So how many IPv4 addresses will disappear?

I was hooked to the question number 13 "Do you know how many total addresses [IPv4] disappears " from the topic Tricky network issues
Not so much the question itself, but the remark of the respected eucariot "Here we are wasteful." It became interesting, but how many percent of the address space disappears in the current conditions of the shortage of IPv4 addresses (about which almost all mass media sounded about a year ago and safely forgotten) - 1%? Or 5%? Or maybe 10%?

Firstly, it is not entirely clear what is meant by the term disappears ? Secondly, which sources to focus on when determining the lists of missing addresses, where to look for the truth?

For simplicity, we will assume that all addresses that are explicitly mentioned (for example, in the RFC) are lost , that for one reason or another they cannot be used for assignment to a node. Losses on masks / gates caused by the need to ensure the allocation of IP addresses to consumers and routing between them will not be taken into account. Losses will be considered in% of the maximum possible number of unique addresses (4,294,967,296 = 2 ^ 32).
')
According to RFC 5735 (which is mentioned by eucariot in the answer to question no. 13) it turned out that 588,514,561 (13.70%) addresses would disappear. The article on the Reserved_IP_addresses wiki has already indicated 592,708,616 (13.80%) addresses. Why the numbers do not match? And because RFC 5735 is deprecated, and the current RFC on the topic is 6890 . Based on RFC6890, there is an IANA IPv4 Special-Purpose Address Registry document on iana.org that is fairly complete and up to date (Last Updated on 2013-05-22), although not perfect (imho).

In the new RFC6890, the network 100.64.0.0/10 (Shared Address Space) appears, and the subnet 192.0.0.0/24 (IETF Protocol Assignments) is detailed in 192.0.0.170/32 and 192.0.0.171/32 (NAT64 / DNS64 Discovery), and also 192.0.0.0/29 (DS-Lite). But the multicast network 224.0.0.0/4 from RFC5735 is not already on the lists. Total, 324 273 411 (7.55%) of missing addresses. And what happened to the range of 224.0.0.0/4? Did they distribute to the suffering? No, the IANA IPv4 Address Space Registry in the Designation column honestly stated Multicast. Therefore, we take into account the range of 224.0.0.0/4 in the list of losses .
As for the wiki, the network 192.0.0.0/24 is mentioned only by the subnet 192.0.0.0/29, which gives an absolute error of 248 hosts, which is not reflected as a percentage due to the meager size.

Thus, if I am not mistaken anywhere, today the total loss of IPv4 addresses is 13.8% of the maximum possible number of unique addresses, and there is an error in Wikipedia. Dear experts, please correct me if I am wrong. Or Wikipedia.

Along the way, a question arose for specialists - how is the shortage of IPv4 addresses, about which the mass media widely announced last year? And how long will the RESERVED for Future use status (from September 1981 already) be in the range of 240.0.0.0/4?

PS: I'm not a networker, if wrong - do not kick much pliz.