Introducing Natively - A free community book sharing site with over 9800 popular books graded by difficulty

…other people can search my gradings specifically? I’m inclined to think that’s a narrow enough use case to not worry about :sweat_smile:

Barring what I suggested above, perhaps a way to filter by type? There are only two results returned for short stories with 朝 but tons when searching all content.

1 Like

While that’s probably true haha, I still think if i’m going to spend time here I’ll fix the exact match issue (not terribly hard to do) rather than adding filters. It’d solve this problem and it’d solve other problems too.

If it is still a problem after I implement that solution, then we could think on additional filters :slight_smile:

At least, seems like the right way forward.

1 Like

Can you add a way to indicate how many times we’ve reread a book? Not the dates of those rereads or anything like that, just a count. Basically the same as what AniList has.

2 Likes

That seems reasonable. I’m not entirely sure if i’d be able to include the rereads into your stats, but at least allowing you to track seems fine. It’d also give me some notion of how common they are too.

4 Likes

Yeah, I figured that it wouldn’t be practical to include in the stats without dates attached to the rereads. Honestly I haven’t reread much in Japanese, but I happened to think of it since I just reread the first volume of Girls’ Last Tour.

2 Likes

PRODUCT UPDATE

As I previously hinted, this sprint was all about surfacing your reading statistics in intuitive way. I think Natively does stats quite well now… better than other reading trackers in my opinion! :slight_smile:

A New 'stats' Tab On Your Profile!


You can now find in-depth metrics on your reading, allowing you to toggle between books vs pages and to split across different time segments. I encourage you to play around!



If you keep track of your days read or log your reading time when you update your progress, you’ll find additional graphs.

Book Page Stats

On book pages, we now have a new section ‘My Stats’ which surfaces more reading statistics for a particular book, such as reading speed over the entire book.

Add / Edit Reading Sessions Manually

On your activities related to reading sessions, you can now edit the reading session… such as changing the date or the pages read. If you want to quickly find your reading sessions, remember that you can filter your feed to only ‘Reading Sessions’! Also, if you want activities for a particular book, you can always view that feed on that particular books activity feed.

You can now manually add reading sessions! You simply need to go to the book page, find the activity feed and click the ‘+’. Do note that you can only add from the book page, but you can edit reading sessions anywhere you see the activity.

Hidden activities

While you’ve always been able to remove certain activities from your feed, you’ll now see you simply can hide activities from others. As we’re beginning to use these activity items to surface editing capabilities (such as edit reading sessions), allowing you to remove the activity from the feed wouldn’t work.

Upcoming

For this sprint, I’ll be adding forums! I’m really not sure how long this will take, but hopefully it’s easy. Afterwards, I’ll get to movies.

I decided to add forums now as I think I’ll want to have dedicated discussion threads for upcoming, separate aspects of the site (movies, then Korean).

In addition, I’m hopeful that forums will be a quick win for growth. Multiple users have mentioned in passing to me that they like to interact on Natively forums and that since they’re not on WaniKani, they’d prefer Natively forums. We’ll see :slight_smile:

Any questions or concerns, of course let me know! Thanks!

Edit: @Phryne you can now add daily reading sessions retroactively! Let me know what you think.

11 Likes

I am so psyched! I’m going to have a go at adding my 2021 data and I’ll let you know how it goes :grin:

2 Likes

I know you’re probably still planning this out, but something I’d really like to ask for in this is a requirement that people rating indicate whether they watched with target-language subs or no subs so that difficulty ratings can (down the line) be sorted out by that. Just being able to evaluate the gradings with that knowledge even would be helpful IMO.

4 Likes

Absolutely. That’s one of the major things to take on and consider. I think it’s pretty straight forward (simply have two statuses, ‘watched w/ subs’ / ‘watched no subs’), but I’ll have to play around with it and make sure it’s intuitive and all makes sense.

I think I’ll have a status for if you’ve watched w Eng subs as well, as I’m sure many people will use that as a wish list of sorts. This status wouldn’t count the video as ‘watched’ of course, just a status.

But yes, I agree with you!

1 Like

Watched with native subs?

1 Like

Yes. As I said there, eng subs would be a different status, operating more as a wish list :upside_down_face:

3 Likes

FYI - I’ve fixed the ‘additional options’ popup to handle marking books finished a bit better. Namely, when you change a books status to ‘finished’ I auto update the Current Page to the total pages with a warning:

All the other places where you can mark a book finished (‘update progress’ popup / simple status dropdown on book pages) should already auto update the current page count.

I will note too that for the 160 user books which were clearly affected by this poor interface (books that were marked finished with a current page of 0), I went ahead and updated them to the total page count. I may additionally update 19 more user books that I suspect were also affected, but you’ll receive an email if I do that update.

If you are concerned and would like to know if you were affected, shoot me an email… I’ve saved the information. It was only affecting people marking their books finished through that particular popup, which wasn’t the majority.

Thanks!

4 Likes

I’ve been wonder for some time about the grading limitations (only 6 count?) And I wondered if it is causing ”grading islands” leading to unusual grades.

