Can be lots of things, there's lots of evidence that data quality is extremely poor right now, but not enough to be able to say accurately what all the data sho

Can be lots of things, there's lots of evidence that data quality is extremely poor right now, but not enough to be able to say accurately what all the data should be. For example, in the Netherlands, in a blood test about 3% of the population was found to have antibodies, while the number of people confirmed to have corona via tests is at 0.15% of the population. We're looking at a 1:20 confirmed/actual rate. That tells you testing is way off, about 20x so. The total tests performed is roughly similar to the US by the way, about 1 in 100. Then there's death rates. The Netherlands currently saw 2 in 10.000 confirmed deaths from corona. This is about double the rate in the US. However if we look at excess deaths compared to a 3-year average for the same period (2017-2019) we see about triple the amount. Apparently 2 in 3 die of corona without it being tested or confirmed. (likely in nursing homes where mortality rates are often around 30% per year, and it's not uncommon for staff to find a patient died in their sleep for example). And the Netherlands is a small country with high-quality public infrastructure. It's quite likely that it is among the countries with the best record-keeping system, and is among the countries with the highest tests per population, and even here we see massive undercounting of infections and deaths. It's good to keep looking at the numbers, but only with a huge caveat. I'm actually completely flabbergasted that the media report on infection numbers without ANY mention of a confidence interval, standard deviation or some kind of uncertainty measurement, as well as some basic stats around testing methodologies, in particular when reporting cross-country comparisons. This should be the norm, but instead we just get country comparisons, world maps and graphs displaying wildly inaccurate and limited data (which is absolutely useful, but criminal to report without caveats). Apart from a mention about 'Chinese data isn't reliable' or 'India isn't testing much, so the problem may be far greater', there's really no consistent discussion about this. There's lots of standard statistical tools to express uncertainty which have been used in science for decades, but they're completely absent in our reporting or discussions at the moment.