All bio-structures are built of carbon, hydrogen, nitrogen, oxygen, phosphorus and sulfur(CHNOPS), and differ only in number and spatial arrangements of these elements. Hence, if we start with the simple self-replicating molecule(evolution's starting point), then the only way to find new proteins, molecular machines, organs or organ systems is by re-arrangement of CHNOPS.
The idea of evolution is based on two fundamental premises. The first one says that mutations cause variations or re-arrangements of CHNOPS. The second one says that the certain variations will be selectively preserved in response to environment. For example, when functional protein exists, and it is beneficial in the current environment, then - it will be selected. That's fine.
But that begs the question: how did this selectable combination of CHNOPS(protein) came to be? This is the crucial and the most important question. There are virtually infinite number of ways in which CHNOPS comprising protein can be arranged, and most are junk, or non-selectable arrangements. For e.g. for a protein 92 AA long, with 10e122 possible AA combinatios, there is only 1 in every 10e63 functional sequence*. On the other hand, published extreme upper limit estimates puts the maximum number of mutations or CHNOPS re-arrangements at 10e43**. So, the total number of evolutionary CHNOPS re-arrangements is 20 orders of magnitude insufficient to find only one selectable state for evolution to preserve - a protein, let alone molecular machines, organs or organ systems.