Many guides haven't been updated since MGLRU was added to the Linux kernel. Before MGLRU the kernel's memory management, especially in low-memory conditions, was pretty poor. The kernel would often choose memory pages that were still fairly active to move to swap which slows things down unnecessarily (since those pages would often just need to be moved back) which is where many of the recommendations on setting a low swappiness comes from (basically to tell the kernel not to swap unless it has to).
Now kernel memory management is much better and the kernel should really only swap fairly inactive memory pages. This ends up being a good thing because those memory pages will go to the compressed memory device (zswap or zram) where they are much "cheaper" to retrieve instead of being written out to disk. This frees up more memory for active things, generally meaning the user can do more things at once before they start seeing performance issues.
Now granted, I haven't done a lot of benchmarks to confirm this one way or the other so I could still very well be completely wrong, but this is my understanding after all of the research I've done on the subject (I've been the Solus kernel maintainer for the last couple of years or so and the fact that Solus has Zram/Zswap at all is something I chose to make sure was enabled).