But as someone who has written for pretty diverse types of sites (from an over-specialized over-opinionated fansite like NMA to a general RPG specialty site like GameBanshee to a mainstream generalist site like GamerNode), I've learned it's not all that simple. There's two factors in particular that stick out like a thumb:
1. How much do you write and what do you focus on when writing? Internet reviews are theoretically limitless, but not even the Watch and GameBanshee audiences would read a 40-page review of Dragon Age or Fallout 3. I feel more comfortable writing a 4-page review for GB than for GN. And given the text-limit reaches no more than 1000 words for GN, I have to pick and choose what to discuss and how much. So how do I choose? I have to guess what my audience is interested in. I could focus purely on what I care about but I'm not writing for myself.
What makes them extra painful is that I'm well aware that easily half my readers read the intro, conclusion and grade, if that, and many only glance at the grade. Obviously, the most important thing in giving out a mark is how much I like a game. But it's also a form of recommendation. If I'm reviewing an Avernum game, it'll usually range in the 7s for GB. I couldn't possibly recommend it to GN readers, so I might have to give it something into the 5s if I'd ever review it there. Which is exactly why I don't.
Honestly, if this is the case there is a problem with your approach. What kind of method would you use to find out what kind of score would be best for me, as a potential GN-reader? You don't know anything about me, except that I might or might not be a regular reader of GamerNode. This says absolutely nothing about me and my tastes. Even if you have (which I assume) access to data about your core readers (or those who bothered to answer the last reader survey), you don't know the first thing about them as individuals. Especially not the silent majority, who never post comments.
So really, it's pure guesswork from your side. You're making assumptions about me as a reader and what I would like, and in reality you have nothing tangible to base these assumptions on. Again, you're not a mind-reader, and you don't know me. No matter what you might think. All your readers are different individuals, and they deserve to be treated as such.
So if you're providing me with a score that's not based on you own opinions, but what you imagine I would like or not, you're doing both me as a reader and the game you're reviewing a huge disservice. I expect to be told what you think about a game, and if you're not telling me that, the review is useless to me. And you're throwing away any credibility you might have as a reviewer, because as a reader I expect you to judge the game for what it is. I expect professionalism and honesty, not futile attempts to please an imagined audience.
The way we do it (I have worked full time on a mainstream site for seven-eight years now) is that we use 1000-2000 words for the review, depending on the situation. Our Risen review was around 2000 words (I didn't write it). Now, we don't expect everyone to read all that. That's why we have pretty meaty conclusions at the end, and give out a score in the first place. The words are for the people who are actually interested (or the people who see the score, and then become interested).
We assume that a large number of the people who are interested know the genre. Our readers are all kinds of people, and fans of all kinds of genres, but they all expect our reviewers to know what they're talking about, whether they're reviewing an RPG or a 2D fighter (seriously… you think certain RPG-communities are angry? try annoying fans of Japanese 2D-fighters with a casual "this might not be for the masses, hence I'm playing it safe and giving it a 6"-review of a hardcore fighter).
So when you write a review of an RPG-game, the readers who matters are the ones who would be interested in buying one in the first place. Scoring a game lower because it's a hardcore RPG and you don't expect the imagined "mainstream gamer" in your head would like that, is like scoring a soccer managing game lower because you don't expect fans of FIFA would like a game where they don't actually get to play soccer themselves (IGN did this. The readers were not pleased, and IGN eventually had to withdraw their review).
So you need to give the Geneforge games the score they deserve, as hardcore RPGs. Because that's what they are, no matter if you're reviewing them on a RPG site or a mainstream site. If you review them on a mainstream site, you need to explain a good few things in the text, because you have to "sell it" (if you think it's good) to a wider audience. So you need to focus on different things than you would were you reviewing it for a hardcore RPG-site. But that's different from changing the score to suit the readers.