I don't follow you. My point is that I don't think the old player averages were wrong. The numbers seemed to bear out that if someone gave the PS2 version of "Chaos Theory" an 8.4 and gave the XBOX version an 8.6, those two ratings would be aggregated accurately into the averages for PS2 and XBOX respectively.

Now, in whatever manner that 8.4 and 8.6 had originally been assigned to PS2 and XBOX respectively, the site redesign erased it. I never said that Gamespot was now manually assigning the ratings - I believe when they say they're using some kind of algorithm to assign the ratings among the versions. But that algorithm, by necessity, involves guesswork if the original version assignments have been lost. And if that guesswork assigns 90% of the ratings to one version, there's something wrong. When that 90% goes to a fringe version like N-Gage, there's DEFINITELY something wrong.

@leon2365:

Shrugging. I don't know, and neither does Gamespot. What I've been told is that even before the site redesign, when a user gave a ratings-only review (i.e. didn't write anything), the rating didn't necessarily get assigned to the version of the game intended by the user. So when GS was tinkering with user review numbers, they developed a new algorithm that supposedly more accurately assigned ratings to versions.

My problem with this version of events is that I still have access to a lot of the numbers from before the redesign, and they look accurate. "Chaos Theory," for example, had about 8.740 ratings for Xbox, about 5,190 ratings for PC, and about 2,640 ratings for PS2. Obviously, the new algorithm reassigned most of these ratings to the N-Gage version, which can't possibly be right. I still maintain that what really happened is that the ratings were accurate pre-redesign, but the version-specific tags on a rating got somehow lost in the redesign. So now Gamespot applies some sort of guesswork in assigning a huge percentage of the ratings to one version or another. Look up any multi-platform game, and you'll see one version that has the majority of the ratings allocated to it, and often, as is the case here, it makes no sense at all.

... but that doesn't mean I'm going to stop talking about it.

"Tom Clancy's Splinter Cell 3D" has 18,949 player reviews.

It turns out that that's an aggregation of all of the various versions of "Tom Clancy's Splinter Cell Chaos Theory." Don't ask me why the Chaos Theory part is left out of the homepage name for this game, or why all the versions are lumped as "3D." That's not even the fun part.

Here's the breakdown of the number of player reviews for each version:

• • PC - 1,649
  • XBOX - 3,682
  • PS2 - 931
  • GC - 254
  • DS - 396
  • 3DS - 34
  • N-Gage - 12,003

No, seriously. I've been told that the Gamespot player review numbers for multi-platform games are more accurate than before the site redesign. Yet I'm supposed to believe that the freaking N-Gage version of this game got almost twice as many people rating it as did all of the other versions of the game combined. Were there even 12,000 Chaos Theory games made for the N-Gage?!?!?!?

Can someone from Gamespot please tell me what utility it is to have obviously inaccurate information about player reviews for multi-platform games? I mean, you might as well be just making up the numbers if you're going to take player ratings that were obviously entered for one version of the game and randomly assign them to another version.

It's so frustrating. I once really relied on these scores, but now they're laughable.

Not sure anyone's paying attention any more, but you can't see reviews for a ton of games - when you try, you get directed to the homepage for the game's blackberry version. "Tom Clancy's Splinter Cell Double Agent" is the most recent example I've found - if this can be fixed, maybe there could be a universal fix.

And it still annoys me that homepages for a ton of expansions have disappeared, or have been subsumed into the original game - "Mass Effect 2: Overlord", "Borderlands 2: Tiny Tina's Assault", etc.

It also doesn't help matters that, as I've said before, for a lot of games, the user ratings for expansions are still being lumped in with the original game... and for some games (like Bionic Commando Rearmed, for example), you can't see any reviews at all - if you try, you get shunted to the page for the Blackberry version, which has no reviews. Still very, very buggy, that.

I've looked at a lot of data I had taken earlier last year from the previous Gamespot site, and compared it to the new data. Here's what I think happened to the user ratings and Average Player Score.

The old site worked fine. If you wanted to give a user rating (rather than a full written user review), you'd just go to the page for the game you played, you'd go to the version of the game you played (PS2, XBOX, whatever), and you'd give a rating. That rating would be added to all the other ratings for that version of that game, and Gamespot spat out an Average Player Score.

When the site was redesigned, for whatever reason, there were no longer separate pages for each version of a game. You could no longer see what the Average Player Score was for any given version - all you could get was the overall Average Player Score across all the versions.

When Edgework and others from Gamespot saw this, they fixed it by separating out the versions. The problem is that the new site still only has one page where you can give a user rating for a game, and that user rating isn't version-specific. So Gamespot uses some kind of algorithm that compares the sales of the various versions of the game, and assumes that you are rating the version that sold best in the given time period. You think the PC version of Batman: Arkham Asylum is bad, so you give it a low rating vs. the PS3 version? Nope. You just rated whichever version of the game sold highest.

