I think the other reply about TX counts vs. people-hours of power convinced me that indeed cryptocurrency is much less efficient than conventional currency even if you put large error bars on that analysis. I do think it's a solvable problem.
Your reply about trust though...
Trust is massively less expensive until it's not. The trouble with trust is that when it breaks "fixing" it is immensely painful, often requiring major political upheaval or worse. In extreme cases people die when trust has to be "fixed."
I do still wonder... what happens when you amortize the cost of trust across say two hundred years time? Reminds me of the cost analysis of nuclear power. Nuclear power is cheap until Fukushima happens, and one Fukushima amortized over even 50-100 years renders nuclear power more expensive than any other energy source.
I do not think it's a coincidence that the cryptocurrency explosion happened right after the 2008 financial crisis. The level of corruption revealed by the crisis and by the nature of the state and financial sector response to the crisis (selective bailouts, bailouts only to the rich, bailouts that preserved the wealth of those responsible, etc.) showed that our trust in the financial sector and possibly in larger institutions is dangerous. People started looking for alternatives. I have doubts about whether cryptocurrency would have caught on to the level it has prior to 2008.
What cryptocurrency needs to be successful is some alternative to proof of work mining. I'm not convinced proof of stake is it since it has a lot of other problems.
What conventional economic systems need to be successful is a housecleaning and a restoration of public trust.