Yesterday we found a pair: the same Ford defect, the same manufacturer campaign code, described in America as a “loss of steering control, increasing the risk of a crash” and in the Netherlands as a car that becomes “difficult to steer”.
One example is an anecdote. So we scored all of them.
Across 69 recalls confirmed to be the same defect in both countries, the American notice reaches for stronger language 26 times, the Dutch one 7 times, and they are level 36 times.
Which warnings appear, and where
| Warning | In the US text | In the Dutch text |
|---|---|---|
| A crash or collision | 60 | 36 |
| Fire | 11 | 14 |
| Injury | 19 | 10 |
| Death | 0 | 1 |
| Loss of control | 1 | 0 |
| Happens without warning | 6 | 11 |
The widest differences
Both authorities’ own references are shown. The Dutch text is quoted exactly as the Netherlands vehicle register publishes it.
SRVRCL26001: Ducati North Europe · US 26V050 · NL MGP260038
United States (3/6): A sudden loss of rear brake function increases the risk of a crash or injury.
Netherlands (0/6): De achterremmen kunnen remdruk verliezen.
21S32: Ford Nederland B.V. · US 21V537 · NL MGP220161
United States (2/6): A rear toe-link fracture can result in a loss of steering control, increasing the risk of a crash.
Netherlands (0/6): Hierdoor kan het buitenste deel van de reactiestang van de achterwielophanging barsten, waardoor het voertuig moeilijk te besturen is.
RP1063: Volvo Cars Nederland B.V. · US 25V453 · NL MGP250233
United States (2/6): Damage to harnesses can cause various electrical malfunctions, such as a loss of power steering assist, increasing the risk of a crash or injury.
Netherlands (0/6): Verschillende electronische storingen waaronder verlies van de stuurbekrachtiging.
25SD9: Ford Nederland B.V. · US 25V860 · NL MGP260085
United States (2/6): A loss of drive power increases the risk of a crash. In addition, a half shaft that is not fully engaged to the primary drive unit can result in a vehicle rollaway if the parking brake is not applied, increasing the risk of a crash or injury.
Netherlands (0/6): Hierdoor kan lekkage en uiteindelijk verlies van aandrijfkracht optreden. Ook kan het voertuig onbedoeld bewegen in de parkeerstand.
R10365: Volvo Cars Nederland B.V. · US 26V136 · NL MGP260093
United States (2/6): An audible warning chime that does not properly alert of an unbelted seat belt can increase the risk of injury during a crash.
Netherlands (0/6): De bestuurder krijgt geen waarschuwing als de inzittenden geen veiligheidsgordel dragen.
18S07: Ford Nederland B.V. · US 18V169 · NL MGP180340
United States (1/6): A transmission fluid leak in the presence of an ignition source such as hot engine or exhaust components can increase the risk of a fire.
Netherlands (0/6): De koppeling kan slippen. Hierdoor is de rijsnelheid gereduceerd, zijn de voertuigprestaties beperkt en verspreidt de koppeling een geur. In extreme gevallen komen fragmenten van de koppeling/transmissie los en raken deze de componenten onder de motorkap.
23S27: Ford Nederland B.V. · US 23V380 · NL MGP230240
United States (2/6): An engine compartment fire increases the risk of injury.
Netherlands (1/6): Hierdoor is het mogelijk dat olie ontsnapt, waardoor in combinatie met de hitte in de motorruimte, brand onder de motorkap kan ontstaan.
23C35: Ford Nederland B.V. · US 23V727 · NL MGP230470
United States (1/6): Without a low brake fluid warning light, the vehicle may be driven with low brake fluid, reducing braking performance and increasing the risk of a crash.
Netherlands (0/6): Verliezen van remcapaciteit, niet kunnen remmen.
What this score cannot do. The score does not read meaning, negation or context - "cannot catch fire" scores the same as "may catch fire" - and Dutch and English do not map word for word. A single pair proves nothing. Only the direction of the difference across many pairs is worth anything, which is why no manufacturer is ranked by it.
Six categories of escalation: crash, fire, injury, death, loss of control, no warning. One point per category present in the consequence text, so the range is 0 to 6. Repetition earns nothing. The English and Duth word lists are published in full so the count can be checked or disputed. Recall text is published by NHTSA and the Netherlands vehicle register. The pairing and the scoring are CarsMultiverse.
What the count actually shows
The clearest single number: the American text names a crash in 59 of the 67 matched recalls. The Dutch text names one in 35. Injury appears twice as often on the American side.
But it does not run one way on everything. The Dutch notices mention fire slightly more often, and are considerably more likely to say a failure happens without warning, which is arguably the more useful thing for a driver to know. Neither authority is simply “better”. They emphasise different dangers about the identical defect.
The method, in full, so you can disagree with it
Six categories: a crash, fire, injury, death, loss of control, and whether it happens without warning. A text scores one point per category present: range 0 to 6. Saying “fire” four times earns one point, because repeating a word is not saying more.
The exact English and Dutch word lists are published in the data at /wp-json/cmvsev/v1/report, under terms. They are the whole method. If you think a term is wrong or missing, you can see it, change it, and recount, which is the point. A severity score produced by a model nobody can inspect would be worthless here.
What this score cannot do
It does not read meaning, negation or context. “Cannot catch fire” scores exactly the same as “may catch fire”. Dutch and English do not map word for word, and a translator would make different choices than a word list does.
So a single pair proves nothing. Only the direction of the difference across many pairs carries any weight, and even then it is a measurement of language, not of danger. That is also why no manufacturer is ranked by this index and none ever will be: the score is not precise enough to carry a name.
Why it is worth having anyway
Because the alternative is that nobody checks at all. Two national authorities describe the same defect to two sets of drivers, and until now there was no way to ask whether they said the same thing, not because it was hard, but because nothing linked the two records together. The join that made this possible is here.
The rest of the record
- The same recall, two countries, two stories: how the pairs are matched.
- A million recalled cars with no repair on record
- How long before anyone tells you your car is unsafe?
- Check your car for open recalls
Every figure on this page comes from a table you can have. 40 rows, CSV JSON All 14 datasets