The biggest technology that people think is working, but it doesn't?

Currency and debt.

Currency is a lot like electricity. It's a very portable form of power, easy to store, easy to convert to something else.

Like you can have a lot of political power, you can have charisma, skills, or military power. But you can't convert that easily into the other, the way you can't convert a river into lighting your house.

But the thing with electricity is that it still follows nature. You can't convert 1 MW of electricity into kinetic power, then convert it back to 1.1 MW. You can't lend it to someone with less, expect them to give you more, and have this system run forever. You can't just print as much of it as you think you can get away with.

Banking is a kind of arms race. If nobody used banks and usurious debt, everyone would be better off. But the people who do it have an unfair advantage, forcing others to as well. Except now we're pushing it as far as the concept can go, with every country taking well near their GDP in foreign debt.

The Bitcoin boom didn't come out of nothing. Some people lost faith in fiat and wanted to give alternative currency a shot. The bubble only happened when crypto attracted the attention of those who broke fiat in the first place.

The biggest piece of technology that seems fine but isn't is automobiles. They are much too big and ponderous for the 90% of the time they are moving one person around a big city at 30mph. Electrification will produce much smaller, lighter, cheaper, safer vehicles for this purpose.

Advertising data algorithms. Usually they are just promoting things you already have recently purchased.

Advertising. There is not much published about how actual money people get on their investment in the major platforms, including online. It's not tech that "doesnt work", but it "doesnt work half the time". But since a lot of online advertising is fueled by the wave of high valuations , people seem to be fine with not knowing whether it works or not.

Science. I can't take any study seriously until it's been replicated at least once, preferably twice. Does anybody know if there's a search engine for replicated studies?

"machine learning" in large corporates, when in practice this is 90%+ just maths/algorithms

facial recognition, especially when used by UK police against certain minorities

ML models making decisions about who gets to receive benefits (according to a bunch of articles in today's Guardian)

Recommendation systems.