Sorry, I was confused by the various ways this information is being presented. The Original Spell grants +5 AC as a reaction, the boxed text for you version says it still grants +5 AC but you say afterwards that it does not. The whole point of the spell, with regard to bounded accuracy, is to increase the odd of negating hits for casters. Negating a hit is important for multiple effects, so why would you want to throw off bounded accuracy and make it easier to hit casters?
I think I am just fundamentally not understanding what problem you are trying to address, and how you are trying to address it.
A few issues.
So shield spell at high levels is a ridiculously efficient 1st level spell slot.
Enemy number of attacks and damage scale, but the cost of negating them does not.
On a character building for high AC AC, having access to shield spell and even a modest number of low level slots is ridiculously strong. Nothing but a critical hit can penetrate it.
The shield spell basically
breaks bounded accuracy when stacked with other high-AC build choices.
Keeping the shield spell up against every attack costs you 1 spell slot (of any level) per round.
Dipping a class that grants shield (hexblade, wizard, sorcerer, some artificers) becomes part of "I cannot be hit" builds, and arguing that shield is always up is very plausible.
At low levels, meanwhile, shield is a decent spell, if expensive. Staying alive is often worth burning 1/2 of your daily spells at level 1.
So I think, at high levels, burning a 1st level slot and a reaction for +5 to AC against all attacks is overly strong.
---
I have seen DMs, when confronted with a moderate-to-high AC character who has shield, ending up resorting to ATK-inflation. Monsters start having higher ATK modifiers, just so that they can land a blow ever. This results in an AC arms race in the party, as anyone who doesn't have insane AC becomes auto-hit just so the DM can sometimes land blows on the insane-AC-plus-shield character.
So fixing shield so it is less "always on" has value (or at least making it cost more!)
---
The next issue is that builds who dip hexblade get a better deal than hexblades at using shield. Shield doesn't care what slot your spell is at all. A hexblade 2 is just as good at using the shield spell as a hexblade 10; and a caster MC hexblade who stops at level 1 is probably smart.
---
Adding these up, making a shield spell that is about as good for low level PCs, but at high levels you'll have to scale the spell slot used, could help.
Burning a 1st level slot on shield is still strong -- you get to negate the attack, and the next attack is more likely to miss, and even if it does hit, you take half damage. But it isn't the virtual immunity (on high AC builds) that the prior shield spell grants.
---
The "resistance to damage from attacks" is a cherry. You can cast shield even if it won't help (like against a critical hit) you get missed, but you halve the damage. This is a life saver compared to the default shield, and (I hope) helps make the change feel less nerfy.
For low-HP low-AC casters (for whom the spell is reasonable), this probably makes the spell better. They are more likely to experience "even a shield cast won't help you".
For shield-casting tanks, this makes it less "I am immune to attacks" and more "I can negate a select number of attacks". And can use it to blunt critical hits, which has value.
---
Suppose you have an enemy that needs a 15+ to hit the "tank" with shield, and a 10+ to hit a normal caster.
Normal shield makes the "tank" get hit only on a 20, and the caster on a 15+. They both have a 1/4 chance per attack for the shield spell to be useful (hit, but by less than 5).
If crits are 2x damage (close enough), then old-shield reduced damage to the tank by 71% (from 5/20 hits 1/20 crits, to 1/20 crits, is 5/7th of incoming damage negated), and to the caster by 42% (from 10/20 hits 1/20 crits to 5/20 hits 1/20 crits is 5/12 damage negated). Both burn about 0.7 slots/round against 4 attacks/round, but the tank gets a much much better ROI.
New-shield, using a slot high enough to defend against all attacks remaining, reduce the damage on the tank by 86% and the caster by 73%. But instead of 1st level slots, you are sometimes burning 2nd, 3rd, 4th or even higher level slots (depending on number of expected attacks before your next turn).
(The caster using the higher level slots gets resistance against the follow-up hits; the tank only gets to mitigate crits with the new version of shield).
