EdTrice: Sliding scale: In Smirf average exchange values are constant. But the positional influence will vary automatically, thus I have not to rebalance those values. When a piece is captured, the evaluation is not only reduced by the exchange value, it will be positionally recalculated.

Pawn Races: Smirf has included a simple estimation of passed pawn values - still not perfect.

Castling awareness: Castling is not individually analysed, it will be selected automatically by positional considerations only, independent e.g. from castling moves.

trying to teach Vortex: I am very distant to such approaches. Because it is nearly impossible for a chess master to describe what exactly he is doing during his considerations. I think that something like this might lead into a conceptual chaos. But why not give it a try? I am on the opposite way, trying to 'feel' the needs of a 'frustrated' and 'madly' working machine, to reduce the needs of its job (danger, could lead to paranoia).