It seems providers scan the mangas themselves. As this one isslightly better: e.g. the hair is not pixelated and the dots on the dress are clear to see.
2 Likes
Like other have said, I found the levels appropriate when there are enough people who graded a given book. On the other hand, I am reading a series graded 40 that should be ~33, and there’s nothing I can do about it (my gradings brought it down from 45 to 40, but that’s as far as it goes). (To be fair, it’s only me + one other person who graded it, so it’s hardly “enough people”).
When you go with stuff with 0~1 ratings, you can find other funky stuff. Based on previews I read, 攻殻機動隊 1 | L42?? seems on par with 銃夢 Last Order | L24??, so, hum, someone must be wrong here. (I guess they both belong to the low 30s)
For the purpose of a book club, I’d say it can be a good base to evaluate the difficulty of something, while leaving some leeway to argue that something is under/over graded.
I haven’t read Chainsaw Man either, but I agree with @seanblue on the difficulty of 放浪息子. It’s more in the flow than in the actual vocabulary or grammar. Assuming that Chainsaw Man follows a typical Shonen manga plot line (again, I have no idea), that can make the difference.
5 Likes
I’ve pretty much stopped choosing my books based on natively levels, it’s just a bit too inconsistent for me. I think it would be nice if there were options to say why you thought a book was hard, but I’m not sure how that would even work with the current grading system.
I read 天久鷹央の推理カルテ which is Lvl Green 35 last year 2023 in April, and I didn’t think it was that hard. Really it was the medical vocab that slowed me down, but otherwise was fine. If I could somehow mention that the vocab contributes a lot to the difficulty level, that would be useful. But again, no idea how this would work, unless I wrote a review for all my gradings.
But I’m in a fortunate position now, having read 150+ LN/novels/Childrens Literature , I can just have a quick read of the sample on Amazon or Bookwalker etc, and decide if it’s too difficult for me.
6 Likes
I think the worst problems arise when someone who typically reads only low-level or only high-level books grades an outlier. They only have very specific levels to compare the book to, so even if it’s consistently higher or lower than all the books it’s being compared to, it will still not land far from them. When more readers get to grade such books, they can only really finetune the grading, not change the level drastically. I’m not sure if this gets corrected eventually as more and more readers grade the same book, but for now I think the initial grading tends to skew the results way too much.
11 Likes
Initially I was very skeptical about the grading process because of the way the grading algorithm works. We had a bit of a discussion about this over at WK, where I referenced an algorithm that I think would work better to produce a relative hierarchy of partially compared items. But I must say I am rather surprised how well it works in practice (for books that have enough gradings).
Having said that, I think there is still lots of wiggle room to improve. Like Naphthalene said, each new grading only corrects the book’s level to a certain extent. So if there is a big outlier in the grading early on, it takes a long time and quite a number of new gradings to correct this. Which makes me wonder what happens if a book was classified L30?? initially but turns out to be in the 45-50 range? How much influence does this initial classification actually have?
Which takes me to my next point: I think the selection process for which books a given book is graded against could be improved. I recently graded a book (世界から猫が消えたなら | L26) and every single book I got to compare it against was on a higher or same level at the time 
(and yes, I have read books that are lower than level 27 
)
I would expect the grading to jump around quite a bit, going higher and lower in alternation, and slowly converging, trying to approximate the book’s level as well as possible? Especially with books that have no gradings yet, and that are maybe initially classified in the wrong bracket, this would provide better results faster, is my impression
EDIT: I think I blanked a bit when I wrote that - what I wanted to say was that I would expect the levels of the books that I’m grading “my” book against to jump around quite a bit, going higher and lower than the current level of “my” book in alternation. What I would ideally want is that the algorithm picks book A to grade against, and if I say that my book is harder than A, then I’d expect a harder book to be chosen for the next comparison, and so on, this way allowing to approach the level range that the book fits in [with respect to the other books I read] to converge quickly.
12 Likes
very easy. much easier than mokuro. you just need the OCR manga reader which is a normal anroid app and your manga. done.
Personally, I take the levels as a guide not as set in stone. I am usually pleasantly surprised when I check my gradings to find them in line with other people’s gradings. But I also only grade against books I read recently.
If a book is level 34 but slice of life, I’ll probably find it easier than a level 30 but it’s sci-fi. Simply because I have plenty of exposure to one but not the other. Just like some people might have read a ton of medical novels and will rate books with medical vocab lower than someone who has never encountered that.
I do read reviews, though and also check what books the gradings used as comprison. If I have read any of them, it gives me an even better idea how accurate a level might be for me.
8 Likes
it’s not that broken - you’re spending money you otherwise wouldn’t have if they didn’t give you a 50% coinback offer! At least keep telling them that so they keep giving us coins.
3 Likes
I definitely agree. You’re right that it would take a long time to move up. However, I also think that sort of situation is exceedingly rare, so in practice it’s a pretty small issue.
I must admit too that I think it’s more resilient than I thought as well. The judge of the rating system is really how self consistent it is… and while you may disagree with a difficulty rating, if you look at the gradings below, it almost always makes sense how it got there ±2.
We could definitely get that ±2 down and we could certainly make it more flexible for books with fewer gradings and more stable for books with lots of gradings.
I’ll just say that I thought 黒 1 | L19 was quite a bit easier than Yotsuba and if you read it you’d probably see why. It’s literally textbook grammar, simple descriptive sentences with super easy vocab. Compared to dialogue heavy, slangy and with some adult speak in Yotsuba, there’s definitely a reasonable opinion on both sides.
