That being said, while Satoshi's idea with reclaiming disk space is not implemented exactly as he describes, we do have blockchain pruning. This however does require still downloading all 110+ GB however it will not all exist at the same time on disk. Currently pruning will delete on the fly, once a block becomes old enough, it is deleted to make space for the next block. Pruning has been around for over a year now, it was first introduced in Bitcoin Core 0.11.
You make it sound like pruning is worse, in fact it's tens of times more efficient than whats described in the whitepaper. Both have the need to transfer data in the first place (note that it's 'reclaiming disk space' not 'avoiding bandwidth usage' 
).To the unanswered part of the OP's post:
Sites like blockchain.info aren't nodes. They are custom databases that take up many terabytes of space. They don't validate things (at least not completely) and often show invalid data. What they do is unrelated to how nodes work.
