This applies to ANY locking mechanism.
Which is why I stressed "When it's designed and executed well..." in my post. If I'm given a choice between two well-designed and executed knives (or for that matter, if forced to choose between two poorly-designed knives), one with few moving parts, and one with many more, I'll take simpler construction every single time, if long-term reliability is the main concern.
There's 'a lot going on there' with any lock mechanism, if it's going to work well. Throw in more parts, and there'll be a 'lot MORE going on', with that much more possibility for failure.
But I think people are confusing complexity with probability of failure. Simpler is not always better.
But, as a matter of engineering common sense, this is certainly false. All modern breaking systems of cars employ a redundant master cylinder design for instance. Electronic fuel injection is far more consistent and reliable than mechanical carburetion despite vastly increased complexity--sometimes as many as 16 fuel injectors over four fuel rails and not taking into the complexity of the ECU controlling it all. The increase in complexity is probably 1000 fold, but the increase in reliability (among other things) is so great that literally every car manufacturer uses it.
Elevators use a secondary fail safe.
Virtually all corporate servers have battery backups, and many have redundant RAID 0 hard drive configurations as well.
The examples of multiple-system redundancy being used in the world of engineering is nearly limitless. The lack of redundancy and fail-safes in a mission-critical part is evidence of an unsophisticated part, not evidence of superiority.
Axis locks are both easier to reliably produce than crosslocks and are proven to be, on average, more reliable. Where a discussion begins to happen is in the 150+ dollar range of framelocks, particularly on up to about 400 with sebenzas, which is really demonstrating how tight the tolerances on these designs need to be to be comparable to axis locks, which have been available in the 50 dollar range for years, yet with an apparently near 0% failure rate.
As a matter of empirical fact, judging from the massive sample size of bladeforums anecdotes, crosslocks (framelocks and liner locks) have a higher failure rate than say, axis locks.
I'm not saying that the anecdotal evidence is conclusive on which lock is superior. I'm saying that this sort of bizarre argument that complexity necessarily increases the failure rate is false theoretically and false empirically.
I mean, I guess you can turn around and try to spin this into the situation that the redunancy built into axis locks harms the axis locks specifically, but on what justification could you possibly base this conclusion? Both theoretically and actually, it is false.
Let's abandon the silly complexity argument and move back to the poll.