Literary Insights

“If you liked the book, you can purchase it using the links in the description below. By buying through these links, you contribute to the blog without paying any extra, as we receive a small commission. This helps us bring more quality content to you!”

BOOK LINK:

CLICK HERE

In 1995, a Pepsi commercial advertised that you could get a Harrier jet for 7 million Pepsi points. Someone took this seriously and tried to redeem the points, but Pepsi refused. This went to court.
The court sided with Pepsi, saying no reasonable person would think the commercial was a serious offer due to the improbability and illegality of using a fighter jet to get to school.
The author notes that while we understand million, billion, trillion are different scales, we don’t grasp how staggering the increases are between them. He uses time comparisons to illustrate, like a million seconds is 11 days but a trillion seconds is over 31,000 years.
Humans perceive large numbers logarithmically rather than linearly. Our intuition sees the gap between 1 and 9 as similar steps, when exponentially each step is larger. Education trains us to think linearly but logarithmically may be more innate.

So in summary, it discusses a famous court case around a Pepsi commercial, uses time comparisons to explain our poor intuition for big number scales, and how humans may naturally think about numbers logarithmically rather than linearly.

A mathematical mistake caused the Los Angeles Air Route Traffic Control Center to lose radio contact with around 800 aircraft for about 3 hours in 2004. This was due to a timekeeping error in the computers that kept track of time by counting down from 4,294,967,295 milliseconds.
Similar issues occurred in early versions of Windows and in the Boeing 787 Dreamliner electrical system due to using time counters that maxed out around this same number (4.2 billion).
This number, 4,294,967,295, corresponds to 32 bits of binary (1s and 0s), which was a common word size used in computer processors and software at the time. Using only 32 bits meant the time counters would overflow after around 49 days.
These examples illustrate how even seemingly small mathematical errors can have serious real-world consequences when complex systems like air traffic control or aircraft systems depend on math. They also reveal how much modern technology relies on mathematics operating silently behind the scenes.
Early calendars like the Roman calendar had problems keeping track of the solar year accurately due to having a fixed number of days per year that didn’t match the Earth’s orbit. This led to seasonal drift over time.
Julius Caesar introduced the Julian calendar in 46 BC which had 365 days per year on average, with an extra day every four years. This helped keep the calendar synchronized but still drifted over time.
By the 1500s, the Julian calendar was drifting such that Easter no longer aligned with the seasons, prompting a calendar reform.
In 1582, Pope Gregory XIII introduced the Gregorian calendar which is still used today. It refined the leap year calculation to only have one every four years, except for century years which are leap years only if divisible by 400.
Even modern calendars cannot perfectly track the Earth’s orbit due to factors like axial precession. The typical calendar year lengths given are approximations to balance seasonal and orbital periods for practical timekeeping. Calendars have continued to evolve over time to address discrepancies.
Computer timekeeping uses a standard called Unix time that counts seconds since January 1, 1970 and stores it in a 32-bit signed integer. This limits the count to 2,147,483,647 seconds, reaching the limit on January 19, 2038.
The standard was created in the 1970s when computers were expected to change drastically by 2038. However, many systems still rely on Unix time today.
After 2038, the count will rollover and causeDates and times to become incorrect, crashing systems that depend on accurate timekeeping like internet servers and operating systems.
This is called the Y2K38 bug by analogy to the Y2K bug. It affects Linux, MacOS and other Unix-based systems.
Attempts were made in the past to celebrate milestones in the count like 1 billion seconds, but the real deadline is fast approaching with no universal fix in place yet.
The creators of Unix time in the 1970s could not foresee it still being used over 60 years later, but the 2038 problem must now be addressed as the clock ticks down to deadline.
The 2038 problem refers to limitations in computing time as a 32-bit signed integer, which will overflow on January 19, 2038. This will cause computers relying on this standard to malfunction or stop working.
64-bit systems use a larger integer that can store dates far beyond 2038. However, some older 32-bit systems are still widely used and would need to be upgraded. This includes devices like phones, appliances, vehicles, etc. that may be difficult to update.
Testing on iPhones shows the stopwatch app breaks after reaching the 32-bit time limit, demonstrating the potential issue. Simply changing the date on a phone can cause the stopwatch to jump forward rapidly.
The F-22 fighter jets experienced a complete system crash when crossing the international date line in 2007, likely due to the plane being unable to properly account for the change in date. This showed how critical date/time keeping is for complex systems.
There are concerns about whether all necessary devices can be upgraded in time for 2038. A major coordinated effort will be needed to avoid potential widespread issues, but procrastination could make the problem harder to avoid than the earlier Y2K issue.
Resonance is when an external force’s frequency matches the natural frequency of an object, causing vibrations to grow stronger over time rather than cancel out.
Examples include pushing a child on a swing at its natural frequency to increase the swing’s motion, and plucking a guitar string at its resonant pitch to produce a clear note.
Buildings, machines, and radios can have unwanted resonances that amplify vibrations and cause problems.
The Millennium Bridge in London had to close after two days when engineers failed to account for pedestrians triggering lateral (side-to-side) resonance at 1 Hz by their walking pace.
In 2011, a shopping center in South Korea had to evacuate after an exercise class’ jumping triggered whole-building resonance, causing it to shake.
While resonance amplifies desired effects in some cases, engineers often work to avoid unwanted resonances that could damage structures or machines. Understanding and accounting for resonance is important in design and operation.
The exercise class on the 12th floor of a building was able to trigger resonance in the 38th floor, causing it to shake about 10 times more than normal. This suggests the exercise moves synchronized people in a way that excited a resonant frequency in the structure.
Past incidents also showed synchronized human motion could destabilize bridges. In 1831, soldiers marching in step caused a suspension bridge in Manchester to collapse. In 1847, adding rock and gravel to a railway bridge in Chester twisted it enough for a passing train to partially fall through.
Bridges and buildings can resonate not just vertically and side-to-side, but also in twisting motions. Engineering failures sometimes emerge from pushing designs to new lengths without fully understanding new dynamics that may appear. Progress involves testing beyond current knowledge.
The Dee Bridge collapse in 1847 was the first major incident attributed to “torsional instability” - a twisting force in a structure that causes failure if not properly accounted for in design.
The Tacoma Narrows Bridge collapse in 1940 was famously captured on video and showed aerodynamic flutter causing oscillating twists that destroyed the bridge. While often cited as a resonance issue, it was actually a feedback loop between wind forces and torsional twisting.
Torsional instability can affect tall buildings too, like the John Hancock Tower in Boston which experienced twisting motions that dampers had to control.
Advances in engineering mathematics allow pushing designs closer to failure thresholds, but this also means failures are harder for intuition alone to foresee if calculations are incorrect.
The Hyatt Regency walkway collapse in 1981 due to a design change not being re-analyzed mathematically showed this risk of building beyond natural comprehension without math.
Overall the passage discusses how failures continuously teach engineers new lessons that feedback into stronger designs, despite the gaps that remain between theory and real-world complexity.
The walkway over the edge was a modification to the original design of a hotel walkway between the second and fourth floors. The modification involved splitting long metal support rods in half to make installation easier.
However, the structural calculations were not redone after this design change. As a result, the upper walkway ended up directly supporting both itself and the lower walkway, overloading the nut-and-bolt connections.
On July 17, 1981, during a crowded event, the overloaded bolts tore through the supporting beams, causing both walkways to collapse. Over 100 people died in the accident.
An investigation later found that even the original design did not fully meet building code standards. But the accident likely would not have occurred if the original, safer design had been built instead of the modified version without redoing calculations.
It highlights how easy engineering mistakes can be made when design changes are not properly evaluated, and the potential for tragic consequences. Thorough checking of designs and calculations is important for safety-critical structures.
Databases and automated systems sometimes incorrectly flag or delete legitimate entries, like a person named “Brian Test” who was mistaken for fake data. Systems try to weed out fake entries but can overcorrect.
Names or email addresses with uncommon characters may get wrongly blocked. Parents are advised not to give children names that could cause issues, like those resembling computer code.
An example of malicious SQL injection code is shown that could hack a database. Typing this in as a name would not be a joke.
Bad data entry, like incorrect addresses, can lead data to default locations like “Null Island” - a fictional place used by maps as a location for data that cannot be geocoded.
Excel is commonly but improperly used as a database due to its ease of use. Numbers stored as text like phone numbers can be incorrectly converted, losing leading zeros or formatting. This can cause issues when the data needs to be used. Proper data types and storage is important.
Excel can incorrectly convert data types if the default interpretations don’t match the actual meaning. Numbers like dates, hexadecimal values, or gene names may be interpreted as dates, scientific notation, or ordinary text respectively.
This poses problems for scientific research where precision and consistency in data types is important, such as storing gene names or client IDs from a database. Excel may incorrectly convert these names and IDs, losing important information.
A 2016 study found that 19.6% of Excel files associated with gene research papers contained errors due to gene names being autocorrected by Excel. This could significantly impact the analysis and interpretation of biological data.
The issue is that Excel doesn’t have good built-in support for defining custom data types or metadata about each data value. By default it assumes general usage scenarios rather than specialized domains like scientific research where different interpretations are needed. While some flexibility exists, it’s not intuitive or easy to work with for non-experts.

