To be clear I understand that hybrids are current best practice. But the hybrid practice seem antithetical to open source agriculture.
BTW would you say that seed producers have no motivation to develop non-hybrids that will perform as well as hybrids?
I disagree, open source ag is completely compatible with hybridization. And if open source couldn't somehow deal with the benefits of hybridization, then open source ag is not a route we should follow.
Farming is a transformation of inputs into food. It's not about seeds and propagating seeds every season... that's just an attachment to a practice like being super attached to, say FTP as the only way to transfer files and deploy a website.
Open Source Ag is completely compatible with maintaining two strains, crossing them, and then distributing those seeds. Having widely distributed access to these seeds, and information about them, gives the chance for more innovation in the crossing, too. The question is whether somebody can get compensated for the efforts in discovery...
Anyone interested here might also wish to read Carol Deppe's "Breed Your Own Vegetable Varieties". She's a Harvard-trained geneticist and amateur vegetable breeder, with special interest in open-pollinated varieties derived from proprietary F1 hybrids. Her book extensively discusses the underlying biology, the practical breeding process, and the legal situation of such work.
For instance the modern Irish Wolfhound was created by an nobleman who, worried about the breed dying out (there hadn't been a wolf in GB in over a hundred years), bought every wolfhound he could get his hands on, and crossbred them with mastiffs and Danes to build up a healthier population (while maintaining stature). Then he selected back to phenotype and went from there.
One of the groups trying to save the American Chestnut is doing the same thing, though I think their methods are recklessly optimistic. They've crossed residual chestnut populations with Asian chestnuts and then back breeding those to try to get just the blight genes and nothing else.
I say reckless because the smartest thing to do would be to identify the resistant genes and test each tree to see if it's carrying the gene. Instead they are infecting every tree and culling the ones that are susceptible. Effectively making a petri dish in which the blight can learn to infect resistant trees over time.
In some ways I think the GMO people may be the less insane of the two, and I'm very much not a fan of the transgenic surface area for pathogen adaptation across species.
If that is true it amounts to the same thing you are proposing. Sounds just like making a speed-congenic mouse.
Are they not selecting with markers at all?
Hence a petri dish for pathogen selection. IE, insane and reckless.