No, I didn’t misunderstand anything. The “real experts” are saying that the denominator (number of infected) is likely much higher than official numbers. But they are not saying that the numerator (deaths) is accurately represented by official figures. Deaths are easy to count if you only count deaths of confirmed infected people. There’s plenty of evidence that people are dying outside of the medical system, and that China is not accurately reporting all deaths from within the medical system as being caused by the virus.
In short, we have no idea, and it’s very dishonest to claim that either number is remotely reliable at this point.
Now the interesting question is which number is greater, unreported cases or unreported deaths. I'd say the number of unreported cases tends to be way higher than unreported deaths. Fits apparently epidemics in general. Also it is kind of logic. Severe cases have a bias to end up in hospitals, and to reported when they end deadly, only leaving the measuring error due to reporting issues.
Less severe cases on the other hand have a tendency to end up unreported, not in the hospital. Neither deadly. Still leaves a number of never reported cases. All in all, unreported cases are likely to be higher than unreported deaths. As a result, qualitatively speaking, mortality should be lower.
I'd say the number of unreported cases tends to be way higher than unreported deaths.
Of course, but that doesn’t mean that mortality will be lower. A much smaller number of unreported deaths can shift the rate upward. At a 2% mortality rate, you only need one unreported death for every 50 unreported infected to maintain the same rate. That might be laughably high or laughably low. We just don’t know.
But the people in the best position to know, the Chinese government, seem to be taking this pretty damn seriously.
In short, we have no idea, and it’s very dishonest to claim that either number is remotely reliable at this point.