They don't take up the "same space".  In order to validate blocks utxo data needs to be quickly accessible as the access speed bottlenecks validation speed, witness data doesn't need to be stored *at all*, since once you've validated it once you can forget it, so the long term cost of witness data is orders of magnitude lower.

Unfortunately the weight calculation can't disregard witness size completely because if its too disregarded it will make nodes bandwidth constrained, because witness and non-witness data are equivalent for the purpose relay at the tip.  Most nodes are far from bandwidth constrained and other known improvements at the time segwit was designed were expected to get nodes 2-4x bandwidth reduction (in terms of their ongoing p2p traffic).

In any case, the perspective you're adopting is a confused one-- I think a confused one specifically engineered by malicious parties attempting to engage in consensus cracking.

There is exactly one metric that matters when it comes to the ability to spam and the cost of spam: the capacity of the network relative to the demand.  Segwit did increase people's ability to add spam, but it did so purely by virtue of adding capacity.  Any other way of adding capacity would have the same effect.  Perhaps the particular spam *encoding* chosen might be different depending on how the capacity limit was constructed, but that's immaterial to anything you care about.   (and fortunately, the weighing scheme has caused the spammers here to encode their bullshit as witness data, which radically lowers the carrying cost of it for the network).