The first strike with collateral is a big deal. You can wipe an army of the same size taking less than half their losses. It's not only the collateral first strike either. Next to the movement problem which is huge the defenders also has walled strongpoints you alway have to lug your siege over to contest. It takes forever and allows the defender to whip more units as indicated above.
I accept that defenders advantage is higher in Civ5 and Civ6. But those games are way overdoing it often leading to static and boring games. It's also not quite that simple. In Civ4 defenders advantage increases over the course of the game as siege units increase in power and enemy culture becomes stronger naturally. Conversely, in Civ5 the late game combination of Stealth Bombers/XCom is completely unstoppable. It will kill the whole map. (These are assuming nukes are banned by hose rules as they often are. With nukes it's different.)
It's possible people play like that but it's not correct in an FFA setting. Early conquests usually aren't worth it as you do not have the economy to digest them, especially after having to sustain the costs of the war itself. Of course, in direct IP mulitplayer people tend to play on tiny maps so the game finishes in less than 8 hours and that changes the evaluation of early aggression. If you only have 5 cities of course you are gonna see if you can increase that on the warpath. What else are you gonna do. Try to research stuff with your 5 cities? On maps where everyone gets a decent amount of cities the game can be quite peaceful for a while. Really large MP games (Pitboss, Pbem) usually go into the modern era. There have been space and culture victories. Direct IP games are almost always too small for that kind of thing though.