ff:
Please see the great article at the link posted by XS, above. Slippage is not part of the HSD solution. At any rate, consider the end result. The Prius is a car that weighs almost 3,000 lbs. During steady state cruising at highway speed, it still gets substantially better mileage than similar size/weight cars with either stepped automatics or manuals. Just a few minutes ago, I finished a run from Pensacola to New Orleans, just over 200 miles, almost all of it done at just under 80 mph, and my Prius averaged just under 46 mpg. During this type driving, there is only a small in/out flow to/from the battery. I have a CAN View unit grafted into my car, so I can monitor a lot more data than the driver of a "stock" Prius. This is an example of one of the many screens CV offers:
The one I use the most has ICE rpm, ICE temp, instant mpg, in/out charge flow to the battery (CV gives you three such screens, and you can select the displayed parameters for each). During steady-state cruising, there is only miniscule charge flow to or from the battery (as compared to what you see at low speeds, or when you floor it).
What you're missing is that it does NOT work on "slippage". If the ideal setting at one particular moment calls for more ICE rpms, that causes MG1 to spin more as well, putting more juice back into the battery for later use. There is no slippage at all in this unit. Power applied goes either to driving the wheels, or generating electricity, which if not used right away, goes to the battery for later use. What many people miss when looking at this design is what would be slippage in a common belt or chain and cone CVT gets "captured" by the motor-generators in this design, in the form of electricity, and re-used later.
Even with the neat, small, Atkinson cycle ICE, and the very carefully done aerodynamics, this almost-mid-sized, 3k lbs car simply could not achieve the highway fuel economy it does with a "slippy" inefficient transmission.