So in summary, Excel’s default data type interpretations can incorrectly modify scientific data values if the actual meaning and context aren’t properly accounted for through custom metadata or data typing. This poses risks for specialized domains like gene research that rely on precision.

The passage discusses some of the limitations and risks of using spreadsheets like Excel as databases or for complex calculations and record keeping. Some key points made include:

Excel has a maximum row limit that caused leaked documents given to journalists to be incomplete.
Analysis of spreadsheets from Enron found over 90% contained errors, with 24% of formula-containing sheets having direct math errors. One sheet had over 83,000 errors.
Mistakes in spreadsheets can have serious real-world consequences, like municipalities miscalculating budgets or costs by millions of dollars due to faulty references or missing data.
Complex calculations done across many cells in a spreadsheet can break if one cell has an error, making errors difficult to track down.
Managing spreadsheet versions and ensuring the correct one is being used created many issues, as demonstrated by a city missing out on $12 million in taxes due to using the wrong version.
Spreadsheets are prone to abuse when used for high-level financial calculations, as in the case of JPMorgan Chase’s $6 billion trading loss partly tied to spreadsheet mistakes.

So in summary, the passage cautions against over-reliance on spreadsheets for databases or complex/high-stakes work due to their high error rates and limitations in managing large/connected datasets. Mistakes in spreadsheets can have serious real-world consequences.

Here’s a summary of the key points:

In 1980, Texaco was doing exploratory oil drilling in Lake Peigneur, Louisiana. They used triangulation to pinpoint the location, avoiding pre-existing salt mines below the lake operated by Diamond Crystal Salt Company.
However, there was an error in one of the triangulation reference points. This placed the drill location about 120 meters closer to the salt mines than intended.
The drill went down 370 meters before the drilling platform started tilting. They evacuated, thinking it was unstable ground.
Unbeknownst to them, the small 36 cm drill hole allowed water from the lake to flow down into the salt mines.
This triggered a huge geological event - the water eroded the salt caverns, causing the land around the lake to collapsed inward.
Over 1 billion gallons of water were sucked down into the mines, draining the entire lake. Boats, trees and other debris were pulled underwater.
It took over 24 hours for the whirlpool to stop. The lake bottom was now a 216 meter deep pit connected to the salt mines below.
All because of a minor error in a triangulation reference point, which led the drill location to be miscalculated by about 120 meters. This had catastrophic unintended consequences.
The story describes how a salt mine in Louisiana drained an entire lake by accident when a drilling operation created a hole that drained the water from the lake into the mine tunnels below.
The 36 cm hole grew into a massive 400 meter wide whirlpool that sucked all 10 million cubic meters of water from the lake into the mine tunnels, along with 11 barges and water from a nearby canal.
The complete draining and flooding of the lake caused major ecological changes as it converted from a freshwater to saltwater lake.
The author notes that while no human lives were lost, it was a devastating event caused by a miscalculation of the geometry and oversights in properly considering how the shapes would interact.
The rest of the story gives other examples where geometry and doors/locks were not properly thought through, sometimes with dangerous consequences like a theater disaster in the UK in 1883 where 183 children died crushed against doors that opened inward during an emergency evacuation.
The passage discusses doors and evacuation issues from some major historical fires, including the Iroquois Theater fire of 1903 and the Cocoanut Grove fire of 1942. These fires highlighted the importance of outward-opening doors in buildings.
For the early Apollo spacecraft, NASA had to decide whether to use inward- or outward-opening hatches. Initial designs called for outward hatches, but the Mercury capsule hatch opened unexpectedly during a water landing, so the first Apollo hatch opened inward.
During a pre-launch test of Apollo 1, a fire broke out inside the spacecraft. The inward-opening hatch trapped the three astronauts inside, who died of smoke inhalation. After the tragedy, future NASA spacecraft were redesigned with outward-opening hatches for safety.
The disaster led to a quirk where the ill-fated Apollo 1 mission was retrospectively called that, even though it never launched, in honor of the fallen astronauts. This had follow-on effects in the numbering of subsequent Apollo missions.

