Psion
Adventurer
3) It may appear that everyone wants a perfected edition, and probably they do. But each of these people has a different vision of how to perfect the game. Each will complain loudly when your game diverges. It is best to keep strictly to the rules you're cloning, with the only deviations being for legal reasons.
That depends on your goal, really. What are you trying to accomplish here? What you'd get here is something that isn't really any better than the books already on my shelves. So unless you have some alternate goal, like being a reference point for creating reference materials (like OSRIC), then that doesn't buy you much.
If the goals at the heart of this product is to "update the 2e gaming experience" and/or "let me use my 2e stuff as-is", then a strict approach really isn't necessary or desirable.
If those aren't goals... well, I, at least, don't see any benefit in the project.
It's true that interested parties might want different perfections (the Pathfinder RPG forums should serve as an illustrative example), I think that it should be possible to come up with a list of most common bugbears and axe them, giving you a slimmed down baseline you can then supplement.
If I can pick on the post before yours as an illustration:
Of course, here are my ideas for 2nd Edition:
Bring back the half-orc and monk.
Wasn't in the core and wasn't in most published materials, so if your ultimate goal is plug in playability, this is stuff that can wait for a supplement.
Make the ability bonus tables make sense (either 3e or BECMI).
That's a good streamlining.
Get rid of level limits.
It's very easy to ignore them if they are there, and at the same time, replicating the table is a copyright sticky-wicket. I'd come up with an alternate optional method for those who want to replicate the experience. (An example might be a list of racial aptitudes that translate in another table.)
Allow the character a small number of S&P points for character customization.
WAY optional, not representative of the "core 2e experience", and a lot of work to boot.
Make all the skills percentile like the thief (ala Buck Rogers XXVc).
Possible, but making all skills/proficiencies D20 is a shorter leap from an OGL product.
Ascending AC and BAB.
Yes. This is a method simplification, not a truly substantial change.
Make the saving throws make sense (probably one per ability score like Castles and Crusades).
Penalize multiclass characters significantly (at least x2 or x3)
Not have the xp tables level off (maybe not a unified 3e table, but make each table advance like that).
All of these sound like major departures of intent. Simplify, streamline, but no alteration please.
Get rid of exceptional strength, give warriors a 1d4 strength bonus, allow ability bonuses with advancement.
Something like that, yes.
So, in short, I think what Philotomy and Mythmere are points well taken, but they advocate more "truth to the original" than I really think buys you anything.