Pkarr does not require the DHT to change topology, neither do bittorrent peers need to change the topology of anything.

I think you are thinking about Pkarr as if it is a holepunching/NAT traversal thing. It is NOT, but neither is Mainline.

Mainline is literally a sharded routing table, just like DNS, but instead of a hierarchy, a flat distribution of routing servers (all of which have port forwarding no need for NAT) .

so now the question is, could one million small nodes cope with the 8 billion people updating their homeserver/hosting provider once a year each? because that is the only Publishing we need.

As for Reads, well just like DNS, we have to cache the shit of it, but that is ok, we only need the DHT to work well enough and caches be refreshed frequently enough just so your followers to detect your super rare event of change your hosting SERVER

Seriously, if the DHT can't scale for this, then nothing can and the alternative is just ICANN, if you think a 100 Nostr relays can scale better than a million nodes AND stay open and censorship resistant, I need you to explain that to me and walk me through what is the innovation in relays that we didn't have before ... because DHTs literally exists because Bittorrent trackers (exactly same as nostr relays) couldn't both scale AND stay permissionless.

I hope that makes sense.