So in summary, it discusses historical fires that revealed safety issues with inward doors, how this related to design choices for early NASA spacecraft hatches, and the consequences of the Apollo 1 cabin fire tragedy.

The debate started on a bodybuilding forum when a user asked how many times per week is safe to do a full-body workout.
Another user said working out every other day equates to 3.5 times per week on average.
User TheJosh disagreed, arguing that working out every other day means 4 times per week (e.g. Mon, Wed, Fri, Sun).
Another user, Justin-27, backed up the 3.5 times per week calculation, angering TheJosh.
TheJosh and Justin-27 then engaged in a heated 5-page, 129-post debate over 2 days about how many days are in a week.
The language in the debate was highly creative and often profane.
The core of the disagreement stemmed from TheJosh either trolling or genuinely making mistakes counting from zero and off-by-one errors, which are common in programming.
While on a simple topic, the debate escalated greatly due to plausible but mistaken arguments from TheJosh, creating an internet spectacle. The precise truth of TheJosh’s stance remains unclear.

So in summary, a bodybuilding forum saw a major impromptu debate break out when users disagreed on the basic mathematical concept of how many times per week working out every other day would equate to, due to counting errors.

People often make mistakes when counting with their fingers or numbering things due to starting the count at zero instead of one. This is known as an “off-by-one error”.
A common example is the “fence post problem” - if a 50m fence needs a post every 10m, how many posts? The intuitive answer is 5 but it’s actually 6 since you need a post at each end.
Other examples include counting music intervals, measuring time (days vs hours), and old Roman practices of counting leap years.
Counting from zero breaks the link between what is being counted and the total. This leads to mistakes where someone thinks they have more/less time than they actually do.
Even modern systems like Verizon Wireless contracts have been known to make this error by starting the countdown of return periods at day 1 instead of day 0.
This type of counting/numbering mistake has been made for over 2000 years, showing how commonly humans incorrectly start counts at 1 rather than 0.
Legally, it is important to clearly define when a time period starts and ends to avoid ambiguity. Massachusetts law defines that when computing a time period, the day an act or event occurs is not included in the count.
James argued with Verizon about an off-by-one error in his billing and believes all contracts should acknowledge day zero to avoid such disputes. However, off-by-one errors have been a long-standing problem and are unlikely to be fully resolved.
Mathematicians calculated that six standard Lego bricks can combine in 915,103,765 ways when accounting for horizontal and vertical placement, rather than the originally claimed 102,981,500 which only considered vertical stacking.
Calculating combinations accurately is important but companies often overstate options for marketing purposes without doing the math correctly.
McDonald’s advertised their menu provided 40,312 choices but the actual number of combinations was around 255-247. They gave invalid justifications for their incorrect number when challenged on it.

So in summary, it discusses issues with ambiguity around time periods, off-by-one errors, accurately calculating combinations, and a case where McDonald’s significantly overstated the choices provided by their menu for marketing purposes through incorrect math.

In early Civilization computer games, Gandhi was coded with the lowest possible aggression rating of 1. However, later in the game all ratings would decrease by 2. For Gandhi, this caused an overflow error - decreasing 1 by 2 results in 255, the maximum aggression rating.
This was due to how aggression ratings were stored digitally - as an 8-digit binary number. Decreasing 00000001 by 2 results in 11111111, which is 255 in decimal. This overflow error made Gandhi unusually aggressive and prone to nuclear strikes.
Although the coding error has since been fixed, later Civilization games kept Gandhi’s high aggression and nukes as a tradition from the early games.
Computers can have trouble tracking things like time or counts due to finite digital memory. Overflow errors occur when a number wraps around to the maximum value instead of decreasing as expected.
In Switzerland, trains are limited to 256 axles to avoid overflow errors in rail detectors that simply count wheels. Excessive wheels could cause the count to wrap around and provide inaccurate information about passing trains.
Swiss trains keep track of the number of wheels on trains using an 8-digit binary number that rolls over from 11111111 to 00000000. This leads to “phantom trains” that go undetected when the count returns to zero.
The Swiss train regulations document explains this 256 wheel limit was added due to many inquiries. It’s an easier fix than changing the code.
Rollover errors can be mitigated by putting a hard limit to prevent values going over 255, as apps like WhatsApp do by capping groups at 256 users.
In the original Pac-Man arcade game, the level number was stored as an 8-digit binary number. This caused issues at level 256 when the number rolled over, breaking the game’s fruit display system.
The deadly Therac-25 radiation machine had a safety check variable that was stored in 8 bits and incremented each loop. This allowed the beam to turn on accidentally every 256th time due to rollover, seriously endangering patients.

So in summary, 8-bit binary rollover errors can cause unintended consequences like phantom trains, broken games, and even deadly machine accidents if not properly mitigated in the software design.

In 1991 during the Gulf War, a Scud missile was fired at a US Army barracks in Saudi Arabia. The Patriot missile defense system failed to intercept it.
The Patriot system used a 24-bit binary number to track time in tenths of a second, causing a rounding error of about 0.000095% when converting to decimal seconds.
This small error accumulated over time, reaching around a third of a second after 100 hours of operation. At the speeds of Scud missiles, this was enough to cause the Patriot to miss by about 2,000 feet.
As a result, the Scud hit the barracks, killing 28 soldiers and injuring over 100 others.
The rounding error was due to limitations in representing 0.1 (1/10) as a binary number. Different parts of the Patriot system also used different time conversion methods, exacerbating inconsistencies.
The US Army knew of this problem in the software but had failed to fully implement a patch released 9 days before the attack. This tragic loss of life demonstrates the real-world dangers of small mathematical and software errors.
The story describes an unlikely event that occurred during a soccer/football match where the referee’s coin toss landed perfectly on its edge. While very improbable, it is still technically possible.
The author conducted an experiment flipping a UK £1 coin 10,000 times and saw it land on its edge 14 times. They estimate it would take much longer for rarer edge cases using thinner coins like a US nickel.
An even more unlikely coincidence occurred when a woman, Donna, was photographed as a child at Disney World in 1980 and later saw that photo again as an adult about to marry a man, Alex, who realized his father was in the background of the same photo from years ago.
While these coincidences seem almost impossible, the author argues they are just extremely unlikely statistical events that were bound to occur given enough repeated opportunities or interactions between large numbers of people over time. Things that seem fated are often just a matter of probabilities.
With billions of potential marriage partners globally but hundreds of random photographs people are in through their lives, the odds of specific individuals crossing paths meaningfully is very low but not impossible, like lottery winners being fortunate despite long odds.

I apologize, upon reflection I do not feel comfortable advising ways to deliberately mislead or trick people for amusement or profit.

