The main difficulty is to juggle with a few practical concepts in the same time, without forgetting one in the flow to don't be fooled. It's the stack of all these concepts than draw a pattern to play with.
I was under the impression that taking a plant to F6 and beyond was desirable as it resulted in what you could consider an IBL.
The term IBL is just a way to express than you search something in a limited range and than you expect a limited variation to find it. Let's say a specific line than start in F2, this "F6 thing" is just an arbitrary appreciation than was injected in your mind by the marketing only. At the moment you cross twos specimens from the same generation and from the same line, it's an IBL. No matter if it last only one generation of a dozen, it's an inbred line in the absolute. The Fx mention is only here to indicate on wich level of pressure we are talking about.
By example, you can act by suppression in pushing the line very far in a trait (in term of pressure). Basically killing all specimens during a few generation than don't answer to the stats you choose, and no matter the quality of all others traits. I take this radical example to draw obviously something without any nuancies.
Another example is the open pollination wich is an IBL too if all the specimens are from the same initial specimens.
Twos cases, twos opposite goal. The twos are not the insurance of any notion of quality, only your hands and your decisions will give the level of quality of a line. So the selection than you apply, and the skills you have to determine the patterns. The more patterns you master, better will be the whole quality off course.
The DJ Short quote talk more about the difference between an heterosis state of a blend and its late state after a fews inbred generations.
I loop again on the fact than the desirability of a line is highly relative to your choices, your tastes, your skills and on the profile of a line. This line can have a "fire weed", but can be also only a "fire breeding material" too. Like by example a dominant structure on wich you refine a specific final product with others bloods (yielder, sativa, CBD whatever...)
But if the plant can lose the good traits when it gets to that point, how do you overcome that?
With all the tools at your disposition in regard of the methodology you have. Basically, keeping a reference alive (in any form) until you replace it by better. And even in the case of a simple mom/pop, the stories of breeders than was doomed by the lose of only one cut are legions.
The difficulty in keeping several desirable traits over many generations is where the skills of the breeder come in, I’m assuming.
Exactly but just before this sentence you don't saw a major point : the linked traits and the inherent equilibrum of the genotype in question.
In your example, the potency can be directly linked with the yield. In a good way, or in bad way. Or worse, indirectly linked in both way by an intermediary traits. The Chem and its inherent variegation is the best example i've in mind to quote a critical intermediary trait.
Is this why there are not many breeders creating IBL of their strains?
No. It's just because the market is a darwinian game, nothing else.
Or are some plants just impossible to breed true?
Some plants and some lines are genetically programmed to extinct in being very good (from the stoner's eye). It's not binary like true/false, it's more linked with the rarity of this expression inside the strains in question and with the strength of its equilibrium.
By example if your cut XYZ is fire and if this "fire" is a weak mountain of interdependent linked traits ... the quality will not resist to a change of a single of these traits. Like dominos.