Compilation of VNDB Ratings

[SilverLi]

 

 

Awesome text from Steve

 

Really good points.

 

But I still like the effort you are doing Warudusu. It's a huge job to do. 

[Warudusu]

One last thing, using the vndb Bayesian rating (the number you are posting) is very misleading, especially for non-translated and new titles.

Take example Irotoridori no Sekai (which you are by the way missing in the list), has an average vote of 8.44 and Bayesian to 8.10 (since only 159 votes on VNDB, compared to like 1000 on EGS). And then take its sequel Irotoridori no Hikari (I wouldn't really call it a fandisc, its literally a full length sequel almost as long as the original game) which is rated on vndb as 8.55 average but Bayesian pushed to 7.76 (well obviously if the original game has only 159 votes, the sequel will have even less since people have to finish the original in the first place). So according to this VNDB rating the sequel is worse than the original game, which is actually not how it is viewed by most people (EGS is like 81 for original vs 86 for sequel with like 1000 and 500 votes respectively). The second game is viewed as the better of the 2 and rightfully so.

 

This vndb rating system puts too much emphasis on the number of votes which for stuff like this causes inaccurate information about how people voted.

This will also never give a chance to newly released games, since they can never achieve so many votes as something like FSN.

 

So I would recommend switching to the average rating if you are going to do this, it won't change much for the top positions, but it will allow newer games to be in the top, where they belong. (and lets be honest if you're gonna be picking between outdated game released in 2001 or something and a game with awesome new effects and graphics and HD resolution, both of which are about equally rated as far as the writing goes - you will probably want to pick the newer one, where as vndb advises you to pick the old one).

I did include the vote count for those who are brave enough to try and stabilize it, ..But realistically I doubt more than 50 people are going to see this list anyway.

The reason I didn't use average instead of bayesian (I did previously, but ugh) is because the bayesian ratings are next to the names instead of having to open each entry in a new tab, It's not impossible to do, but laziness is what mankind specialises in.

By the way, I'm going to upload something I made a few days ago but decided to scrap.

It provides a look into how the two systems differ in ratings

Pretty sure it's very incomplete.

50+ Hours

30-50 Hours

10-30 Hours

Averages VNDB, Median EGS (Since that is what fuwanovel lists)

Except in the cases where even EGS won't vote on something.

While the average VNDB does reflect the actual score better, I can't in good conscious list something as 8.00 with 4 votes, even in those cases EGS may or may not have votes on it either.

Do you think the EGS  median is more fair or should I also just use the egs average?

I can also favor one over the other, the images are -20%A + 20%B, though maybe a "fair" weighted score is the best option (First column)

Irotori Data

Sekai

VNDB: Bayesian 8.10, Vote Count 159, Average 8.44

EGS:   Median    8.10, Vote Count 843, Average 8.00

Hitori

VNDB: Bayesian 7.76, Vote Count 57, Average 8.55

EGS:   Median    8.40, Vote Count 49, Average 7.90

In Hitori, you can see that the two are almost opposites, If you followed averages for EGS, you would still come to the conclusion that the sequel was worse, which I'm going to take your word for it, is better.

Irotori Rating Data

Sekai:VNDB Bayesian + EGS Median = 8.10 (Same)

Sekai:VNDB Average  + EGS Average = 8.06

Sekai:VNDB Average + EGS Median   =  8.40

Hitori: VNDB Bayesian + EGS Median = 8.05

Hitori: VNDB Average  + EGS Average = 8.24

Hitori  VNDB Average  + EGS Median  = 8.48

Oh, and lastly, the Irotori is at the bottom of the image because as far as I know, the website is down and thus the status is unknown

 

 

 

[Steve]

Yeah median on EGS is fine, and a good filter is using like 50 minimum data, even newer games (worth playing) should get 50 votes within first 2 months so you can see those and it filters those doujins and stuff where the dev votes 10 on the day of the release xD

So filter like this: http://erogamescape.dyndns.org/~ap2/ero/toukei_kaiseki/attlist.php?median=&count=50&from=1900-01-01&to=2100-12-31&game[pc]=01&submit=on#kekka

PC-only also filters some of the re-releases but not all, lMLA has like 4 versions there for all ages and stuff like that.