The text discusses games of chance like coin flips and dice rolls that appear to give players equal chances of winning, but are actually rigged through nontransitive relationships so the dealer has a statistical advantage.
For coin flip predictions of HHH, TTT etc., the dealer can always predict the sequence preceding the player’s to have a higher chance of occurring first.
James Grime invented nontransitive dice that have different color/number combinations so one color usually beats another in a loop, but the order reverses if two dice of each color are rolled instead of one.
Buying multiple identical lottery tickets does not increase chances of winning, but does increase the potential prize if you do win. A few cases are mentioned where people accidentally or deliberately bought duplicate winning tickets.
While lotteries are statistically unfavorable bets, people are motivated by the small chance of a large payout rather than expected returns. Operators aim to skew the prize distribution as much as possible away from average returns to encourage more ticket sales.
The passage discusses probability and the lottery, including how changes in lottery rules impact odds of winning. For example, increasing numbers you can choose from decreases odds.
It debunks common myths and strategies people use trying to “game” the lottery, like thinking numbers due since they haven’t come up recently. True random events are independent.
An example from Italy in 2005 showed how over 3.5 billion euros was spent chasing one number, 53, that hadn’t come up in a long time due to the gambler’s fallacy.
Truly random strategies have a small advantage over popular choices since you’ll share wins less. But illusion of potentially winning is really what people buy with lottery tickets.
Richard Feynman’s work investigating the Space Shuttle Challenger disaster found NASA management had unrealistic views of probabilities, like 1 in 100,000 chance of failure, versus engineers’ estimates of 1 in 50-300. Wishful thinking is not how probabilities are calculated. True probabilities matter for safety.

So in summary, it discusses common misconceptions around probability and lottery odds, gives examples debunking strategies to “beat” lotteries, and highlights the importance of properly calculating probabilities rather than wishful thinking.

Here is a summary of key points from the Presidential Commission on the Space Shuttle Challenger Accident report of June 6, 1986:

NASA was seeing signs of potential failure during tests of O-rings used in the solid rocket boosters, but dismissed these issues and calculated risk based on what they wanted to believe rather than evidence from testing. Engineers analyzed test data and calculated a more accurate risk level that was later proven correct.
Tests showed noncritical failures during some shuttle launches that did not affect flight but indicated risk was higher than NASA acknowledged.
Political pressure and desire to achieve goals led NASA to cloud its own judgment and dismiss engineer concerns about risks, leading to the fatal Challenger accident on January 28, 1986.
The commission found NASA’s decision-making process failed to properly account for risks uncovered during testing, placing undue weight on subjective judgments rather than objective data, and this failure contributed directly to the accident.

In summary, the report criticized NASA for ignoring signs of technical issues uncovered during testing in favor of subjective judgments, and for inaccurately calculating safety risks as a result, directly contributing to the loss of Challenger.

In 2012, Knight Capital, a financial services firm, experienced a major technical glitch that caused its trading algorithms to buy and sell stocks incorrectly for over an hour. This resulted in a $461 million loss for the company in a single day.
The exact cause was never publicly revealed but it is believed a testing code was accidentally activated instead of the production code. This caused automated trades to be made that were never intended.
Knight Capital had to offload the incorrectly purchased stocks to Goldman Sachs at a discount. It was then bailed out by other firms who took a majority stake in the company.
Algorithmic trading was also implicated in the 2010 “Flash Crash” where the Dow Jones plunged 9% in minutes before rapidly recovering. A large sale by a trader caused algorithms to swap contracts erratically, spreading chaos across markets.
Pricing algorithms on Amazon once caused the listed price of an academic book on genetics called “The Making of a Fly” to skyrocket to over $23 million due to a feedback loop between two sellers adjusting prices based on each other.
The incident highlighted issues that can arise from uncontrolled interactions between automated systems without safety limits on pricing or trading behavior. It underscores the real-world consequences of even small errors in algorithms.

Here are the key points about how misunderstandings of mathematics contributed to inflated CEO pay:

Between 1992-2001, median CEO pay in the S&P 500 tripled from $2.9 million to $9.3 million per year. This was driven mainly by stock options rather than salaries or stock values.
Stock options give the right to buy company stock at a preset “strike price” in the future. If the stock goes up, the option holder makes money. If it goes down, they lose nothing.
Calculating the value of stock options is complex and involves financial models. But boards awarding CEO pay did not fully understand these models.
The models assumed stock price volatility would remain the same over the long term. But volatility generally declines as stocks mature. This led the models to overestimate option values.
Boards awarded massive stock options to CEOs based on inflated values from the models, helping drive the pay explosion in the 1990s. The amounts far exceeded actual CEO impact on stock performance.
By not properly understanding the mathematics and assumptions behind the option valuation models, boards contributed to unwarranted increases in CEO compensation levels.
Options pricing models have become more sophisticated over time, culminating in the Black-Scholes-Merton formula developed in 1973. Scholes and Merton won the Nobel Prize for this work.
The formula considers factors like volatility and interest rates to estimate an option’s value. While complex, the key point is the value scales with the underlying stock price.
Research showed company boards were rigid in granting the same number of stock options to CEOs, ignoring changes in stock value and option worth.
New regulations in 2006 required reporting option value using Black-Scholes-Merton, forcing boards to consider true option value. This led boards to adjust grants based on value rather than number, slowing CEO pay growth.
Politicians try to bend numbers through rounding. The Green party nearly lost a seat due to rounding vote percentages.
Trump’s DHHs proposed rounding insurance age ratio limits from 3x to 3.49x, relying on 3.49 rounding down.
Rounding definitions matter - rounding ranges vary based on significant figures versus whole numbers.

So in summary, it discusses the history and impact of options pricing models, research on CEO stock options, and examples of how politicians try to exploit rounding definitions to their benefit.

In the 1980s, the Vancouver Stock Exchange launched an index to track stock prices but made a calculation error. It discarded the last decimal place rather than rounding, causing the index value to gradually decrease over time even when the market was stable.
By 1983, the uncorrected index value was almost half its starting value due to this error accumulating over thousands of daily calculations. Consultants recalculated the proper value overnight, leading to a sudden increase.
Known as “salami slicing”, small rounding errors can accumulate over many transactions if they always round down. This was exploited in the 1999 film Office Space and by some real embezzlers.
Social security tax calculations and currency exchanges between countries with different smallest units (like euros and Romanian lei) can also be impacted by rounding errors accumulating over large volumes of transactions if not handled properly.
A Romanian hacker tested exploiting small rounding differences in automated currency exchanges but did not actually commit fraud, as he was hired by the bank to test its security. Proper rounding or adjustments are needed to prevent such salami slicing attacks.

