Case No. 001 · The Reveal
One case. Two files. A 50 point verdict gap.
Every juror in Liebeck v. McDonald's, Retried was randomly handed one of two files. Neither group knew the other file existed. The files differ in how much of the record they contain, and any known defect in either file is disclosed in the record addendum below rather than quietly corrected.
The Folklore File
0% found liable
Read this file
You have been assigned the case file as the public knows it.
In February 1992, a woman bought a cup of coffee at a McDonald's drive-through in Albuquerque, New Mexico. Riding in the car with her drink, she took the lid off the cup and spilled the coffee into her lap.
She was burned, and she sued McDonald's. Her claim: the coffee was too hot.
A New Mexico jury heard the case in 1994 and awarded her nearly $2.9 million. The verdict made headlines around the world and became the most famous example in the American argument over lawsuits: to some, proof that anyone can sue over anything; to others, a case everyone talks about and no one has actually read.
The question before you is the question that was before that jury: was McDonald's liable for her injuries?
The Evidence File
50% found liable
Read this file
You have been assigned the trial record.
Stella Liebeck was 79 years old in February 1992 when she bought a 49-cent cup of coffee at a McDonald's drive-through in Albuquerque, New Mexico. She was a passenger in her grandson's car. He pulled out of the drive-through lane and parked so she could add cream and sugar. Holding the cup between her knees, she pulled the far side of the lid toward her, and the cup tipped into her lap.
Her sweatpants absorbed the coffee and held it against her skin. She suffered third-degree burns, the kind that destroy the full thickness of the skin, across six percent of her body, with lesser burns over sixteen percent. She spent eight days in the hospital, underwent skin grafting to her thighs and groin, lost twenty pounds during recovery, and required care for two years.
Evidence at trial established that McDonald's corporate policy required coffee to be held at 180 to 190 degrees Fahrenheit. Coffee served at home is typically 135 to 140 degrees. At 190 degrees, liquid can cause third-degree burns in under three seconds; at 160 degrees, a person has more than twenty seconds to react. McDonald's own quality-assurance manager testified the company knew the coffee was not fit for consumption at the temperature served, because it would scald the mouth and throat.
Documents produced in discovery showed more than 700 prior reports of burns from McDonald's coffee between 1982 and 1992, including third-degree burns to children, and settlements the company had paid in earlier claims. McDonald's had made no change to its holding temperature.
Ms. Liebeck initially did not want to sue. She wrote to McDonald's asking it to check its temperature practices and to cover her costs, roughly $20,000 in medical bills and lost income. McDonald's offered $800. A mediator later recommended settling for $225,000; McDonald's declined.
At trial, McDonald's argued that customers buy its coffee hot precisely because they want it hot, that Ms. Liebeck's own handling caused the spill, and that her age made her skin more vulnerable than the average customer's.
The question before you is the question that was before that jury: was McDonald's liable for her injuries?
What the real court did
The Record Addendumwhat each jury saw, and what none of them did
This case was a blind split. Its seventeen jurors did not all read the same thing, and the difference between what they read is the finding. This note sets out exactly who saw what, what neither group saw, and one flaw in our own instrument that we would rather disclose than bury.
The two files. Each juror was assigned one of two versions at random and could not see the other, or know that another existed.
- The folklore file ran to roughly one hundred and twenty words: a woman bought coffee at a drive-through, removed the lid, spilled it in her lap, sued, and was awarded nearly $2.9 million. Every fact in it appears in the trial record. It simply stops there, which is where the public conversation about this case has stopped for thirty years.
- The trial-record file added what the first left out: that she was seventy-nine, that the car was parked, that the coffee was held at 180 to 190 degrees Fahrenheit against a domestic norm of 135 to 140, that liquid at that temperature destroys skin in under three seconds, that McDonald's own quality-assurance manager testified the coffee was not fit to drink as served, that more than seven hundred prior burn reports had reached the company between 1982 and 1992, that she first asked only for roughly $20,000 in bills and lost income, and that the company offered $800.
What neither file contained. Neither group was told how the 1994 case was resolved beyond the headline figure. The compensatory award of $160,000, the twenty percent comparative fault assigned to Ms. Liebeck, the judge's reduction of the punitive award from $2.7 million to $480,000, and the confidential settlement that followed were all withheld until voting closed. They now appear on the case page as the real-court outcome. Any opinion on this page that reasons from those figures is reasoning from facts that arrived after the vote, and says so where it does.
A flaw in our own file, disclosed rather than corrected. The folklore file said Ms. Liebeck was "riding in the car" when she removed the lid. The trial record is that her grandson had pulled out of the drive-through lane and parked. Whether the car was moving is one of the most common public misconceptions about this case, and our own summary carried a phrase that invites it. Our published method says nothing in either file is false. That phrase was at best ambiguous and at worst reproduced the very error the experiment exists to measure.
We are not editing it now. Changing the file after the votes are in would corrupt the record more thoroughly than the flaw does, and a quiet fix is how a research instrument becomes a marketing instrument. The honest statement is this: the folklore group's result may be partly an artifact of our wording rather than purely an effect of omission, and the size of that contribution is unknown. Any future blind split gets its files read specifically for implied facts before it opens.
Sample. Seventeen jurors: nine on the folklore file, eight on the trial record. That is small, well below the threshold of one hundred and fifty at which this house is willing to publish raw counts as a citable finding. The counts appear here because a founding entry with an honest denominator is worth more than a percentage with a hidden one, not because seventeen people settle anything.
Notice to the jurors of the popular-story arm. The court would ordinarily correct each affected juror directly. It cannot: jurors vote without accounts, and the court holds no contact information by design. The correction is therefore published in every place a returning juror lands: on the case page, in this addendum, and in the press release announcing the verdict. Any juror who wishes to revisit their verdict with the full record may read it on the case page; the record of what each file said at voting time is preserved unchanged.
How the error got through, and what changed. Before the case opened, both files were checked against the record for the facts they stated. The popular-story file passed that check because it was read for what it stated, not for what it implied: the phrase about riding in the car was treated as scene-setting rather than as a factual claim about whether the vehicle was moving. The court's process has changed accordingly. Every blind-split file is now read a second time for implied facts, before it opens, by a reader whose only assignment is to ask what a juror would take each sentence to assert.