By this logic, should tiny have a smaller penalty than small? Huge and large have the biggest gap, so why are they both at 5%?The way I would justify it would be that, simply, there are more tiles per player on bigger maps, therefore, room for a couple more cities. Of course this is dependant on map generation, which is why I think it's important to know how many cities a player is expected to have. But a simple calculation of total tiles gives the following results.
Tiny - 504 tiles per player
Small - 462
Standard (8 players) - 520
Large (12 players) - 554
Huge (16 players) - 640
Even beyond that, why does having enough land for 1 more city mean that my other each produce dramatically less science? I don't see any connection.
Here is an example. Let's say the players get 6 cities on standard and 7 on large, which seems reasonable. On standard, policies have a cost of 135%. On large, they cost 130%. So for culture and science to be roughly the same speed, the large needs to produce less total science while having more cities. It doesn't make sense.
Extending to small, an empire with 5 cities would have a cost of 150%. So this player needs substantially more total yields from a smaller number of cities.
So even if the standard player's 6th city (and large players 6th and 7th cities) earn 0 science ever, they still discover techs faster than on small.