The author brought up more examples besides PNG:

* Afghanistan

* Nigeria

* Congo

* South Sudan

* Eritrea

* Chad

* Somalia

* South Africa

Enough that "a lot" seems to be a fair characterization.

Also - while he implies this, I think it's important to mention explicitly - there's obvious fakery in the number of significant digits. If the numbers are approximations to the nearest ten million (or worse), it's a form of scientific fraud to provide a number like "94.9 million".

The only one of those that is an example is Nigeria. All the others are just listed as examples of countries that have not conducted a census in an extremely long time. While that's a good reason to think the numbers are probably inaccurate, it's not a good reason to think they are fake.

> there's obvious fakery in the number of significant digits. If the numbers are approximations to the nearest ten million (or worse), it's a form of scientific fraud to provide a number like "94.9 million"

The numbers aren't approximations to the nearest ten million. Just because they're inaccurate doesn't mean they're imprecise. For comparison if my bank statement is missing a large transaction it may be off the true value by hundreds of dollars, but that doesn't mean they didn't count the cents for the transactions they're aware of.