when we bet the farm on a single set of policies, we should prefer a policy that eliminates the possibility of World War III or widespread famine over a policy that offers the possibility of a utopia (but might end in World War III instead).

When we look at analysis – we must ask whether we want to predict the unusual event or the mundane event, and in the latter case what ramifications the Black Swan event can have on our business or our life.

To err is human… except in medicine

Physicians are held to higher standards than anyone else. To some extent, this is natural – lives are at stake. However, society and law has taken this to a ludicrous extreme, with the expectation that any mistake that happens is due to negligence. When any undesirable outcome occurs, the blame is placed squarely upon the doctor, who it is assumed has “botched the job.”

Who Botched the Job

Even by a higher standard, doctors cannot be expected to be infallible. Not even machines work perfectly. Error rates can be very, very low – as low as 1 per million – but the chance error is not just possible, it is inevitable.

Many studies have been conducted to identify error rates in medicine, but let’s take one such study as an example: that for every 1000 prescriptions written, there are four errors. Is this negligence? No. This is the inevitability of probability. No one can be perfect, no matter how hard they try.

Now, there may be ways to reduce the error rate. Cajole doctors into writing neater, require a second doctor to review the prescription before it can be filled, develop a computerized system, etc. But any technique in quality control is about *reducing* error – not about *eliminating* error.

Consider, then, the malpractice lawsuit. A patient had an adverse outcome to a prescription written in error. The error was obvious post-facto. That patient happened to be one of the four unlucky ones. Did the doctor botch the job – display gross negligence, even incompetence – because she, as an average doctor, has an error rate on par with average?

The legal system has turned medicine into a casino. Errors happen, often not because of negligence but because of statistical inevitability. But when statistical inevitability is treated as negligence, when the average doctor is sued three times over the course of her career, the entire field begins to take a defensive posture, which ultimately reduces efficiency and quality of care.


Why Good Decisions are Hard to Make (Part 1)

What is a difficult decision that you needed to make recently?

There are any number of reasons why a decision might be difficult…

  1. The problem lacks structure, and it’s not clear what the distinct options are or what the possible outcomes may be
  2. There is no quantifiable data that can be used to create a rigorous analysis
  3. There is too much data – which leads to contradictions or confusion
  4. After going through a logical process of reasoning, the conclusion seems counterintuitive. Should you believe the results?
  5. Not able to reach consensus among multiple decision-makers

In the last fifty years, the field of decision analysis has emerged as a viable technique for structuring problems so that they can be analyzed through a well-defined process (c.f., Big Sky Associates).

The science of decision analysis makes it a mathematical problem of solving an equation to find the right answer; however, there still remains an art: clarifying ambiguous problems and creating consensus among stakeholders.

The Language of Mathematics

The esoteric fundamentals of mathematics originate in set theory and at its most fundamental level – symbolic logic. Theoretically, one could use symbolic logic to express all formulas and functions in mathematics. And much of what can be encoded into symbolic logic can be translated into mathematical notation as well.

Mathematics is a tool through which complex logical relationships can be encoded. It is not the only tool. Computer programming languages, for example, also use syntax that originates in logic but builds in a different direction.

A primary requirement for a field to move from a “soft science” to a “hard science” is the ability to develop a rich symbolic notation that can be used to encode, store and modify the information releveant to the field of study. As subjects such as psychology and biology move from the “soft science” category to the “hard science” category, they change from a reliance on jargon to an adaptation of symbolic notation from other disciplines (e.g., statistics, computer algorithms).

Ultimately, a discipline can mature only when it develops its own symbolism. The suitability of mathematics or computer science can only go so far, and a hodge podge of semantic tools will constrain advancement of a discipline.