I see what you mean, but it's not necessarily true. You could pick an initial direction and keep jumping that way. I don't know what kind of environment would favour this behaviour, or if the beans can "jump the same way as last time" but it's certainly plausible.
Agreed, but the bean would probably very quickly find an obstacle over which it cannot jump. It would have to change direction then, but that's a much more complex strategy.
Another problem is that the bean probably can't keep a direction even if it "wanted". It uses random walk, quite possibly because it's the simplest to achieve acceptable results.
So you're saying we should CRISPR in some pigeon magnetic direction finding genes to the jumping beans so that they have more choice in their shade-finding strategy?
Perhaps, but even in genetics, there is no free lunch - all that hardware you CRISPR into the larva inside the bean will need energy and material to be constructed and operated. Those resources will come at the expense of something else. It might damage another vital system, or just compete with everything else for fixed metabolic output.
Ultimately, it may turn out that natural selection will remove your "upgrades" simply because metabolic load makes the organism statistically less likely to survive long enough to procreate. Turns out - it's something I hear is particularly apparent in bacteria and viruses - natural selection likes simplicity too :).
Takeaway being, it's easy to splice in arbitrary genes these days - but it's hard to make it an actual improvement.
