The below has now been carried out. If there's a feeling that random-noise should be merged into randomness this will always be possible to be done later (possibly after a meta discussion).
As it stands the tag name is not really optimal.
This is because it could refer to both the noise family of protocols as well as the more usual variation in measurements / the randomness introduced to mask / hide actual values (eg in FHE).
Looking through our search results for the term "noise" it appears we have quite a few questions, so it seems sensible to create a random-noise which would be a specialization from standard uniform randomness to some usually more gaussian distribution (we have some questions e.g. about Perlin and other forms of noise).
Additionally it seems like we also have a few questions about the noise protocol / framework and creating noise-protocol (optionally synonymed with noise-framework) also seems like a good idea.
As for noise itself I think it is too ambiguous to be really valuable here and should thus be removed.