You’re right though, it depends on what you’ve learned. I also wonder too if the non-kanji enjoyers get scared off and people self-select themselves into reading it 
Great to hear that people are finding the levels generally useful!
9 Likes
How come you think so? I very rarely put initial levels, so most of my submitted books start at L30?? but I’d expect a fair number of them to be 40-ish, even. Do other users put initial levels so often?
I think that might be a good option.
Actually I reread my above post and I think I went a bit brain-afk while writing the final section, thus I rewrote it:
In other words: It would be nice if we could get more spread in the gradings for books with fewer gradings, to allow books to move faster initially 
5 Likes
I feel like my gradings do this. But that might just be my perception.
1 Like
Oh sorry! I totally missed the ‘??’. No, that initial classification has no impact on the rating system as a whole. No points are lost or gained by the books it’s compared against. And the whole temporary rating system is it’s own algorithm. It can easily move a book from level 30?? to 45. If you keep saying it’s ‘harder’ it will keep prompting you until it can put some sort of bound on the level… either an ‘equal to’ or both a ‘harder’ and an ‘easier’. After it has bounds, then it takes another 5 gradings or so to make it ‘non-temporary’.
Now, obviously that system isn’t without flaws (what if you compare it to a lvl 38 book that you think is under leveled for instance…), but honestly it seems to work really quite well.
That’s reasonable. Although, that’s more about finding your personal true difficulty rating for the book. I think that’s an ideal approach, but with my simple elo model, I’m afraid that’d introduce a lot of volatility to the gradings (it gives each user more power to sway the ratings). If I had a better rating algorithm, like we’d chatted about in WK, I think that’d be a doable approach yeah.
At least that’s my current thinking 
9 Likes
Ah ok that’s good to know. 
Ah, I see. I was not expecting an “equal” grading to be sufficient for a bound. But if that’s the case with your algorithm, then I understand why I got these results with my recent grading.
Hmmm I actually just wanted to reiterate what you yourself said:
so now I’m a bit confused
But anyways, like I said, I’m surprised how well the grading actually works, although I sometimes find it a bit irritating to see a number of books slowly crawl across the levels at a speed of maybe half a level per month
4 Likes
I take this back. I just found a good OCR app this morning. Just didn’t look hard enough before I guess.
I agree. That is exactly what I ran into with 日本人の知らない日本語 S1 | L26. Granted, it did have a temporary rating and was corrected once I added my ratings, but the displayed rating was a fair ways off at the time because nearly everything it was rated against was L30+.
I’ve also noticed a similar problem with people who may consume easier content, but their true level is way above it. In that case, you get a crap shoot of gradings. Anything within 4-5 levels is all over the place. I noticed that with くまクマ熊ベアー S1 | L22 in particular. It was graded L19, but was definitely harder than similarly graded shows. I looked at some of the gradings and noticed that 1 or 2 of the graders simply graded it easier than everything in site and that was what pulled the level down.
Both of these come back down to the number of people grading the book/video. As they get more graders the levels tend to get corrected fairly well.
At least on video, I’ve noticed the exact opposite. Many times when I watch a show and feel like the level is off it is already corrected before I watch the required 4 episodes. The video section is a lot newer than the books section though, so that may be due to shows having fewer gradings on average and thus being easier to move.
1 Like
@Megumin もしもの世界ルーレット
You led me to to believe this would be around Magic Treehouse level. I’ve been wronged.
3 Likes
I stand by all those ratings after reviewing them.
Generally because they have very specific language like Isekai Nonbiri Nouka with the farming and Nagasarete Airantou with the spiritual stuff going in the later volumes.
Jashin-chan difficulty for the puns.
I didn’t feel this problem with the Roulette LN.
Of course as I mentioned earlier, if you choose to ignore the specific terms or the puns not hitting but just a basic understanding then maybe you could argue otherwise, at least IMHO
The problem is that I think those levels shouldn’t be that low, and I don’t think it was this low before
5 Likes
I haven’t read those series in particular, so I can’t speak for them myself. I definitely think it’s an ongoing issue with a lot of series though.
I’m not sure they should let us grade manga against novels or children’s books in the first place. I feel like the mediums are too different. For example, for whatever reason I’ve found manga like Inuyasha and Komi Can’t Communicate to be a lot easier than the time I’ve had with もしもの世界ルーレット.
I also found the first volume of Strange Catalog easier. I think the longer format of ルーレット has actually been detrimental for me. I’m reading slower because it’s taking me too long to get to the engaging parts in the story that make me want to read more. It’s been frustrating. 
I’m sure this whole thing is also an ADHD problem now that I think about it.
3 Likes
I can see the engagement being an important part, but I actually found the longer stories to be easier in this case.
Not changing completely the setting every chapter was a huge progress lift for me, as I had to spend less time understanding what was going on before focusing on the story.
I do agree I have mixed feelings about those being mixed.
Maybe having a global grading and a specific category grading would make more sense for comparisons, but it would make the system not as intuitive for first timers.
3 Likes
Care to share? I’ve tried a few and uninstalled and gave up looking for more.
I have used this a lot. Was easy to install. Easy to use.
3 Likes
I found this: https://play.google.com/store/apps/details?id=ca.fuwafuwa.kaku
It’s not a reader app, but it is able to do OCR over-top of any other app. It creates a resizable window on top of other apps that you can use for OCR. You can continue using whatever reading app you were using before.
I’ve only used it for a moment to check it out, but it seems to work pretty well. Might have many problems that I haven’t ran into in the 2 minutes I was looking at it before work.
2 Likes
