Ask a simple question about cardiac-arrest readiness in the UAE — how many public-access defibrillators are there, and where — and you will not get an answer. Not a wrong answer. No answer.
As of July 2026, OpenStreetMap, the largest open geographic dataset in the world, records five AEDs in the entire country. Nobody believes there are only five. There are certainly thousands, in malls and offices and hotels and schools. But "certainly thousands" is not a number, and a system you cannot count is a system you cannot manage, cannot audit, and cannot honestly claim to have improved.
This is the denominator problem, and it is more corrosive than it looks.
Why the missing number matters
Every meaningful statement about readiness is a fraction. What share of arrests happen within reach of a working device? What share of devices are actually functional on the day they are needed? What share of a district can be reached inside the few minutes that decide whether someone lives? Each of these needs a denominator — a known total to divide by.
Without it, claims become unverifiable. "We have distributed thousands of AEDs" is a procurement figure, not a readiness figure. It says nothing about how many still work, how many are reachable, or how many sit locked in an office that closes at six. A cumulative count of things bought is the easiest number to grow and the least honest one to cite, because it only ever goes up.
A number that can only rise is not a measurement. It is a marketing instrument.
Counting is the intervention
The instinct is to treat the register as bookkeeping — something you do after the real work of buying and installing devices. That is backwards. The register is the real work. Until devices, responders, and outcomes are counted and kept current in one place, nothing else can be trusted: not coverage maps, not response times, not survival claims.
This is why we start here. Not with more devices, and not with a louder campaign, but with the plain, unglamorous act of maintaining a denominator — and being honest about how small the verified number is while we build it.
Five, today, in open data. We would like to be able to tell you the real figure. That we cannot is the entire point.