By using this site, you agree to our Privacy Policy and our Terms of Use. Close
maenthoven said:
Volterra_90 said:

What I was wondering about OC method is the sentence that says: "38% of reviewers recommend this game" or something like that. This percentage is based on the actual score (<75 is considered a "not recommended" review, for example) or is it based on actually reading the reviews and evaluating if the review is negative or positive?

Yeah, we're still testing this one. One very, very important thing to understand is that "recommended" is very hard to generalize. It's hard to set even a philosophical bar. Should it be "recommended for fans of the series"? "Fans of the genre"? "All Gamers"? "Even non-gamers"?

For us, "recommended" means that the critic would unconditionally recommend the game to a general gamer

To us, that's intentionally meant to be a very high standard that's difficult to achieve. And it's important to understand this philosophy. "Recommended" can mean different things to different people. If a friend asks you about a product, your "recommendation" is almost always personalized to that friend. With critic reviews, recommendations aren't personalized, and so we have to speak in general terms.

We have it set to automatically parse it at 8/10 and higher. We also include non-scoring publications that issue a non-numeric verdict. Eurogamer, for example, rates games as "Avoid, No Verdict, Recommended, Essential," the last of two count as recommended. AngryCentaurGaming uses "Buy, Wait for Sale, Rent, Never Touch," and only the first is considered recommended. In these cases, we worked with the publications to draw the lines. Publications that do not issue a verdict are still excluded (AV Club, Totalbiscuit, Rock Paper Shotgun).

Having that said, we also have made some adjustments. It might sound ridiculous, but someone at OpenCritic has sat down and read the entirety of every single review. If we feel like there's an error (a 7.8 that sounds like a glowing recommendation), we can adjust it separately from the score.

One benefit of the system now is that it produces a completely uniform distribution. There are roughly as many games from 5-10% recommended as there are from 55-60% and 75-80%. Contrast that with average scores, where the move from 73-->75 jumps over way more games than the move from 54-->56, even though it's a 2 point difference in both cases.

Okey, so it's a harsher metric than I previously thought! If that's the case, I understand that the percentage for SF: Zero is low. A lot of reviews have pointed that they recommend the game to SF fans, but they were not sure about the general gamer. Also, lots of them are below 80. I'll remember that when I see another OC page about a different game then. If you're talking about unconditional recommendation, it seems about right. I thought it was more like "it's alright, give it a try" recommendation. Also, kudos to the guy who read all the reviews haha. I love to read reviews, but I'd find tiresome to read all the SF:Zero reviews! :P