There appears to be a common assumption in Blue-Red Hackenbush that the Blue player should make whichever move will maximize the position value of the game, and the Red player should make whichever move will minimize the position value of the game, as though a greater position value corresponds to the relevant player winning by a "better" margin.
However, it stands to reason that assuming optimal play (recursive in this definition) for the remainder of the game, any of the Blue player's moves which leave the position value non-negative are equally optimal, and any of the Red player's moves which leave the position value non-positive are equally optimal, since they win deterministically, and conversely any losing move is equally optimal to any other.
How strong of a conclusion can we state regarding the strategic benefits of optimizing the position value of Blue-Red Hackenbush as opposed to hacking just any winnable edge? Are there any arguments for hacking another edge actually being better in some positions? How does this conclusion extend to the general assignment of surreal numbers to combinatorial games?
Citation needed! However, there is a reasonable sense in which it is often a good idea to focus on better-valued moves when thinking about a Hackenbush game.
Remember that the assignment of values to Hackenbush games comes from game addition: $G$ and $G'$ have the same value iff for every game $H$, the games $G+H$ and $G'+H$ have the same "basic type" (= win for blue, win for red, or win for second player).
With this in mind, we often think of a Hackenbush game $G$ as not existing in a vacuum but rather as part of a larger game $G+H$. Even if $G$ is a win for Blue (say), Blue moving in the $G$-part of the sum $G+H$ may be a losing move depending on what $H$ is; the set of good "$G$-part" moves for Blue in $G+H$ is however (even if empty) always closed under increased value for Blue.