This is a principle of human nature pointed out to me by ScottAdams and his PointyHairedBoss.  There is a Dilbert strip where the PointyHairedBoss works out a schedule for Dilbert, and bases it on the assumption that anything he cannot understand is easy (magic).  Thus, he commands the poor drone to build a worldwide networking system in six minutes.

If you can understand something, you can reasonably evaluate it.  If you can't understand it (either it is beyond your comprehension, or someone has "hermetically sealed" it so you can't see), you can't reasonably evaluate it.

The phenomenon of "HermeticallySealedStuffIsMagic" is that, if a person can't understand a thing and thus can't evaluate it, one often assumes that it is magically perfect.

ProprietarySoftware gets a big win this way.  Shipping without source code makes you hermetically sealed, so people expect that it is perfect.  Even software houses, which should know better, assume that their vendors' code is perfect, and that all bugs exist in their own local code.  They make this assumption because they can see all the flaws under the hood of their own code, but can't look under the hermetically sealed hood of other code.

Of course, hermetically sealed stuff is not any fundamentally better or worse than visible stuff.  It's just our perception that changes.

----

''Even software houses, which should know better, assume that their vendors' code is perfect''

My ex-company used to chide me for finding and reporting compiler bugs. I stopped reporting them and just left the work-arounds in the code without comments because I got fed up with dealing with it. Remember, it's your fault if you ruin the magic show for the little kids.

----

This has a 'simple' psychological explanation (explained e.g. in JudgmentUnderUncertainty): The AvailabilityBias leads people to estimate probabilities by the ease they can come up with examples. If you (intentionally or not) take this ease away (and most bugs you can imagine come from the source code you can inspect), then the probability is consequently estimated lower (if the seal is perfect, then the estimate is zero). -- Gunnar Zarncke

----

See CharlenesMagicMethod for a brief anecdote of what happens when you have to fix a magic blackbox. You need more magic.

----

PerfectMagic can be a good thing though. Magic is bad unless it's perfect.

----
See also MagicTechnology