In the article I wrote, I didn't want to imply what would be "right or wrong", just inform what happens, and that to make the most of it, both methods are necessary. It's important that the actual breeding pair should be representative of the focus of the linebreeding characteristics aimed for.
The 10-20% figure is pretty subjective, and totally dependant on what's bred to what after all. Your best genes may go into 25% of the pool, but not necessarily into the same quarter of your litter, into the same pups. Depends on how many separate heterogenous characteristics you're hoping to make homogenous. Then you have to re-select from the results, and continue refining.
As a general rule, linebred dogs can be better for
producing, and outcrosses are often better
performers. And often, the best performers are not necessarily the best producers (depends on the mate matched with).
This distinction of producers vs. performers can even be seen between littermates. I guess my favorite example would be Ellute von der Mohnwiese and Eros von der Mohnwiese. One's a better producer, the other a better performer. They're linebred, but hardly in any significant way. 5-4 on one particular dog, but through alternating sexes (M>F>M>F>M--M>F>M>F). Their dam was somewhat linebred, but not in a very influential way.
If you have a subject that is very outstanding, but want to plan some improvement, linebreeding is a good way to really take advantage, and in a lesser amount of time. Then follow with a suitable outcross, an equally important component.
It's not any kind of magic formula, but it definitely helps keep a trait or characteristic
within the line that may exceedingly be rare among the general gene population. It seems we are noticing some shortages out there for balanced-drives, genetic-grips, good-hips. Challenging to find one or two, much less all three traits in a single dog. Throw in a couple other desireable characteristics, and the math is less in your favor (10-20% might be optimistic).
That low a percentage just wouldn't warrant the years of line breeding.
linebreeding is actually the "quick-fix", it would take much longer with only very carefully selected outcrosses... lifetimes perhaps, depending how refined you want to go
The idea is to fix traits and every outcross adds 50% new genes to the mix and negates the linebreeding until the offspring can be bred back and reset them. EVERY outcross will do this.
the outcross is the "finished product"