The not so good news:
A software bug can remain well-hidden due to the numerous lines of often complex code that is best understood only by the original programmers. Then there is the programmer's intended interaction with the not-so-finite real world which becomes practically an ethereal matter. The resulting trail from said defect may only be visible to a select few, if there is any trail left at all.
True, indeed. Software generally does not fail-soft. The heavy dependence on software that is 'undocumented' and inaccessible to all but a select-few company insiders means no accountability, no transparency, and no independent verification of the critical control processes and underlying assumptions that operate most functions in modern aircraft.
With engines, airframe, and avionics coming from several sources, the potential for hidden flaws in integral control software compounds with each design change and each 'model' variation thereon.
Imagine further how the design integrity of these aircraft may degrade over decades of 'minor' control and subsystem repairs intended to deal with specific operational problems encountered along the way, many of them done on the cheap without the top-to-bottom integrated-systems evaluations that look at the bigger picture of overall vehicle reliablility in light of each and every modification.
The various civil authorities need to develop enough backbone to compel full disclosure ( confidential, where necessary ) for ongoing technical evaluation of algorithms, principles, and all other operational data about the software controlling ALL airbourne controls, systems, and subsystems used in civil aviation. Only when this is done will the true facts be available for inquiry and independent evaluation when incidents arise.