I'm trying to define the exact nature of the problem. Your analogy is helpful, but it doesn't answer the question on what *you* mean by 'understand how the machine works'. What if ET has a stockpile of older discarded machines and as chief scientist I have the opportunity to look into the construction of those models (assuming they have limited functioning)? Wouldn't that be a good start since I can see what led up to the present machines?
What I need to know, is do you want to discuss the machine that ET left, or do you want to discuss the earliest traces of a machine possessing some form of knowledge? Looking at all the previous models which attributes of knowledge are you most concerned with (i.e., this ET culture is an ancient civilization going back 4 billion years and we see they've made machines going back almost 4 billion years)? For example, are you concerned with the ability of the machine to possess memory? recognition?, cognition?, sensitivity to inputs?
It's easy to throw out a whole conundrum of ideas and ask that it be answered as a whole, but these 'machine functions' are not one attribute. All the functions are visible in the one leftover ET machine, but looking at the past collection of machines we don't see that. We need to distinguish which attribute (one by one) that you are most concerned about. Without this analysis we are just throwing out riddles without any serious attempt to understand how those riddles may be answerable.
Warm regards, Harv