There is some truth in this, but be careful of sweeping statements. TLS/SSL encrypts and can slow down internet connections, but noticeable for general public, I am not sure?
Again though, what appears to be and what is are not always the same. The xor stuff we do is hard to grasp, but does indeed provide many amazing things, the ability to replicate/emulate huge networks in very small segments as an example (the group sees the network from only the group perspective). So to get a transaction done for instance we can do it in a group, fully encrypted end to end, way faster than bitcoin (not sniping or competing here, just a point) where the transaction needs transmitted to everyone.
I will give you an interesting part of our history here. A few years back we did some experiments and tests with the NHS in our local county. This was to provide a single server for diacom (xray) images. I was disappointed in the single server as it meant using pretty much only our encrypt library and very cut down versions of the rest (no network).
We used a multi disk ssd server with 24 cores. So we could test encrypt and use drive in a manner that was multi user.
Those drives were 300Mb/s and we could write to them at that speed with unencrypted data (it basically streams). We then put the data through our system and the encrypt library was able to process it.
So we has a 300Mb/s system and added encrypt.
Our system then processed the data at nearly 1Gb/s !! so we added chunking, encrypt and real time de-duplication and went over 3X faster
What? we add several steps and the process is faster, how so? It is due to the parallelism we could get, rather than streaming a file start -> end, we could split it across cores and then use the multi threaded aspect of SSD drives to write multiple parts at the same time.
So we encrypted, went faster, but it was not over. Even on a single server with not so compressible images (dicom) we were compared with the existing de-duplication system, Microsoft. Microsoft deduplication (which I believe is not real time) on that single box achieved 8-9% de-duplication figures. We achieved (with real time deduplicatiion) over 48% savings.
So all is not always as it seems, encrypt for instance can be done at a throughout of 800Mb/s with some ciphers so if there is a decent pipeline used in the process and writing to disk or network is slower than 800Mb/s (they pretty much all are) then this slower link will not be the part of the chain that slows you down (so add or don’t it will make no difference).
So a disk (old magnetic disk) has a xfer rate of 30-40Mb/s, ssds are around 300Mb/s and if a cipher is 800Mb/s then the impact can be near zero, in fact just do what we do and do not use convention and the normal ways (parallelism) can prove we can add more functions, features and process and go faster.
SAFE is hard to explain and these parts are hard, but give me great hope that as we progress, things like this show us that is seems we are onto something very big as we find more and more really nice surprises like this along the way.
In terms of that commenter, I am sure there will be elements of truth there somewhere, just like any good misinformed self proclaimed expert, generally will have When he stops telling folk their IQ is not good enough to talk to, or he actually presents some real info rather than cat nap dreams I would take the info provided with a pinch of salt.