That's the basis for "conservatism": recognize that many processes, rules, barriers, morals, norms, etc. have been hard-won and hard-learned at great cost (money, lives, time), that we may not remember/understand quite why they're in place but they're there for good reason nonetheless, and abandoning/destroying them out of ignorance may doom generations to suffer until they re-discover why those were enacted in the first place.
That's also why rebuilding systems from scratch (the old "just rip it all out and do it again") usually doesn't work: there is a great deal of wisdom, however poorly implemented or documented, buried within that system. Starting over abandons those minute but critical bits, and must be re-discovered the hard way.
And yet the crowning achievement of conservatism in the last thirty years has been tearing down the processes, rules, barriers, morals, norms, etc. that were put in place during the Great Depression to prevent capitalism from tearing itself apart again.
Actually, although I don't agree with the idea, this is part of a conservative process. What is supposed to be conserved is the established order (Rich over poor, work over lazyness, God over men, etc.). The premise is that order has value, even if unfair, and that an established order is better than no order at all.
It is the blessing, and the curse, of culture that generations have suffered to find a balanced set of behaviors and rules. Change them at your peril! Its kind of like a system of natural selection - societies must be conservative to survive long-term. Rules change over generations, variations are few since the survival of the collective depends on (some of )them.
However, the world and the society we live in is changing so quickly (technology, cultural mixing, global warming, you name it) that the new norm for changing the rules is not generations but years or months. Using Facebook? Change your settings today or lose your entire life history to strangers! and so on.
I remember just such an occasion a few years back where someone working for me did exactly that. When asked why he hadn't noticed the associated unit tests were failing the reply was "those weren't compiling so I commented them out".
My favorite is when there is a fence with a sign saying, "Erected to prevent buffalo from stampeding
our corn fields. It is OK to remove after 1850." Meanwhile there is now a neighboring cattle ranch
and the fence is still actively performing a blocking function :)
That can also lead to a hell of a lot of analysis paralysis, however.
I know because I share your approach. Having worked with a great many legacy / inherited systems, most with little or no documentation, and rarely with much in the way of contact with the system designer, it's a bit of an occupational hazard.
Sometimes you've just got to blow down the fences to see what happens. Sometimes it's nothing, sometimes it's the end of the world.
Fortunately, in technology, resurrecting fences is generally fairly straightforward, but not always.
That said, the worst fences are the ones you take down without realizing they were there ... which turn out to be critical.
What do you do if the man that put up the fence does not know why he put it up himself, or if he put it up for all the wrong reasons but is not there to explain them to you?
Hence the proper usage of "question authority": with all due respect[1], investigate the reasoning behind the authority's action and either be enlightened as to the reasons or confirm that the decision makers were, in fact, ignorant/idiots. Alas, the term got misconstrued to validating "ignore those who likely know better, and act confidently in your own impulsive ignorance."
[1] - "all due respect" has also been stripped of proper meaning, relegated to a barely-veiled insult.
"I don't know why this is here and I've tried to find out but hit a dead end so I'm getting rid anyway" is better than "I don't know why this is here so I'm just getting rid".
"Never take down a fence until you understand why it was put up."
I used to use that one a lot with developers trying to improve on a predecessor's code...