You can run an experiment in a single large nation, such as the US. Policies are set at the national level in any case.

Virtual child porn might well crowd out a market for real porn.

This speculation seems unfounded, considering this has not happened in adult porn.

The production of real adult porn is as legal as virtual adult porn. Since the production of real child porn would remain illegal, one might expect a difference.

It should still be possible to follow the money to the producers. One could consider making the purchase of such material illegal but not its possession.

No. Cryptography and covering your tracks by using anonymization services is trivial.

These methods are available in today's environment too where child porn possession is illegal. There are still a lot of convictions. If we divide the world into "those who can use tracks-covering services reliably" and "those who can't", we could argue that the first group is already consuming its fill of child porn and the second group would be as uncertain in covering financial dealings as they are in covering downloads today.

But to my knowledge, making a video of it which a person uploads or sells is not an additional crime, and possession is not a crime. Consider real, existing video of hostages being executed. People watching them creates the demand for their creation, but we don't even think about banning possession of such things.

I don't think legalizing one harmful thing because other harmful things are legal is a good argument.

That's fair enough, but we can also consult our intuitions about how we'd like to handle that case. Would you with enthusiasm support efforts to make possession of such videos illegal? My reaction is, "Let's not go there, and just let people possess those videos."

Cryptography and covering your tracks by using anonymization services is trivial.

It is many things, but trivial is definitely not one of them.

As a related example, consider Bruce Schneier's opinion that it is non-trivial to maintain as simple a thing as an air gap.

I'm a little amazed that you're managing to lose this argument, Hypor.