pull down to refresh

They don't sound invisible to me. The translator transaction has a distinctive footprint. If pools censor/filter translator transactions (i.e. transactions which publicly decrypt a previously inscribed cyphertext) then you're in pretty much the same boat as the inscription people are currently in.
They don't sound invisible to me.
The ciphertext transaction is invisible. The translator transaction isn't, and that's intentional. Part of the use case of inscriptions is on-chain discoverability. If my wallet software mints an NFT, software running on other nodes should be able to detect that new NFT without an external communication channel or database. This implies the inscription must be distinguishable on-chain, eventually.
If pools censor/filter translator transactions (i.e. transactions which publicly decrypt a previously inscribed cyphertext) then you're in pretty much the same boat as the inscription people are currently in.
If I perform inscribing first in secret, and reveal the inscription occurred only after it has already been mined, this nullifies any incentive to censor/filter in the first place (beyond spite), because 99.9% of the damage (block space usage) has already been done.
Take street art as an analogy. Inscriptions are like graffiti, except permanent. Currently, vandals are painting buildings in broad daylight: Their progress is easy to observe, and thus easy to interrupt if we wanted to.
However if we start to interrupt them too much, then the little punks will just wait until dark, and paint in the shadows while we're asleep. We wake up and find their works were completed in secret, and by then, nobody can remove them. The damage is already done.