I believe it means 1) not being afraid to outcross to another related breed or to a non-registered 'landrace' dog if the gene pool needs refreshing, and 2) do not breed for extreme breed type, as this almost always weakens working ability.
This type of outcross has been done to save some breeds such as the basenji and the husky.
What I find especially compelling is that (in)breeding for show type and (in)breeding for performance are merely two sides of the same coin. In huskies, breeding for sled dog speed (race performance) did the breed no more good than breeding for fluffy coats. The analogy to the gsd are those working lines folks who feel that any narrowing for working ability is a good thing, unlike the superficial inbreeding for show type, which most reasonably intelligent people readily acknowledge to be a problem.
One needs a much broader vision, a patience to temporarily lose track of traits you like and want to preserve, and a willingness to defy the breed organization if they are rewarding genetic disease through diminished genetic diversity. Fortunately, the gsd gene pool is wider than many breed gene pools, but this does not mean this is not a problem for us.
In the American gsd show ring, extreme breed type is palpably obvious. The simple fact that show ring gsd's and working gsd's in every country are seemingly drawn from different color gene pools is also an alarming observation, suggesting that something is wrong.
Yes, that's right, beetree, but I believe the dogs used had no pedigrees. Same with the dogs used to increase diversity in the Chinook sled dog. The dogs were ancestral to the registered Chinooks.
I was surprised to find there were problems with diversity in the racing sled dogs. For 2 years, I lived a short distance from a very large sled dog kennel. The owner of the kennel used a very wide diversity of breeds to cross with huskies to get dogs suitable for both racing and touring.