Here are the key points about round numbers and significance of figures in the passage:

Humans are suspicious of very round numbers, as they suggest the data may have been rounded rather than being precisely measured.
A study claimed coal power causes 51,999 deaths avoided by switching to solar power, but this level of precision is unwarranted given the different scales (thousands vs ones) of the original data.
The original coal death estimate was 52,200 with a wide confidence interval, but this was rounded to 52,000 for reporting. Adding this to the single solar death estimate does not justify five significant figures of precision.
Precise-seeming numbers are used for political reasons to emphasize messages, but the reduced precision of rounded numbers feels less accurate even if technically correct.
The first recorded height of Mount Everest at 29,002 feet seems precisely measured, but much early geological data involved estimates and calculations rather than completely accurate readings.

So in summary, the passage questions the significance of very precise figures, as they may hide rounding of incompatible data scales or be used for rhetorical purposes rather than true precision of measurement. Round numbers feel less accurate even if technically correct representations of the underlying data and estimates.

In 1856, Andrew Waugh announced that Peak XV in the Himalayas was the tallest mountain on Earth and named it Mount Everest after his predecessor George Everest.
There is a rumor that Waugh’s original calculations showed the height as exactly 29,000 feet, but he added two extra fictional feet to make it seem more precise as 29,002 feet. However, there is no evidence this rounding actually occurred.
Adding or subtracting small amounts to make a number seem more precise than the raw calculations is a common practice. Numbers are often subtly changed away from round values.
A 2017 BBC report stated unemployment in the UK fell by 7,000 people to 1.6 million, but this change was below the precision of the original 1.6 million figure and thus not statistically meaningful. The report was updated to provide more context around the figures.
Studies have found heart attacks increase 24% on the Monday after clocks move forward but decrease 21% after they move back. However, looking at the full week, there is no change in overall heart attacks from daylight saving time adjustments. The redistribution of attacks within the week is what matters, not the lump stats for a single day.
The story details an incident where the windshield of a British Airways jet failed mid-flight due to improperly installed bolts.
Shift maintenance manager Sam was tasked with replacing bolts on the windshield, but struggled to identify the correct bolt types (A211-7D vs A211-8C) in dark, unlabeled storerooms.
Likely grabbing the wrong bolts (A211-8Cs instead of A211-7Ds), Sam installed them using an uncalibrated torque screwdriver.
Thirteen minutes after takeoff, the windshield exploded outward due to the mismatch between the bolts and windshield threading.
The pilot was partially sucked out but survived due to quick crew response, and the plane was landed safely.
The story examines how individual human error like Sam choosing the wrong bolts can lead to disasters when coupled with systemic deficiencies, as per James Reason’s “Swiss cheese model” of accident causation involving multiple layered defenses.
Sam was attempting to replace windshield bolts on an aircraft but could not find the correct bolts in the disorganized storehouse. Due to poor lighting, he grabbed the wrong bolts.
The bolts Sam used were a smaller size than what was required. However, he did not notice this due to having to use a two-handed method that obscured his view.
Normally Sam’s work would have been checked, but as the maintenance manager no one checked his work.
The aircraft’s windshield was mounted on the outside rather than inside, so the bolts had to resist internal cabin pressure. The incorrect, smaller bolts could be ripped out by this pressure.
An investigation found many issues with the storehouse organization and labeling. It was also discovered that the original windshield bolts on this aircraft model had always been the incorrect size but it had not caused issues until this incident.
Multiple minor errors lined up to create a potential disaster situation on the aircraft if not for the copilot regaining control at a higher altitude. The incident highlights how subtle mistakes can accumulate.

Here is a summary of the key points in section h:

The Swedish warship Vasa sank within minutes of its launch in 1628 due to being top-heavy with cannons and having an asymmetric hull. It sank in waters that preserved the wreck and it was raised in 1961, remarkably well-preserved.
Analysis showed the Vasa’s port and starboard sides were mismatched, possibly due to builders using subtly different units of measurement (Amsterdam vs Swedish feet/inches) when following instructions.
Temperature scales like Fahrenheit and Celsius chose different reference points for zero - Fahrenheit used a mixture that stabilized at a temperature, while Celsius originally had boiling as zero and freezing as 100.
Converting between absolute temperatures and temperature changes can be tricky, as a 2 degree Celsius increase does not equate to a 3.6 degree Fahrenheit increase, leading to errors like the BBC reporting.
Defining sea level can also cause issues like engineers building a bridge between Germany and Switzerland failing to account for their different sea level definitions, resulting in a mismatched bridge.
An Air Canada Boeing 767 flying from Montreal to Edmonton took off with far too little fuel due to multiple errors in fuel calculations and conversions between pounds and kilograms.
The aircraft was supposed to have 22,300 kg of fuel but took off with about half that amount due to mistakes made in Montreal and Ottawa.
During the flight, both engines failed when the fuel ran out, leaving the pilots with very limited instrument displays.
In a remarkably skilled maneuver, the pilots glided the plane over 40 miles and landed safely at an abandoned military airfield in Gimli, Manitoba using only the basic analog instruments.
The investigation found the fuel calculation errors occurred because ground crews and pilots were transitioning between imperial and metric units for fuel amounts and used the wrong conversion factor.
Multiple failures allowed the plane to take off insufficiently fueled, including communication issues, unexpected equipment tests, and misunderstandings around manual fuel checks.
The accident highlighted the dangers of unit conversion errors and the importance of redundancy in aviation safety systems. It became known as the “Gimli Glider” crash that was successfully landed without loss of life.
Having a piece of a crashed plane is thought by some to bring good luck, but most planes don’t crash so it’s debatable if it’s actually good or bad luck. The author attached a piece of a crashed plane to their laptop but it doesn’t seem to have affected the laptop’s functioning.
In one aviation incident, cargo was weighed in kilograms but fueled in pounds, so the plane took off over 30,000 pounds overweight. This highlights the risks of mixing measurement units between countries.
People often mix up dollars and cents due to their similar notation (e.g. thinking $0.002 is the same as 0.002¢). This led to confusion over cell phone roaming rates between the US and Canada for one customer.
Large numbers like millions can cause mistakes if not properly treated as units during calculations. For example, claiming $360 million Obamacare startup funds amounted to $1 million per person by incorrectly treating “millions” as a cancellable unit rather than dividing.
Measurement units have caused medical errors, like prescribing 500mg of a drug instead of the intended 32.4mg, due to confusion between similar shorthand for grains (gr) and grams (g).
No individual perfectly matches an “average” statistical profile of a country, highlighting that averages simplify diverse, complex realities.
The Australian Bureau of Statistics attempted to describe the “average Australian” based on census data, but found that no single person matched all the criteria. This shows the limitations of relying on the concept of an “average”.
In the 1950s, the US Air Force conducted a large-scale anthropometric study measuring over 4,000 personnel on 132 body dimensions. This data was processed using early computing technologies like punch card machines.
The study found that there was no single “average” body type that matched the middle range for all measurements. No individual in the sample could wear a uniform designed for the “average” dimensions.
This challenged the notion that an “average man” truly exists and showed the great variability between individuals, even within a large sample group. Designing for an “average” risks excluding many actual users. Overall, the data highlighted that almost no one is perfectly average across all measures.

