Answers for Random Wizard

The original questions may be found here.

(1). Race (Elf, Dwarf, Halfling) as a class? Yes or no?
Most of the time, no.  I used to think absolutely not, but more recently as I considered monsters as classes (e.g. dragons, vampires, etc.) it made more sense to consider some races as their own class, when the inherent nature of the race was sufficiently "strong" as to warrant that kind of treatment.  I understand it's a fuzzy definition, and it depends heavily on a clear conception of what, exactly, race is.  These days, the more I think a race is relatively close to human, the less likely to see it as its own class.

(2). Do demi-humans have souls?
Depends on your campaign, and the deeper assumptions of the world you've created.  In planning for my next OD&D+ campaign, Aldwyr, I've decided that elves and dwarves have souls, but are so intertwined with their physical natures that if they die, they cannot be resurrected.  For dwarves, this is because a dwarf is too tied to the living rock of the world to not return to it when they die.  For elves, their intrinsic nature binds their soul up with their physical form, which is related to their longevity - but also means that when they die, their entire corpora goes through dissolution.  Halflings are closer to humans in having souls and may be resurrected.

That having been said, in my White Box game, The Hall of Forgotten Gods, all demi-humans lack souls as understood by humans.  Their gods relate to them differently than human gods do, resulting in different intrinsic natures.  As I mentioned before, it depends on the campaign and the assumptions you make.

(3). Ascending or descending armor class?
Descending.  It's how I was taught, and for me, it's easier.  I can see the relative simplicity of ascending armor class, but it's difficult for me to think that way.

(4). Demi-human level limits?
Again, it depends on the nature of the campaign you create.  Generally, I see them as a mechanical "fix" for dealing with the relative advantages of demi-humans vs. humans.  Rather than making a hard-and-fast limit, I tend to make it simply more difficult for demi-humans to advance - or give humans a bonus on earned experience.

(5). Should thief be a class?
Sure, so long as it is understood that the activities of the thief are things that any character can attempt - thieves simply do them better.  In some cases, thieves are capable of things that other classes simply can't replicate easily.

(6). Do characters get non-weapon skills?
It's been argued that the thief percentage chance increases are a kind of skill increase, and so are the camel's nose under the edge of the tent, making D&D into an endless quest for crunchy mechanical markers of success.  I tend to see non-weapon skills as the real camel's nose - what are they for, in game terms?  I'm not sure.  I would rather have players tell me something about their character that I can say yes to, and have THAT be the basis for their character than look at endless ranks of scores on a character sheet.

(7). Are magic-users more powerful than fighters (and, if yes, what level do they take the lead)?
This question hearkens back to the oft-made observation that magic-users start off weak, but become powerful as they gain levels.  Some people suggest that this shift is too powerful, as higher-level magic-users are often capable of dealing immense amounts of damage via spells.  I actually think that this is A Good Thing, since by the time magic-users get that powerful, a referee ought to have thought of really more challenging things for them to deal with than simply being a form of heavy artillery.

(8). Do you use alignment languages?
No.  In this, I am a student of Prof. M.A.R. Barker: languages are a reflection of the cultures and societies which use them, so an "alignment language" presupposes that alignments have direct, immediate, and on-going effects on social interaction.  Since "alignment" itself is a more cosmological element in the games I run than any sort of social divider, the entire notion of "alignment languages" seems a bit off, ontologically speaking.

(9). XP for gold, or XP for objectives (thieves disarming traps, etc...)?
Both.  The original D&D campaign I played in had a differential reward system for gold, monsters killed, spells cast, and other actions, depending on your character's class.  It made sense at the time, and I might use it again, because it rewarded characters for doing the things their class was supposed to be about.  More recently I have been less committed to a defined system like that, preferring instead to use gold as the main reward, with well-played objectives as "icing on the cake" - something more free-form and more amenable to rulings rather than rules.

(10). Which is the best edition; ODD, Holmes, Moldvay, Mentzer, Rules Cyclopedia, 1E ADD, 2E ADD, 3E ADD, 4E ADD, Next ?
The one that best allows me to build the game and campaign I want to run.  I tend to view all role-playing games as toolkits, so the more rules there are, the more likely they are going to get in my way.  I grew up playing Original D&D and then some of the others.  Original D&D's sprawling, inchaote nature lends itself to retooling and refining, which I have always enjoyed.  If I were to start all over, I suspect I would go with Holmes or possibly Moldvay, as they are more clearly edited without having ever-growing sets of assumptions attached to ever-lengthening books of rules.

Bonus Question: Unified XP level tables or individual XP level tables for each class?
Individual XP tables for different classes.  Keeps 'em all guessing, especially since I discourage meta-gaming involving explicit discussions of stats and levels in-game.