TRIZ paints a straightforward and naive picture of evolution of implementation of a function:
For example, the means of transportation evolved from carts to carts + ships to carts + carriages + ships to carts + carriages + autos + ships + planes to carts + carriages + autos + ships + planes + rockets, etc. Every new implementation does not necessarily eliminate the previous ones. In most cases it just fine tunes their use. For example, advent of autos did not eliminate carriages. It just narrowed the scope of their use (for tourists only, for instance). Also carts are still in use in some places and under certain circumstances.
The progressive diversity of the ways to implement a function is the only real direction of evolution. All others, such as transition from a macro- to a micro-level, or from a mechanical to electro-mechanical and then to electronic schemes are imaginary directions. They are the product of the distorted and partial perception of the process of redistribution of the areas of use of various implementations.
Consider, for instance, railroads. In XIX they were the primary implementation of the long haul transportation function. Railroads were used for transporting everything (from passengers to cargo). And they accounted for the most of such transportation. Then automobiles and planes came about. And redistribution of areas of use of all ways of transportation took place. Airlines picked up the most of passenger traffic. Long haul trucks picked up most groceries deliveries. Railroads still transport grain and a small fraction of passengers.
Similarly, when electro-mechanical devices were invented, they took over from purely mechanical devices in some areas. This is what TRIZ takes for the trend of evolution. But in fact the trend is increase in the number of possible implementation of a function and redistribution of areas of their applicability. The difference between two visions is huge. The first urges inventors to replace mechanical schemes by electronic schemes everywhere (or replace solids by fluids by gases etc). Whereas the second vision suggests that there is a room for all depending on circumstances. Attempts to push the envelope may result in corrections down the road when, for instance, electronic devices are replaced back by mechanical ones in certain applications. Similarly premature transitions from a macro to a micro-level (or from solids to fluids and gases) may entail the flurry of the opposite transitions some time later.
There are no unidirectional trends of replacing older implementations by newer ones, which TRIZ seeks. There is only a unidirectional trend of increasing the number of available implementations.