That’s cool! I will check it out.
Maybe, for those of us in small growing spaces, mass selection of the whole population is the answer. Just save and replant the most seeds from your favorite plants, and a few from every other plant that successfully makes seeds (unless you actively disliked a plant for an important reason). That’s something a grower in a large space couldn’t easily do, but a grower with a small space can.
If you really wanted to do group selection, then one idea would be to save the seeds from each mother plant separately. And then, in each spot, plant exactly one seed from each mother. Then save seeds from all the plants in the group that you judge performs best.
If the group that performs best is in the same spot two years in a row, then it would be time to figure out a test to see if that spot of the garden is better than the others, rather than it being a genetic factor that made the group there superior.
I’ve been doing a lot of thinking about the group selection thing, and here’s what I’ve come up with so far.
In the first years of any landrace, selecting from the best plants is probably the way to go; I think the group competition/cooperation effects are probably more subtle and will be obscured by all the other differences. Once the plants are all reasonably adapted, it would be time to start thinking about this potential issue.
For some plants, it doesn’t matter. A more competitive, taller kale simply means more kale greens.
For other plants, the way they are planted minimizes inter-plant competition. Squash hills are not going to shade each other much.
For plants where it does matter, I think there could be two approaches:
Plant the main crop close together, but plant the seed plot at a wider spacing, and interplant with a low-growing crop; that way, you’d still be selecting for competitive vigor against weeds (for which the low-growing crop is a proxy) but not for competition against other crop plants that would cause increased height.
Plant the seed plot plants in clumps of three or four, as @UnicornEmily suggested, and harvest the seed from each clump separately; save seed from the clumps that had a better than median yield, and repeat every year going forward. The clumps would still have to be spread apart so they didn’t compete with one another.
Edited to add: maybe we should also apply this to companion planting. So if we’re planting, say, pole beans on corn, we should select for seeds from the hills where the average yield of the corn and beans was highest; we wouldn’t want to select from individual plants where the yield of either corn or beans was highest, or just bulk select, or we could inadvertently be selecting for successful competition of the crop plants against each other.
Yes to selecting for weed tolerance by planting edible crops with them! That’s exactly what I want to do! I’m willing to weed, but not everyone wants to, and – far more importantly, from my perspective – I want to select for crops that are happy being grown in a crowded polyculture, so I can grow more crops. 
She’s referring to a study done, in which a group of chickens was selected for individual egg laying, while another group was selected for overall egg laying. In the long term, the second group out-performed the first. If I remember correctly, what they ended up accidentally selecting for was aggression of the hens and the ability to bully all the other hens to stop laying.
They stopped the test after less than two years. It is important to note that this was a human controlled study, with a specific goal. In general you are correct, but in the context of the study deliberately selecting for high egg production on an individual basis resulted in lower production overall.
Interesting! So if you want just one laying hen, an aggressive one might be a good fit. If you want a laying flock, you should optimize for hens that cooperate and form a supportive community.