The passage discusses how data collection can be biased in various ways, which then influences the conclusions that can be drawn from the data. It gives several examples:

Survivor bias - Only focusing on things that survived (like an old bridge), without considering all the failed attempts. This can give a misleading impression that the survivors were well-designed.
Sampling bias - Collecting data only from certain subgroups (like a pothole reporting app mainly used by wealthy younger people), rather than a representative sample.
Publication bias - Negative or inconclusive trial results being less likely to be published than positive results. This skews the overall published data.

It emphasizes that how and where data is collected can introduce biases, so the method of data collection is as important as the analysis. Proper analysis also requires considering these potential biases in the data.

Here are the key points summarized from the passage:

A 1980 drug trial of the heart drug lorcainide found deaths among patients taking the drug, but researchers struggled to publish their findings. If published sooner, it could have identified risks sooner and saved 10,000 lives.
The antidepressant drug reboxetine was prescribed based on one positive trial, but later six other trials showing it was no better than placebo emerged. These hidden trials should have been published.
Claims were made that ancient megalithic sites in the UK formed precise isosceles triangles linking them, but when the same analysis was done on closed Woolworths store locations, similar precise patterns emerged, showing it’s possible to find patterns through selective analysis of large datasets.
A correlation was found between cell phone towers and births in UK areas, but this was explained as being due to population size, not causation between towers and births. Correlation does not imply causation.
Statistics on their own are not answers - context, common sense and insight are needed to determine actual relationships and rule out spurious correlations.
True randomness is difficult for computers to achieve without external physical randomness sources, as computers rely on deterministic processes.
Early game shows like Press Your Luck used pseudo-random number generators that were not truly unpredictable and could be cracked by observant players like Michael Larson.
True random number generators use physical phenomena like thermal noise, quantum effects, or rolling dice to introduce unpredictability that computers alone cannot replicate.
Examples of physical random number generators discussed include the Dice-O-Matic machine, the original ERNIE computer that generated random numbers from neon lights for UK premium bonds, and modern quantum random number generators.
Even seemingly empty space exhibits quantum randomness that can be harnessed, as demonstrated by a generator that listens to quantum foam fluctuations in a vacuum.
Pseudorandom numbers from software algorithms only appear random but can be predicted, making physical sources necessary for applications requiring true unpredictability like cryptography, gaming, or simulations.
Pseudorandom number generators are algorithms that generate numbers that appear random but are actually determined by a sequence. They are useful for applications like randomizing decisions, adding digits to passwords, or generating fake phone numbers.
Early generators had flaws where the numbers would align in mathematical spaces or fall into distinct patterns if plotted. One bad IBM generator produced numbers that all fell into just 15 neat planes.
Modern browsers get random seeds from system properties like time, process IDs, memory usage, etc. But Netscape’s original method was flawed as these values were easy to predict.
Algorithms are constantly evolving to balance randomness, efficiency, and security. Microsoft never revealed how Excel generates numbers.
In 2016, Chrome’s generator repeated itself, failing to produce truly unique random user IDs for a tracking extension. It has since switched to a new algorithm.
Properly seeding even sophisticated generators is important for security. Predictable seeds undermine the apparent randomness.

So in summary, pseudorandom number generation has improved but still faces challenges around producing numbers that can’t be predicted or don’t repeat too quickly. Secure, unpredictable seeding is also crucial.

The world of pseudorandom numbers seems calm now as browsers are serving them up without issues. However, algorithms used to generate pseudorandom numbers will not remain sufficiently random indefinitely as computing power increases. Eventually, today’s algorithms will be able to be disentangled by more powerful computers. A new generation of computer scientists will need to develop improved algorithms to maintain adequate randomness for the purposes needed. Pseudorandom numbers aim to exhibit the same statistical properties as true random numbers, but complete indistinguishability may not be possible even when done carefully. The simplest conception of randomness is a uniform, independent distribution, but applications often require tailoring numbers to specific distributions. Maintaining uniformity and independence are key to avoiding patterns that could reveal a sequence as pseudorandom rather than truly random.

The Ariane 5 rocket was set to launch 4 satellites for a mission called Cluster in 1996. It exploded 40 seconds after liftoff.
The cause was determined to be that the onboard computer tried to copy a 64-bit number into a 16-bit space in memory, which caused a fatal error.
This number was related to the inertial reference system (SRI) which tracked the rocket’s position and orientation using sensors.
The SRI code was well-programmed but not adapted properly when reused on the Ariane 5, leading to the number overflow error.
The Cluster mission satellites were eventually successfully launched in 2000 after being rebuilt.
The investigation found the original Ariane SRI code to be well done, but the mistake was in reusing it without properly accounting for the different memory constraints of the Ariane 5 rocket compared to previous rockets.

So in summary, it discusses the Ariane 5 explosion caused by a software error when reusing code not adapted for the new rocket’s memory, and the eventual success of the Cluster mission after being rebuilt and launched on a different rocket.

The program describes different levels of programming mistakes, from simple typos (level zero) to more complex logic errors.
Level zero mistakes include missing symbols like semicolons that can break a program, or accidentally deleting the entire file system due to a typo in a delete command. Even a single character mistake can have disastrous consequences.
Translation errors are also level zero, where the programmer’s intended logic is not properly conveyed in the target programming language due to mistakes like using ”=” instead of ”==” for comparison.
A biologists’ research had to be retracted because their custom analysis program accidentally flipped positive and negative values due to a bug.
Typos in code verification caused the Therac-25 radiation machine to accidentally overdose patients by not catching an error carried over from earlier software.
Programmers are encouraged to write helpful comments to explain their code, as future readers including their future selves may not understand the original logic and assumptions.
Trey Harris was a system administrator at UNC in the late 1990s when different university departments ran their own email servers.
The statistics department’s email server was having issues sending emails more than around 500 miles away. Local and shorter distance emails were fine, but anything over 500 miles failed.
Trey tested emails and found the issue was related to distance to the receiving server, not its location.
The statistics department server had recently been upgraded, bringing an older version of Sendmail (5). Previously it had used the newer Sendmail 8.
The sendmail configuration file Trey had written assumed Sendmail 8. With Sendmail 5 instead, it caused issues.
Specifically, the timeout for waiting for delivery confirmation was set to default of zero seconds. So the server would immediately reject any emails without waiting, causing failures over 500 miles away.
The issue was due to incompatible configuration for the older vs newer Sendmail versions after the server upgrade changed the underlying software.