 

BTW the EGS scores on Fuwa are fairly outdated, they are from the time the game was added on fuwa and don't update automatically, so the number of votes changed (the votes usually stay similar though)

 

(Also on the EGS filter I would also check the eroge only [not a fan of all-ages], wakan [not a fan of rape either] and probably even no doujin [not a fan of inferior graphics xD] so my filter would look like THIS + then I would select like 1 heroine tag [loli or imouto for example], you need to be careful with clicking too many tags since those are not added for every game and depend on users adding them, unlike the categories like wakan/eroge/doujin which are added when the game is added to EGS, but usually stuff like loli is there [usually heroine types are] so I would select that loli tag And lastly I would also filter the games to only list those released since like 2010 as that is when the big change to the new flashy artstyle I love so much happened - so in the end my filter for VNs looks like THIS, well turns out I'm just now reading the #1 - I then check tags on VNDB [usually better organized] if it for example has only good endings and its not too utsugish, since I don't like bad ends, if the loli is the main heroine and she is cute... and that practically leaves me with FAVORITE titles xD well gg, seems like ill be reading a VN every like 1.5 years as that is the average speed FAVORITE releases its titles xD well who knows, maybe one day a new company making true route happy ending quality loli VNs with awesome artstyle will appear xD - this actually shows you how few good VNs fitting your exact taste there actually are )

 

 

 

 

 

 

Hitori
VNDB: Bayesian 7.76, Vote Count 57, Average 8.55
EGS:   Median    8.40, Vote Count 49, Average 7.90

In Hitori, you can see that the two are almost opposites, If you followed averages for EGS, you would still come to the conclusion that the sequel was worse, which I'm going to take your word for it, is better.



Oh, and lastly, the Irotori is at the bottom of the image because as far as I know, the website is down and thus the status is unknown

 
Not sure where you are taking data for Hikari from, but they are wrong - http://erogamescape.dyndns.org/~ap2/ero/toukei_kaiseki/game.php?game=16919
maybe you were looking up some wrong game, its not Hitori but Hikari (ヒカリ)
 
as you can see
中央値 (median) 86
平均値 (average) 84
データ数 (votes) 426

[Warudusu]

One of the biggest problems to overcome is inconsistent naming, you can give two english titles and a japanese title to every game, but only one is the most recognized.

I'll do the 50 sort like you suggested for now and make a separate table for entries with less than 50 entries, kind of like a please play these and rate them table..something of that nature.

I can make no promises about how long this will take.
 

[SilverLi]

Just do what you feel is the correct thing to do. It is in the end your work. We are giving feedback and suggestions.

[Warudusu]

Here's a preview of EGS median

94 Yu-No, The Girl That Chants Love
93 Muv-Luv Alternative
90 Stein;s Gate
90 Clannad
90 Fate/Stay Night
90 Sengoku Rance (Rance 7)  
89 Sharin No Kuni, Himawari No Shoujo
87 Monster Girl Quest 3
86 G-Senjou No Maou
85 Little Busters
85 Grisaia no Kaijitsu 
85 Tsukihime
85 Kamidori Alchemy Meister
82 Saya No Uta (Song Of Saya)
80 Rewrite
80 Uminneko No Naku Koro NI Episodes 1-8
78 Umineko No Naku Koro Ni Episodes 1-4

I linked them to their VNDD pages, but I assure you I got the scores off EGS.

[Steve]

Then again I must ask, what is the point of this anyways? xD

Why writing the numbers down when you can just link people to the list on those websites where those numbers are updated in real time and you can add your own filters too, like English translated only, loli only etc.

Seems to me like lot of work for no real benefit and without a script you cant possibly keep it updated.

 

If anything you could make a script using the VNDB api or DB dumps to get the values ( http://vndb.org/d11 ) since writing everything down manually is really not the way to go

 

But I guess if you're having fun doing it its fine

[Warudusu]

You know, I thought about it and I came to the conclusion that there can only be one true purpose to this list, for those who can't read japanese but still want EGS scores, you cannot easily determine the true vndb score anyway, so by combining EGS scores with VNDB links we have the most important part of both sites.