Saving Throw Derivation and System Assumptions in Old School Games

I’ll try to keep this brief. Saving Throws have been present in five of the six main iterations of D&D (Original/Basic, 1e, 2e, 3e, and 5e), and represent active avoidance of dangers, where AC represents a more passive defence. Really, of course, the two blend – AC represents dodging, and Save vs Spells/Will Saves represent mental endurance and resilience as well as active defence (thus Dwarven resistance).
 
An interesting insight into the mechanical assumptions behind different D&D editions is achieved via looking at how Save values are derived (or, indeed, in 4e, how non-AC defences are derived).
 
In OD&D, Basic, 1e, and 2e, there is one key defensive stat derived from an Ability Scores: AC. Raw AC is improved by a Dexterity-derived moodier. Having high Dexterity means you dodge better. You can then add on armour, shields, and so forth. The other defensive stats – the five Saving Throws – are not derived from Ability Scores, but by Class. (Save vs Spells is often altered by a Wisdom-derived modifier, but not always; in 2e, Save vs Death/Poison can be altered by a Constitution-derived modifier, but only vs Poison, and only with very high Con.)
 
You can very often get the “base” class you want, given the core Ability Score requirement for the four “core” classes which define Saves (Fighter, Wizard/M-U, Thief, Priest/Cleric) is 9 in the relevant Score. If there is any choice about arranging Scores, a player can basically guarantee entry into their choice of the core four.
 
Now, in practice, old school Save arrays are sometimes a bit abstract – why is this class better than that class at this slightly vague Save category? There’s a lot to like in the threefold Save system of 3^rd Edition (lese-majesté!). But the essential point I want to make is: early editions actually put a big choice in the player’s hands. The newer systems flatten that out (though free stat arrangement means that in 5e people focus on Wis and Con Saves rather than Str or Int).
 
This leads, I think, to two conclusions: Firstly (and this is fairly uncontentious), class differentiation is more important in old school systems than base statistical differentiation. Secondly, I think old school systems often *do* emphasize player customisation in a way that gets neglected in some of our discussions. I’ll probably write on this more soon, when I defend 2e Kits (gasp!).