Either because the old data was lost, or for some internal policy reasons, Gamespot then applied this algorithm to all of the old user ratings and old Average Player Score. So if, for example, a game came out on Gamecube, Xbox, and Playstation 2, because Gamespot no longer knows which user ratings go with which version, the algorithm automatically assigns most of the ratings to the highest selling version of the game. That's why you can look at any older game and you'll see 90% of the user ratings under one version of another. Go ahead - I challenge anyone to find a multi-platform game where the number of user ratings assigned across the various versions makes any kind of logical sense.

It's frustrating, because the Average Player Scores can't be remotely accurate any more - and regardless of what I've been told by Gamespot, the data indicates that the numbers made sense before the redesign.

I remain of the opinion that I'd rather Gamespot just start the ratings-only user reviews over again from scratch than give us admittedly inaccurate information. It sucks that Gamespot had the largest database of user ratings of any site on the internet, but at least with IGN and Gamefaqs, the user ratings appear to be accurate.

I've been vocal in my criticism of how user ratings were changed since the Gamespot redesign. At first, all user ratings were lumped together across platforms for any given game. Gamespot fixed this, and now you can find each version of a game with its own user ratings and Average User Score.

(Whether those user ratings and Average User Scores are more or less accurate than prior to the redesign is open to debate. I've again been vocal in saying that the number of user ratings for each version made sense before the redesign, and now they don't - often, 90% of the user ratings get allocated to one version now - but edgework from Gamespot has already told me his position, and it is what it is - I know he worked hard on this issue, and even if I disagree with his conclusions, I don't need to repeat that debate here).

Now, I'm finding that for many games that had expansions, the exact same problem exists - the user ratings are lumped in across the various expansions. If you look up the Average User Rating for Dragon Age: Origins on PC, for example, you're not getting just the user ratings for that game. All the user ratings for that game's various expansions are included in the average. This can have a profound effect on the overall user score when a game's expansions don't live up to the quality of the original. It also messes with the data for the original game - often, I'm finding that the release dates are based on the expansions, or that you can't even see reviews for the original or an expansion.

So far, I've found this problem in Borderlands, Dragon Age: Origins, Galactic Civilizations II, Lord of the Rings, Battle for Middle Earth II, Lord of the Rings Online: Shadows of Angmar, Mass Effect 2, Sims 3, Space Rangers 2, Spellforce 2, and Warhammer 40,000: Dawn of War, but I'm probably just scratching the surface. I started reporting this on the game-specific bug reporting topic in this forum, but this seems like a systemic problem that merits its own topic.

I think there's a problem across many, many games where the reviews, game information, etc. is being mingled between the game itself and an expansion.

Just a short list so far:

Borderlands

Dragon Age: Origins

Galactic Civilizations II

Lord of the Rings, Battle for Middle Earth II

Lord of the Rings Online: Shadows of Angmar

Sims 3

Space Rangers 2

Spellforce 2

Warhammer 40,000: Dawn of War

I'll keep editing this list as I find more examples. For a lot of these, it's like the expansion takes over the entire page from the original, including reviews, release dates, etc. But generally, for each of these games, their pages are just all screwy and all user reviews are mixed together.

A final note - you mentioned GTA III. I don't know what the Mac user review number was before the reboot... all I've got is the PC and PS2 numbers - 7,590 and 16,575 user ratings, respectively, as of September of this year. So the PS2 version got something over twice as many numbers as the PC version. That sounds about right to me.

Now? PC 1,903, PS2 22,679. I mean, tell me honestly - which set of numbers feels more accurate to you?

Shrugging. For more recent games, there's not a whole lot of difference between versions, so whether the majority of ratings get arbitrarily allocated to one over another probably won't affect the average anyway. This is really only a problem for people like me, who are compulsively trying to find a way to judge older games against one-another. Gamespot was once my best resource for doing so - I can't say that it still is. Still, thanks, as always, for your input and your efforts.

I understand that you're having to do the best you can with what you have now. My problem is that I don't understand it when you say that the problems originated before the site reboot. My recollection was that you could make a ratings-only review for a specific platform. And the numbers made sense before the reboot.

I'll give you one example (though I promise I could give literally dozens more):

Batman: Arkham Asylum - here are the numbers from August of this year:

PC: 5,899 user ratings average 9.2

PS3: 7,329 user ratings average 9.3

X360: 7,994 user ratings average 9.3

Here are the numbers now:

PC: 1,878 user ratings average 9.1

PS3: 1,854 user ratings average 9.1

X360: 18,751 user ratings average 9.2

I admit, I haven't looked up the sales figures for Arkham Asylum, but I'm pretty sure they didn't sell ten times as many X360 games as PS3 games. And the pre-reboot numbers, which were approximately the same between the two systems, seem much more reasonable to me. And that's my problem - if you're doing the best you can now because information was lost, okay, it is what it is. But I just don't buy that the numbers were inaccurate pre-reboot. The numbers looked accurate before the pre-boot, and now they look arbitrary.