A few general approaches and ideas:
1) If I wanted a variety mostly similar to a hybrid, except just an op version, after the initial cross (to something as similar as possible) I would usually backcross at least once to the hybrid before starting to select. That way I would have material that was genetically ¾ like the hybrid, and that would be already segregating for and expressing most of the genes in the hybrid including the recessives.
2) I especially tend to backcross at least once to whatever variety that has the flavor I want, since that’s hardest to select for, and is usually influenced by many genes. A delicious distinctive flavor may not reappear even in thousands of F2 or F3 plants if many genes are required to make it. The odds go up considerable if you enrich preferentially for the genes involved. Also, it’s considerably easier to select for visible traits than flavor and cooking characteristics; so when a particular flavor or cooking characteristic is a major desired characteristic of the new variety, I generally backcross at least once to the variety with the magic flavor and start selecting from there.
3) If I want one or more new varieties that are thorough combinations of characteristics from two or more parents, I generally proceed by going to the F2, then selecting gently from the F3 on.
4) You can’t afford to eliminate all but the best plant in any generation, because the incompatibility system means you’ll get little or no seed from that one plant. However, you can afford to cull to, say, your favorite six plants and save seed from just your favorite, if that favorite is uniquely worthy, and if it’s close enough to the others so as to cross with them, and if this is in the early generations of selection after outcrossing. Generally, keeping the numbers up in later generations really matters in or to avoid inbreeding depression and end up with a vigorous variety.
5) I often select very gently only one characteristic at a time. So if I’ve crossed a blue, tender-veined kale to one that is broad-leaved and very sweet, and I ultimately want a blue tender-veined broad very sweet kale, in the F2 and F3 generations I might eliminate all the greens and keep the blues, whatever their other characteristics. My reason for choosing color first is partly that is is easiest and obvious, and likely to express fully early. I once crossed a sweet mustard to a very cold hardy one, and it was 5 generations before I got anything sweet. I selected first based upon cold hardiness and leaf breadth and didn’t worry about the fact that everything tasted awful for the first 4 generations. I could select for cold hardiness automatically by just planting thousands of seeds in a few square feet each year, letting winter cull most of them, then weeding out all but the biggest in spring.
By “gentle selection” I mean eliminating, say, half the plants, the half with the narrowest leaves, for example. And leaving dozens of plants. I try to move the frequencies of characteristics I want gently gradually in the desired direction, while maintaining a representation of most of the other genes (until later generations when it is their turn for attention).
6) You don’t start selecting in the F1, and sometimes not even the F2, because you’re mostly seeing a reflection of dominance relationships among the genes at that point, and many genes you want in your final line are recessives that are hidden until later generations.
7) Don’t select arbitrarily. There is no reason to select for green versus brown brassica seed, for example. Various genes and characteristics are linked genetically and physiologically. So you might accidentally throw away the gene that could have conferred something important because it was linked to the seed color you selected against for no reason. I was tempted once to select the deep green plants that were segregating out of a squash cross. I could argue that such plants had more chlorophyll, so might well be better at growing. But I didn’t select arbitrarily like that, based upon just my speculations. Good thing, too. Those deep green plants attracted such huge numbers of cucumber beetles that the plants were badly stunted. And the fruit was somewhat bitter, too. Apparently, the segregation had uncovered some genetic combination, probably, that made more cucurbitacin, and that was linked to the deep green color, somehow.
Be curious. Be flexible. Notice things. Ask the plants what is happening as it happens. If I’m unsure what I want, or want two different things, I might cull the greens at one end of the row and the blues at the other, so that greens interbreed mostly among themselves, and blues also. I might have “wanted” blue to start with, but almost never would I be so rigid as to just eliminate all the green without exploring further. So I save two batches of seeds, one from the greens, one from the blues, and plant them at two different ends of the row or plot the next year, and see how each does, and taste them, and use them different ways, including different ways than I was initially intending. The blue might all turn out to have a delicious flavor cooked, but the green to be more delicious raw, for example. Before I eliminated the green, I would give it a chance to show what it can do, even if that wasn’t quite what I had in mind to start with. I wasn’t quite what my parents had in mind either.
There is an almost supernatural relationship between the plants and the plant breeder as they talk with each other about future possibilities. You will get more ideas about what could be by looking at and listening to the plants and walking through the patch and tasting everything than you ever will sitting indoors designing a plant breeding project.
9) With respect to the cytoplasm: When doing the initial cross, it matters what direction you do it, that is, which variety is the mother. Keep in mind that mitochondria and chloroplasts are integral to energy metabolism, photosynthesis, etc. And they have some of their own genes. And those genes come only from the mother plant. Most plant breeders make their crosses in whatever direction is the most convenient (usually, the direction that makes it easiest to recognize hybrids). I think this is a mistake. I am guessing that ability to grow in cold or hot weather are going to be partly genetically determined by cytoplasmic genes. So I usually do the initial crosses in the direction that preserves the cytoplasm of the cold resistant or drought resistant, or disease resistant variety, or more vigorous variety. I can make good arguments as to why genes in the chloroplasts and mitochondria might sometimes be involved in those characteristics, and they are known to be involved in some cases of disease resistance. I think the role of cytoplasmic genes is going to turn out to be very much more important than is currently realized.
10) If one cytoplasm is unusual and another is overused, I might deliberately retain the cytoplasm of the underused variety so as to increase the diversity of cytoplasms in the agricultural environment. So, if I were crossing a widely grown hybrid and a (rare) op, I might keep the cytoplasm of the op.
The hybrid mustard might have been created by cytoplasmic male sterility, in which
case it may not make seeds. Look at the flowers. If they don't have anthers, then they
are male sterile.