Back in 6.0 days, according to the rulebook elements with falls and jumps "landed on two feet" were supposed to earn no credit. This guideline, as written, would apply primarily to free skates.
Of course there was no way of knowing exactly how each judge considered various types of failed jump attempts, since all the scores for all the elements and many aspects of the skating skills were all boiled down into one mark.
In the short program, the deductions for various errors were more specific, closer to the negative GOE guidelines in IJS:
Note that the penalty is lower for landing only one jump in a combo on two feet, or for falling on the second jump.
"Less than required revolutions" is the same deduction as a fall on a solo jump -- and the same for one jump in a combo as for falling on the second jump of the combo. (Presumably, if both jumps were less than required, e.g., a single-single combination, the 0.4 deduction would apply.)
And the penalty for omitting an element altogether is higher than the penalty for a fall.
I'm not aware of specific written guidelines about how to set the base mark for a short program -- i.e., the starting mark for Required Elements before taking the deductions. But surely the difficulty of the jumps attempted was one of the considerations.
E.g., a short program with a 3Lz+3T combination would start with a higher base value than a program with a 3T+2T combination. Then if they both had the same error (e.g., a fall) with the same deduction, the former program would end up with a higher Required Elements score both before and after the deduction.
Because the short program is about required elements, and there are minimum revolutions required for the short program jumps, there has always been more reward for actually going for at least the required difficulty than in skipping an element or watering it down to less than what was required.
To the casual viewer, an easy program with no blatantly disruptive errors looks "nicer" than a harder program with an obvious error or two. But even in 6.0 free skates a clean program with 4 triples and a bunch of doubles, for example, might lose to a program with 5 clean triples and 2 failed attempts. Or maybe not. A lot depended on everything else about the program, not just the jump count or the visible error count.
In a short program, a program where the skater at least attempted all the required elements would generally fare better than one in which the skater completely omitted an element or singled jumps that were required to be at least double or triple.