"Security through obscurity is a myth."
It works just fine assuming you get the details right, and while it's true most implementations don't, some do, and they are wildly successful. Saying that it's a myth is a generalization that is simply wrong.
When was the last time you heard about Firewire 1394 DTCP encryption getting cracked? Millions of people have cable STBs with firewire 1394 output. DTCP has been protecting firewire video streams for a dozen years, and no serious attack on it has ever been revealed. DTCP relies a great deal on security through obscurity. Details of the M6 cipher it uses are unknown. Selection of the ECC curve is unknown.
You might think that maybe someone IS working on it, after all, the related (but not the same) HDCP has been cracked wide open, but there is no evidence anywhere that indicates DTCP has been cracked. If it was cracked in some underground lab somewhere you would at least expect to see HW or SW available somewhere that would take advantage of the crack.
But there is nothing, only crickets chirping in the dark while a great example of security through obscurity continues to perform flawlessly year after year after year. And that's no myth.