I still believe that it's a misdesign that we can assign walker to walker to turn. We copied this from L3, but L3 offers direct control over each lemming, not indirect control as in all other Lemmings games.
I'm not sure what is meant by "direct control" here... then again, I haven't played much L3.
The L2 Walker also reverses the direction of a currently-walking lemming, and since SuperLemmix has somewhat un-deliberately been aligned to L2 in a few of its features (Jumpers bounce off walls, Shimmier-Climbers turn and climb instead of dropping, we also now have the Ballooner skill, etc), the existing Walker skill fits in nicely here as well.Since walking is a "state" rather than a "skill", it seems perfectly acceptable that something should happen when the Walker
skill is assigned to an already-walking lem.
With that said, it is quite an overpowered and unpredictable action, making it difficult (but not impossible) to backroute-proof. So, I can see what you mean about it being somewhat questionable from a design point of view, sure.
If you add this to the shimmier, you'll proliferate this design. Where then to draw the line, for consistency?
To be honest, I'll most likely end up leaving things as they are. As much as I personally like the idea of same-to-same skill resulting in
some effect, it does become messy quite quickly. Currently, in almost all cases, same-to-same results in nothing happening at all, which is predictable and consistent and therefore desirable.
Ah, and: In L3, you can assign walker to swimmer, and the swimmer will turn. This costs a swimming tool; if he has none left, he'll drown.
Hmm... why would it cost a Swimmer if a Walker is assigned?

Interesting that L3 gives some way for Swimmers to turn, though. Worth further investigation before we write this topic off altogether, perhaps.