Two Ways to Count the Dead

We all remember the daily number. It came out every afternoon. Everyone read it, everyone argued about it, and then we all went to bed.

I went back to look at those numbers properly. The first thing I found is that the number we all watched was the wrong one.

Two Ways To Count The Dead

The official count is 66,864 people. That means someone was tested, the test said yes, they died, and a doctor wrote COVID on the form. All four things had to happen.

There is another way to count. You look at how many people usually die here in a normal year. Then you count how many actually died. The difference is called excess deaths. It does not care what the form says. It just asks: are more people dying than usual?

That number is 290,774.

So 223,910 people died on top of the usual, and were never counted as COVID.

Some of them had COVID and nobody tested them. Some died of other things because the hospitals were full and there was no bed. I cannot tell you how many are in each group. Nobody can, from this data. But they are all people who would probably be alive without the pandemic.

Why The Death Rate Number Is Slippery

People love to quote the death rate. Ours looks like 1.6 out of every 100 cases.

But look at what that sum does. On top: people who died. On the bottom: people who tested positive. Not people who got sick. People who got a test.

Here is how I know the bottom number is too small. When testing is going well, only about 5 out of every 100 tests come back positive. That is the level health experts aim for. Ours hit 45.6.

When 45 out of 100 tests are positive, you are only testing people who are already very sick. Everyone milder is out there uncounted. So the bottom of the sum is far too small, and the real chance of dying was lower than 1.6.

A number can be correct and still not mean what people think it means. That one is my favourite example.

How Many Waves?

People say four waves. I wanted to check, so I had to decide what a wave even is.

I picked a rule and wrote it down. A wave is a bump in the smoothed daily cases that goes over 1,000 a day, and that is separated from any bigger bump by a proper dip in between.

That rule gives five. Pick a slightly different rule and you get four, or six. That is not me being unsure. It is that "how many waves" has no answer until somebody says what a wave is.

The two big ones are worth putting side by side.

Wave 2 was Delta, in 2021. It had 2,354,103 cases and killed 41,694 people.

Wave 3 was Omicron, in early 2022. It had a much sharper peak. The worst single day of the whole pandemic was in that wave, 38,867 cases. But it killed 9,402.

Fewer cases, four times the deaths. That gap is the clearest thing in this whole dataset about what the vaccines did.

A Backwards-Looking Number

There is a score for how strict a country's rules are. Zero is nothing, 100 is everything shut. We hit 100.

I checked whether strict rules went with high cases. They did. The link came out positive.

Read fast, that sounds like lockdowns made things worse. They did not. It is the other way round: the government locked down because cases were climbing. The rules follow the cases, not the reverse.

This is the trap in every chart of two lines. Things moving together tells you nothing about which one pushed.

Next To Our Neighbours

Against the five other big countries in Southeast Asia, we are 2nd worst for deaths per million people.

But we are near the bottom for cases per million. Singapore recorded about 14 times more cases per person than we did.

Singapore did not have 14 times more illness. Singapore counted. We did not.

Lots of deaths plus few recorded cases is not a good sign. It is what missing the cases looks like.

What I Took Off The Page

The old version of this page had charts for cases by region, vaccine brands, ICU beds, and cases by age.

I could not find any of that in a file I trust. The health department did publish it once, but that download is gone now.

So I took those charts off. A chart with made-up numbers is worse than no chart, because it looks like an answer.

One Small Thing That Bothered Me

All the numbers here are read on the same day. That sounds boring. It is not.

The old page had a case count from May 2023, a death count from January 2024, and a vaccine count from December 2022, all sitting in a row like one snapshot. Each was roughly right on its own. Together they described a country that never existed.

So now the date rides along on every single row of every file. It is the least clever thing I did here and probably the most useful.