Not sure what to do for sensor ghosts of enemy players. Actually does the ship count on the Empires window include ghosts you didn't dismiss? Then probably it's only fair to include them also in IP costs

I don't think the tallies for the empires list knows about ghosts or dismissing them.

Otherwise, #4764 might fix it?

In MP25 turn 226 (present turn as of writing), without these changes I see 4 IP for myself in the Empires window. With this change, i see -8.3. But I have 92 ships, so it should be -9.2 difference, no? Dunno where do extra 3 IP diff come from.

There are also some slight science point generation differences for some reason

Upper is with this change applied, lower is without. But you know, server doesn't have those changes so... dunno

OK so I've added the !this_client_known_destroyed_objects.contains(obj.ID()) && !this_client_stale_object_info.contains(obj.ID()); condition to the mix so those aren't counted (quick and dirty added to owned_by_empire) and it looks better - no changes in science (probably some stale/destroyed data from ships in blizzards?) and IP also more like expected I think? though it's still 8.6 rather than 9.2 but I've noticed monster ships have Influence meter of 0.0 even though Target Influence is -0.1, and I have 6 of them, so it matches that I don't get that extra -0.6 penalty (so, a new bug/issue perhaps, why those monster ships don't change my influence?)

so I don't know what those stale data are but it's probably that rather than destroyed, cause I can see destroyed on object list and they don't have science or influence attached. Could prolly attach debugger and confirm it comes from the stale.. or delete just the stale part of condition

An option would be to just show the current values of a few empire statistics as calculated on the server for the graphs in the pedia.

graphs I think leak total information and are routinely disabled in multiplayer games for that reason, meanwhile "Empires" window is calculated based on what you know, which sounds better (just needs to take stock of this "stale" data issue)

Yes, they are different. Using the statistics is still an option, though.