I like implements (I'm a 4e player!) but would decouple them from the proficiency rules. Instead, set all saves as STAT vs STAT (like you said in your post, this was how at least some earlier packages did it) and choose an approriate base number (I'm attracted to 8+STAT, so a caster has to choose to target weaker stats to gain an edge). Then say that a caster without an appropriate implement reduces that base number by 2 (say from 8 to 6) or even 4 (say from 8 to 4).When spells DC and ST didn't scale, the only difference was in stat bonuses: the Wizard could still target the Barbarian with a Feeblemind, but the difference was determined only by the Barbarian's Wisdom score, which is limited to a certain amount (and occasionally, she might run into a wise barbarian).
<snip>
If I were in the design team, I would consider eliminating implements completely.
<snip>
If wanting to keep implements in the game, maybe they could grant proficiency bonus only to a certain school of spells, or only to DC which target one specific ST.
For a 20 stat caster, the DCs for my most brutal proposal are 13/9. For a character with no stat or other bonus, that is 40%/60% chance to save, which looks OK to me for the hors-de-combat effects. If they're worried that this would make damage spells too weak, just adjust the dice of damage upwards - that's easily done. We don't want to fetishise particular dice expressions for those spells and then set DCs so high that the save-or-suck spells become overpowered.
I had that in mind, because I am assuming that Next will be closer to AD&D and 3E, than to 4e, in its assumption about using player builds for building NPCs too.We also have to see it both from PC's and NPC's point of view, i.e. in this case it is both the Wizard PC not able to blight the Barbarian NPC, and the Barbarian PC being able not to be blighted by the Wizard NPC.