So, just to farther illustrate that as reviewers/staff members we all have different approaches and don't always agree, let me chime back in here.
To glassjawsh's comment - I'd say the number is BY FAR the thing I spend the most time agonizing over. Obviously in terms of time spent, no, since it takes longer to play the game and write about it, but my approach is typically to write the review with the score in mind, look back at what I wrote, and think about adjusting my scores. Part of the process of reviewing a game vs. simply playing it is you are FORCED, by nature of the process itself, to really examine aspects of the game that you might not otherwise examine or that you might simply shrug off as not important to the experience. But when you are forced to provide a score and you know that people might be making PURCHASING decisions, sometimes with limited resources - well I take that very seriously.
The biggest trouble I have with scores is that a score can only capture how you feel about a game at that specific moment in time, and that score gets captured by all kinds of aggregation engines (like Metacritic) and that becomes "your opinion" on the game. Obviously we appreciate our readers here and believe they are a cut above the average video game player and we try to write detailed reviews respecting that - but a large number of people will only ever interact with RPGFan through sites like Metacritic. And that can be frustrating.
What's also frustrating about a score is that quite often I look back and think, in hindsight, a game had more/less staying power than I thought it did and should have been adjusted up/down accordingly. For example, take my respective scores for Torchlight 2 (91%) and Ni no Kuni (95%), two games I just loved.
Now, I REALLY DID agonize over that 4 percent difference. In hindsight, Ni no Kuni for me was still an Editor's Choice caliber game, but Torchlight 2 is the game I've sunk far more hours into and still play on a regular basis. For me Torchlight 2 probably scores higher if I'm scoring on my personal scale!
But then, factor in that you're trying to predict whether the readership of the site IN GENERAL will like one game or the other more. I suspect that my burning love for Torchlight 2 might be a corner case and Ni no Kuni may have the broader appeal. I just don't know. You start to see right away how this task can cause a person to lose hair.
EDIT: Another thing about Ni no Kuni is I stand behind my comment in my review about this being as close to the Platonic form of the JRPG as we've gotten in many years - even if I am more likely to fire up Torchlight 2.
Now, on the topic of comments - we've debated quite a bit about this as well. I come down on the belief that we've got a really cool community on the site that resides in the forums, and if you want to talk about a game you can do it there and I kind of like it that way. We are, again, a volunteer staff, and even with our powers combined I don't think we'd be able to spend the energy necessary to keep comment sections on review pages constructive and civil - many sites have entire staff for that. HOWEVER, I do agree with Damacon that there are often some really good points that come out of comment sections on reviews at other sites. So I'm willing to be persuaded on this point. But I'd say the biggest reason we don't have comment sections has to do with the resources/time it would require to keep them reasonably under control, whereas we still get great conversation about the games we all like in the forums, facebook, etc anyway.
Hopefully this all explains a little bit behind my thought process anyway - I don't speak for the other staff members. :-)
EDIT: Here are links to the reviews if you'd like to see what I mean or provide critique on my scoring, but hopefully I've done at least a decent job illustrating a few of the troubles with scores.http://www.rpgfan.com/reviews/Ni_no_Kuni_Wrath_of_the_White_Witch/index2.htmlhttp://www.rpgfan.com/reviews/Torchlight_II/index.html