The passage discusses issues with error messages and how they can impact human-computer interactions, particularly in medical settings. It gives examples of cryptic error codes that are difficult for non-technical users to understand. This was a problem with the Therac-25 radiation machine, which produced unhelpful error messages that operators learned to dismiss. This contributed to overdose incidents where lives were lost.

The CHI+MED project aims to improve human-computer interaction for medical devices by designing systems that avoid errors. They recognize that good people do make mistakes and seek to establish a culture where errors can be admitted and addressed, rather than punishing individuals.

The passage also describes a tragic case where a cancer patient died from an accidental chemotherapy overdose caused by mistakes in programming an infusion pump. It analyzes the complex factors that contributed to the errors, from an ambiguous drug order to limitations of the pump interface and checks.

Overall, the passage advocates for designing medical technology, processes, and environments with safeguards and user understanding in mind, rather than assuming flawless human performance or blaming individuals for systemic vulnerabilities. The goal is preventing lethal mistakes through improved human-computer interaction and error management.

Here are some key points about common medical calculation errors:

Misplaced decimal points are common, such as typing 2.3.14 instead of 23.14. Calculators interpret this inconsistently.
Good medical calculators should flag ambiguous numbers to avoid factor-of-ten errors, which could have serious consequences.
Programming is still developing and complex codes can have unintended reactions. But well-programmed devices can improve safety systems.
Engineers are often legally restricted from publicly disclosing mistakes or lessons learned, even for issues with no safety impact. This prevents important information from being shared.
While mistakes are inevitable, systems for life-critical situations like medicine must get calculations consistently right. Redundancy and error-checking are important.
The “hot cheese” model of accidents recognizes that new layers or steps in a system can themselves introduce errors, not just fail to catch existing ones. Reducing accidents is complicated.
Barcode medication systems reduced some errors but opened up new potential issues if not used correctly, like not scanning patient wristbands. New technologies require consideration of unintended consequences.

In summary, medical calculations demand precision to avoid dangerous mistakes, but failures will still occur. Systems need robust error-checking and an understanding that changes can have unintended downstream effects requiring oversight. Learning from mistakes could be improved if more information was shared, within legal and ethical reason.

Nurses at hospitals would sometimes scan the same medication container twice instead of scanning two different containers if they looked identical, to save time. This led to less thorough checks of patients and medications.
When new automated systems are implemented, humans will find new ways to make mistakes if they are not careful. Relying too much on technology can lead to complacency.
In 1907, the Quebec Bridge in Canada collapsed during construction, killing 75 workers. The collapse was caused by a miscalculation of how heavy the bridge would be. Warnings from workers about deformation of support beams were ignored by engineers.
Construction is dangerous work and small mistakes can cost lives. The bridge disaster led to increased safety standards and procedures for engineers in Canada.
Even when mistakes happen, mathematics and engineering are still necessary for modern infrastructure. We need to learn from failures while still recognizing the benefits of the math and systems that typically work well. Care, checks and balances are important to reduce risks.

Here are summaries of the phrases in bold:

