I usually from the school that unknown shape/material bonuses must be discovered to be used (otherwise, there's no point to the virtue that lets you do that), but an anvil is such a crucial and iconic component of a forge that it (IMO) stands to reason that it must have already been done.
And that being said, your idea is good, IMO 
Well, I understand the "writing" specialty (knowing more words when writing than speaking is...daft), but there are language specialties, like dialects, giving you an effective skill of 5 in that language.
The most common latin specialty is thus "hermetic usage", which mean you get skill 5 for discussing matters of hermetic magic (you have the most vocabulary there) and 4 for the rest (like asking where's the bathroom)
Actually, I suffer from this in daily life, since there are entire spans of non-RPG-related vocabulary I am totally lacking 
The exact level of soak is debatable, but, IMO, it is quite needed in order to have creatures of might be anything if not canon fodder for players.
Please, read my simple exemple above. Read the numerous threads on the forums. Ask the question. Do as many simulations as you like. But don't let them be as vulnerable as they currently are. It totally breaks my suspension of disbelief.
OTOH, I'm not fan of considering exploding dices as a factor, otherwise you could say that your lvl 50 CrIg spell is useless because some peasant will roll and exploding dice on his soak and take absolutely no wound. It could happen just as easily.
You could also say that, on round one, the demon will roll an exploding die and kill the might-stripper magus. Or that the magus will botch on his spell.
I could also argue that a creature may flee, heal his wounds, and strike later. A creature that had its might score drained is permanently weakened. The equivalent for a magus would be, for exemple, a permanent lowering of his art score. This is heinous, and a clear advantage of might-strippers over damage spells. Not that this is wrong, but just a reminder since you seem to say that, all other things being equal, damage spells are better than Might Strippers.
It's another possibility. Its problem is that emptying the might pool stops the creature from using most of its powers, which is quite weakening.
If at round 2 your demon is forced to rely on physical attacks because his might pool is 0, it won't do him any good if he's the "magic, not sword" type.
Correction:
Round 2: No powers, save those rare powers that cost 0 might
Round 10/18: Dead (you forgot to count the 2 rounds spent draining the might pool)
IMO, the "dead" time in 8 or 16 rounds is right for an "on the fly" weak spell (lvl 10!) cast without any preparation at such a (supposedly dangerous) opponent. It's the "deplete might pool in 2 rounds" bit that cause me problems. This totally cripples all but the most physical adversaries.
I'm also bothered by the fact that you'll need powerful physical spells, which mean you'll require planning, searching for arcane connections... but that the weaker might strippers will always be 100% efficient, even against an Elder Dragon, which means you don't need planning and all
Now that I think of it, I'm not fan of divisions by 4 or 8, this is hard enough to slow down things.
=> We could combine both, with a soak of, say, 2*(Might/5), with score being diminished by 1/2 points.
For that demon, this still means immunity to your lvl 10 DEO.
A lvl 20, like the Pilum, cast twice per round, will drain 08 might, meaning:
Round 5: Bye bye powers (assuming he uses nothing)
Round 15: Hello death.
Wrong.
Attack advantage doesn't add to aimed spells damage, the roll is just for hitting. This was confirmed by David Chart: