When you look at any dataset, even a single outlier value can be a red light preventing you from moving forward. Here’s five examples of why this type of number is so important.
1. Outlier inputs turn a medical device into a death ray.
In March 1986, a radiologist at the East Texas Cancer Centre prepared to administer 180 rads from an electron beam to a patient - something they’d done hundreds of times on the machine. The procedure was marked correctly on the screen, but when the treatment ran, the machine threw an error and did nothing. The operator pressed ‘P’ to proceed, and immediately the patient leapt up screaming, beating on the door to be let out. His body went into shutdown, and within days he was dead.
The machine was THERAC-25; an advanced device that combined the ability to both take X-rays as well as provide electron therapy. A year earlier, the maker had reported that they were moving to software-based safety controls for the first time, but there weren’t yet standardised tests to check for reasonable ranges of output.
Thanks to a bug causing an error that only occurred randomly, the machine could be set to deliver radiation thousands of times higher than intended. The Texas patient, along with five others globally, received lethal doses of radiation to treatment areas, frying their organs. A check on whether the dose was an outlier compared to all others delivered by the machine would have prevented the disaster, and such safeguards are now mandated.
2. An account with too many nines catches the eye of the SEC
In 1985, Kenneth Lay formed the energy giant Enron. The company was incredibly successful, with a diverse set of investments in almost the entire energy supply chain. By the end of 2000, their share price sat around $80 USD, and then it all unraveled. By November 2001, the price had dropped to $1 and Enron was bankrupt. As the market scrambled to prevent the crisis spreading, investors and regulators wanted answers.
During the investigation that followed, prosecutors noticed something unusual. When they looked at the values entered into accounting records, they saw that a huge proportion of them ended in ‘9’. Far more than would be expected based on Benford’s Law - a well-established phenomena that shows numbers ending in 1 should be far more common than numbers ending in 9. The values themselves weren’t outliers, but the number of times the digit ‘9’ appeared at the end of values was. It was hard statistical evidence that someone had been adding fake data to the books. Prosecutors were able to show that this pattern warranted a deeper investigation, which eventually ended in criminal prosecutions for Enron leadership for their crimes of covering up bad investments.
Outlier occurrences of specific digits in accounting records, in opposition to Benford’s Law, are still used in tax litigation cases today.
3. A missed pressure spike causes an environmental catastrophe
When you operate underwater drilling machinery, you’re running everything based on readings from sensors. The number of important sensors an operator is required to monitor is mind-boggling, and it’s always possible that something is missed if outlier readings aren’t flagged automatically.
It’s a lesson that the operators of the Deepwater Horizon oil drilling operation learned the hard way in 2010. The day had proceeded relatively smoothly, with routine maintenance and upgrade operations conducted during daylight hours. Operation logs show that testing was run on well stability to ensure no issues had come up during the work - everything returned normal.
But technicians working on the rig had forgotten to complete one step when switching from testing to operations - reopening a safety valve designed to protect the rig during tests. As they moved through the procedure to restart drilling as soon as possible, they missed a critical reading.
A report of the incident found that at 9.17 PM on the 20th of April 2010, the indicator for Pump 2 showed a pressure spike to an outlier value of 6000psi - enough pressure to smash a thick concrete wall and well more than enough to cause a pipe to explode, even at the incredible depths of the well.
Without an alarm warning of the outlier, nobody noticed the reading. The pipe exploded, and the worst ecological disaster in the history of the Americas ensued. Thousands of kilometers of coastline were impacted, with researchers detecting lowered fertility rates for animals in the area more than five years later.
4. A deluge of strange bets busts a global crime syndicate
As online sports betting is becoming more popular, there are more and more cases in which unusual activity catches out someone trying something tricky. One of the cases most recently in the news is that of NBA player Terry Rozier and his mysterious case of sore feet.
In 2023, Rozier was having a great game against the New Orleans Pelicans. Then, seemingly without warning, he left the court during a time out. It was an innocuous move - in a sport like basketball that takes its toll on the ankles and knees of exceedingly tall athletes, it’s common for injuries to occur. But analysts watching online bets saw something else that told a different story.
With modern digital books, it’s possible to bet on basically any aspect of the game - from which minute the first point will be scored, to the colour of a player’s shoes. This is where Rozier and his conspirator’s plans were focussed, and ultimately what brought them down.
Usually as games happen, sports betting apps see a fairly predictable rate of bets being placed. The more specific a bet, the lower the rate of users placing money on it. For example, almost everyone on the app will bet on who wins the game, but only a couple will be willing to bet on a specific player’s performance. So when a stream of bets on whether Rozier would finish the game arrived at a rate that was an outlier compared to all the other betting rates across the game, it immediately raised alarms for analysts. Their alarm was confirmed when Rozier, seemingly unhurt otherwise, left the court.
The incident began a years-long investigation into criminal betting fraud that discovered multiple current and former NBA players working directly and indirectly with New York crime families to rig NBA games and provide insider knowledge to gamblers. In many cases, statistical analysis showing outliers in the rate of bets being placed on online sports betting apps served as critical evidence in investigations.
5. Spiders Georg and the spider-filled cave
No discussion of outliers can be truly complete without noting the bizarre but often referenced world of fictional character ‘Spiders Georg’.
Spiders Georg originated on the niche counter-culture site Tumblr, first posted by user Max Lavergne (username ‘reallyreallytrying’) as a joke. The post attempted to debunk the claim that ‘the average person eats seven spiders in their lifetime’ (an often quoted, but incorrect factoid) by remarking that in reality, that average was lifted by Spiders Georg, who lived in a cave and ate 10,000 spiders every day, and was thus an outlier who shouldn't be counted.
The mix of mathematical play with absurdist imagery was exactly the type of thing Tumblr users were hungry for. The meme became ubiquitous in the last half of the 2010s, with the example being recounted in endless formats and revisions. The meme’s ‘jumping the shark’ moment occurred in 2023 when Business Insider referred to the US President Trump as the ‘Spiders Georg of world leaders’ for his high number of felony indictments.
Even today, invoking the name will bring a smirk to the faces of those who were there.
Now you’ve seen how important outliers are to discovering the secrets of your data, it’s time to see whether your own data contains this issue. You can search your own data for outliers using our laboratory.