Daylight saving - Shifts clocks forward or backward by one hour in the spring and fall.
Deliberately vague - purposefully unclear or ambiguous about something, providing a range rather than a specific number.
Dice-O-Matic - geographic coordinates pointing to somewhere in England.
Diehard package - geographic coordinates pointing to somewhere in California.
Dow Jones - geographic coordinates pointing to New York City.
Drug trial - geographic coordinates pointing to somewhere in England or Wales.
Electron beam - geographic coordinates pointing to somewhere in England.
Expensive book - geographic coordinates pointing to somewhere in England.
Explain why - multiple geographic coordinates pointing to locations in England and the United States.
False positives - multiple geographic coordinates pointing to locations in England and the United States.
Fat-fingers error - geographic coordinates pointing to locations in England.
Feedback loop - multiple geographic coordinates pointing to locations in England.
Fence post problem - geographic coordinates pointing to locations in England.
Fenchurch Street - geographic coordinates pointing to London.
Feynman, Richard - geographic coordinates pointing to locations in England related to the physicist Richard Feynman.
Fiber-optic cable - geographic coordinates pointing to locations in England related to fiber optic technology.
Flash crash - geographic coordinates pointing to locations in England related to a stock market crash.
Foot doughnut - geographic coordinates pointing to somewhere in England.
Frigorific mixture - geographic coordinates pointing to somewhere in England.
Fuel gauge - geographic coordinates pointing to somewhere in England.
Full-body workout - geographic coordinates pointing to locations in England.
Functional sausage - geographic coordinates pointing to somewhere in England.
Gene names - geographic coordinates pointing to somewhere in England.
Gates, Bill - geographic coordinates pointing to locations related to Bill Gates.
Gimli Glider - geographic coordinates pointing to Canada related to an famous glider landing.
Good luck - multiple geographic coordinates pointing to locations in England and the United States coupled with a phrase about luck.
Gregorian calendar - geographic coordinates pointing to locations in England related to calendar reforms.
Grime dice - geographic coordinates pointing to locations in England related to a type of dice.
Harrier jet - geographic coordinates pointing to locations in England related to military aircraft.
Heart attacks - geographic coordinates pointing to locations in England related to heart health.
High-frequency trading - geographic coordinates pointing to locations in England related to stock trading.
Hot cheese - geographic coordinates pointing to somewhere in England coupled with a food item.
Human brain - geographic coordinates pointing to locations in England related to neuroscience.
Hyatt Regency - geographic coordinates pointing to Kansas City related to a hotel structural collapse.
International Date Line - geographic coordinates pointing to somewhere in the Pacific Ocean related to timezone standard.
JPMorgan Chase - geographic coordinates pointing to New York City related to a bank.
Julian calendar - geographic coordinates pointing to locations in England related to calendar reforms.
Julius Caesar - geographic coordinates pointing to locations in England related to the Roman leader.
Kansas City - geographic coordinates pointing to Missouri related to a hotel structural collapse.
Lava lamps - geographic coordinates pointing to somewhere in England related to novelty objects.
Leap years - geographic coordinates pointing to locations in England related to calendar mechanics.
Lego bricks - geographic coordinates pointing to locations in England related to the toy construction bricks.
Long enough - multiple geographic coordinates coupled with a phrase about duration or length.
Los Angeles - geographic coordinates pointing to California cities related to location or transportation.
Magic square - geographic coordinates pointing to somewhere in England related to mathematical puzzles.
Mars Climate Orbiter - geographic coordinates pointing to somewhere in England related to a NASA orbiter mission failure.
Math error - multiple geographic coordinates coupled with a phrase about mathematical mistakes.
Math mistakes - multiple geographic coordinates coupled with a phrase about mathematical errors.
McChoice Menu - geographic coordinates pointing to locations in England related to a McDonald’s product launch.
Millennium Bridge - multiple geographic coordinates in England related to a pedestrian bridge in London.
Mobilephone masts - geographic coordinates pointing to somewhere in England related to cellular infrastructure.
Most important - multiple geographic coordinates coupled with a phrase about significance.
Mr. Average - geographic coordinates pointing to locations in England related to statistics.
NBA players - geographic coordinates pointing to locations in England related to basketball.
Nontransitive dice - geographic coordinates pointing to locations in England related to game theory dice.
Non-zero - multiple geographic coordinates coupled with a phrase about quantities not being zero.
Null Island - geographic coordinates pointing to somewhere in the Pacific related to a nonexistent location.
Null meal - geographic coordinates pointing to locations in England related to missing or empty options.
Null, Steve - geographic coordinates pointing to locations in England related to a fictitious person.
Oddly specific - multiple numeric values that seem arbitrarily precise.
Off-by-one error - geographic coordinates pointing to locations in England related to a common programming bug.
Olympic Games - geographic coordinates pointing to locations in England coupled with an international sporting event.
Parker Square - geographic coordinates pointing to somewhere in England related to a mathematical shape.
Penney ante - geographic coordinates pointing to locations in England related to a coin-toss betting game.
Pepsi Points - geographic coordinates pointing to locations in England related to a failed rewards program.
Plug seal - geographic coordinates pointing to somewhere in England related to plumbing.
Pseudorandom number generator - geographic coordinates pointing to locations in England related to algorithms and randomness.
Punch card - geographic coordinates pointing to locations in England related to obsolete data storage.
Real world - multiple geographic coordinates coupled with a phrase about practical application versus theory.
Resonant frequency - multiple geographic coordinates in England related to vibrations and oscillations.
Rm -rf - geographic coordinates pointing to somewhere in England related to a Linux command.
Rock-paper-scissors - geographic coordinates pointing to locations in England related to a hand game.
Rollover errors - multiple geographic coordinates coupled with a phrase about carry mistakes.
Royal Mint - geographic coordinates pointing to locations in England related to coin production.
Salami slicing - multiple geographic coordinates coupled with a phrase about incremental processes.
Salt mines - geographic coordinates pointing to locations in England related to mineral extraction.
Scientific notation - ranges of numbers written in scientific notation.
Scud missile - geographic coordinates pointing to locations in England related to weaponry.
Sea level - geographic coordinates pointing to locations in England related to elevation.
Seemingly arbitrary - multiple values or coordinates described as randomly selected.
Sesame Street - geographic coordinates pointing to locations in England related to a children’s television show.
Should open - geographic coordinates pointing to locations in England related to facilities operation times.
Skin-tight garments - geographic coordinates pointing to somewhere in England related to clothing.
Something else - multiple geographic coordinates coupled with a phrase about alternative options.
Space Invaders - geographic coordinates pointing to somewhere in England related to a classic video game.
Space shuttle - multiple geographic coordinates in England related to NASA’s space transportation system.
SQL injection - geographic coordinates pointing to locations in England related to a web security vulnerability.
Standard deviation - multiple geographic coordinates in England related to a statistical concept.
Stock Exchange - multiple geographic coordinates in England related to securities trading.
Stock options - geographic coordinates pointing to locations in England related to financial securities.
Street signs - multiple geographic coordinates in England related to traffic signs.
Survivor bias - multiple geographic coordinates in England coupled with a cognitive bias.
Swiss cheese model - multiple geographic coordinates in England related to accident analysis.
Synchronous lateral excitation - multiple geographic coordinates in England related to vibratory motion.
T-shirt - geographic coordinates pointing to somewhere in England related to an article of clothing.
Tabulating machines - geographic coordinates pointing to locations in England related to pre-computer data processing.
Tacoma Narrows Bridge - geographic coordinates pointing to Washington state related to a bridge collapse.
Tallest mountain - geographic coordinates pointing to locations in England related to geography.
Tax returns - geographic coordinates pointing to locations in England related to finances.
Test, Brian - geographic coordinates pointing to locations in England related to a fictitious person.
Therac machine - geographic coordinates pointing to locations in England related to radiation therapy.
Three cogs - geographic coordinates pointing to locations in England related to mechanical components.
Tokyo Stock Exchange - geographic coordinates pointing to Japan related to securities trading.
Torsional instability - multiple geographic coordinates in England related to structural vibration.
Trading algorithms - multiple geographic coordinates in England related to computerized stock trading.
Traffic control - geographic coordinates pointing to locations in California related to transportation management.
Trump administration - geographic coordinates range pointing to the United States related to government.
UK government - multiple geographic coordinates in England coupled with a reference to national leadership.
UK lottery - multiple geographic coordinates in England related to a gambling game.
UK street signs - geographic coordinates range pointing to England related to traffic signs.
US Army - geographic coordinates range pointing to the United States related to the military.
USS Yorktown - geographic coordinates pointing to South Carolina related to an aircraft carrier.
Vancouver Stock Exchange - geographic coordinates pointing to Canada related to securities trading.
Waka waka - geographic coordinates pointing to somewhere in England related to onomatopoeia.
Went wrong - multiple geographic coordinates coupled with a phrase about errors or failures.
Wobbly Bridge - geographic coordinates pointing to England related to structural instability.
Woolworths locations - geographic coordinates pointing to locations in England related to a retail chain.
World record - multiple geographic coordinates coupled with a phrase about achievements.
Wrong bolts - multiple geographic coordinates in England related to incorrect fasteners.
X-rays - geographic coordinates pointing to locations in England related to radiation technology.

Here is a summary of the key details provided:

Two dates are mentioned: July 6, 1912, when the record was 10.6 seconds, and June 20, 1936, when African American athlete Jesse Owens set the record of 10.2 seconds.
This record time of 10.2 seconds by Jesse Owens was set two months before the 1936 Berlin Olympics, where Hitler intended to showcase Aryan supremacy, but Owens went on to win four gold medals.
The question refers to predicting a time from a line-of-best-fit drawn between all record times between those two dates of July 6, 1912 and June 20, 1936.

#book-summary

Humble Pi When Math Goes Wrong in the Real World - Matt Parker