It's not censorship to not want them.
Discussion
Is it censorship to remove the ability to put them on Bitcoin? I'm guessing it's both yes and no based on the definition of censorship. But ultimately it's just a design choice and the line has to be somewhere.
It'd be pretty archaic to get rid of segwit and opreturn. I can see a subset of people pushing for that as a solution, and others magic hand waving, but I'd rather have the features and figure out ways to adapt.
But I'm speaking from personal desires too. While much of the image data is rubbish, it's also like a Easter egg hunt, and stuff in a node can be referenced in a sovereign way. It's also a form of permissionless censorship resistant communication which has value.
I'd rather have a 4MB jpeg filling a block, then blocks with tx increasing utxo size by 50000 per block, or a ton of operations, or worse, encoding data as value amounts in outputs if there was no other way to embed.
This is why I prefer focusing on ways we can mitigate the negatives while retaining positives and acknowledging there's always going to be a gray area.
After all.. the overwhelming majority of blockspace didnt benefit me before inscriptions or stamps as I wasn't on the sending or receiving end of most of those transactions. But there's residual value in the network effect. The more others use it without directly benefiting me, the more potential benefit I derive through an active system.