Yes, that's exactly the point. Bitcoin is insanely inefficient because PoW is insanely inefficient. Unlike other "inefficiencies" where there is a logical relationship between "energy spent" and "useful work accomplished", with PoW there is no such relationship since the energy isn't spent doing useful work (i.e. processing transactions). Yes, more power into the network makes it "more secure" by making it "more expensive" to perform a 51% attack. The increased expense is distributed across the network in the form of increased difficulty. If a 51% attack was already implausible (which most bitcoin enthusiasts would say is true), then making it even more expensive doesn't actually do anything useful.
Trust-less consensus may be the same. Nature doesn't care either way.
Well it depends what you are considering. Arguably, yes, your computer finished the HTTPS connection long before you got around to reading the words on the page itself. But the scale of an idle CPU compared to a 100% CPU is not even close. Mine is sitting at 2% use while I type this.
You could put that money into mining actual gold, and literally make 4x as much as mining bitcoin [0].
[0] https://www.theguardian.com/technology/2018/nov/05/energy-co...
I can't imagine the processing being done by my computer right now is all strictly necessary to read and post these comments, and I'd wager this site is more efficient than most these days. Still I accept reality and pay these costs in order to talk with you on the internet because this is the world we live in. It looks like these nice people at MIT are working on some of the inefficiencies of cryptocurrencies, so let's rejoice.