More efficient, but much more expensive. I'm sick of people handwaving $100 per kWh. That is two orders of magnitude off where it needs to be to do anything more than virtue signal.

Meanwhile multiple grids are now paying renewable to curtail, because guess what, the variability is correlated (it's the exact same damn mathematics we used to fuck up the entire global economy in 2008, which is why I'm so surprised people are handwaving that too, but whatever). If you want to minimise cost without relying on gas to save you on dark still days, you want a cheap use for the surplus, round-trip be damned.

100$/kwh on a battery that does 1000 cycles is 10c/kwh, 5000 cycles ("Claimed" lifepo4 these days), that's 2c per kwh. These aren't that unreasonable, albeit one would need to account for cost of capital and so on increasing these effective numbers.

Batteries are already economical in most grids where they can arbitrage daily prices of 0-10c during the day to 10-30c during the night, with the occasional outlier event contributing dollars per kwh.

They will never load-shift across seasons, agreed, but for daily loadshifting they are already economical, and being 90%+ efficient (and very simple/easy to deploy and scale) is part of why they're popular. It opens up power shifting opportunities that aren't just daytime solar too.

Kilowatt hour of capacity and kilowatt hour delivered are two very different numbers. Sources rarely distinguish, and you're almost certainly confusing them if you think batteries have to get down to $1.

Correlated errors are a problem in all sorts of places. Most statistics assume everything is independent; super important to verify that before drawing conclusions.