For example, if you start adding new books and give them say too high of an initial grade (let’s say 28 for a childrens book). Then only rate those against each other (or harder than other easy books) won’t that cause an issue? It’s not that uncommon a scenario, given the vast quantities of books. There are lots of books with single readers or gradings.

I have one book, that I simply could not reduce the rating for. I even deleted a bunch of old ratings to try and reduce the score, to no avail!
Only one other person has since read it and also commented that it was too high, their grading managed to atleast reduce it by one level:

I also wonder how this affects other books. If you read a genuine level 27 book, you’d rate it harder than the above book. That would push your new book towards 28, whilst having no effect on the old one? (As it has already reached the grading limit) This is what I feel I have been experiencing, other books trend towards to 27/28 whilst this one won’t budge…

Obviously the more people read, the less this is an issue ー but there are soo many books and niches that some will end up with isolated gradings.

So I wonder, should that ratings limit be removed or at least be adjusted for books with only a few users grading them?

4 Likes

Thanks for the note! Really appreciate the insight :slight_smile:

I first just want to correct this thought. Both books are affected. The genuine level 27 book would take rating points from the poorly grading book, moving the genuine level 27 towards 28 and poorly graded towards 26. When I say that you only get ‘6 gradings’ per book, I mean you’re only prompted for 6 gradings when you marked it finished. You can accumulate many more gradings on a particular book when grading other books. I do see the book moving down from your gradings, but I may need to investigate further.

The ‘initial grade’ you estimate it at doesn’t make much difference because of the temporary level system, but to a certain degree this can happen, yes. If a book gets marked fully graded at too high of level, then for a particular user it may have an island effect like you’re describing. I think it’s a little less likely than you expect however. Most people have a good amount of books they’ve read that are already graded and on platform and for any new temporary books, you can only grade them against fully graded books, meaning one bad fully graded book in your catalog shouldn’t have a super large impact.

And of course, every time you grade it easier against properly rated books, all those bad rating points do inflate the properly rated ones, but it’s spread out over many books.

Coming back to this book though, it does seem like an outlier (when I use my level estimator purely from the grades that I use for temporary leveling, I get level 25). I’m not entirely sure why it is ranked so high - perhaps some of the of the grades you deleted weren’t properly refunded? I’d have to check if I have time… but right now I feel comfortable making a rare exception and moving it down to level 25, which I only do if I feel the system isn’t internally consistent.

Right, thankfully the more people read the better the level, which is why it’s not too concerning! And thankfully, if people think a book’s level is significantly off, they like to mention it in the review, which further mitigates the issue. I agree that opening up grading to more than 6 would help mitigate the issue as well but it introduces other ones… namely giving the last grader on a book too much power.

You mention that Wakaman’s grading only moved it down by one level - that is by design. A user, at most, shouldn’t be able to move a fully graded book by more than 1-2 levels, otherwise the ranking becomes more the last user’s opinion rather than the aggregate.

Ultimately, reconciling all these sorts of issue is very difficult and may be mathematically impossible… you always need to compromise somewhere. I’d love to dedicate some more time to improving the grading algorithm and many here have given great ideas how to move forward. I do try to monitor situations like this, but it’s difficult to prioritize such an ambiguous, daunting task such as algorithm improvement when there are so many high value tractable tasks such as movies & more social features. I’d only prioritize if we see many such situations you’re describing, which would greatly diminish the overall value of the system.

So, keep noting these as you come across them! Definitely helpful for monitoring, thank you. I’ll think about if it makes sense to open up the limit when there are few graders… that could be an interesting solution.

And as one last note, I’ll just say that Elo is much more resilient than you might expect. So even if things are a little wonky in a few places, my experience suggests that it all ends up working out alright :slight_smile:

4 Likes

Ah cool, from older posts I thought only the first 6 headings counted to avoid heavy readers having too much influence.

I’ve been skipping gradings recently :face_with_spiral_eyes:

I believe there were ratings against manga that I deleted. I read the first 3 volumes of JoJo (33, why?) a few months before, and didn’t think it too different at the time (both needing tons of lookups) so probably graded it as similar.

I deleted a lot of my older manga/book ratings and owned tag. As I feel a lot of manga ratings just don’t make sense (full furigana 少年 stuff in the 30s). So that might be related?

Edit: so many typos… stupid iOS :smiley_cat:

1 Like

I prefer people to skip gradings if they don’t feel confident about them - no issues! Just happy your grading and reviewing at all :slight_smile:

Yes I see that. It’d take a little effort for me to delve into what exactly went wrong, but I could figure it out. Not sure if I have time to, but we’ll see :sweat_smile:

I’d be curious to hear more about that. Obviously manga in the 30s are more likely to be compared to non-manga, making their grades potentially more noisy. I don’t think full-furigana is really a dis-qualifier for a hard manga, but if you broadly think manga that are in the 30s are poorly graded, I of course would love to know! Do you not feel there’s a difference between the high lvl 20s manga and the lvl 30s manga?

2 Likes

It’s hard to compare manga and novels. There are just some outliers that make me want to go read a volume from those series to improve the ratings, but I prefer novels at the moment :sweat_smile:

Books marked around 30 (so average?) include some be pretty commonly known ones: キノの旅 (29)、ビブリア古書堂の事件手帖 (30)、ようこそ実力 (31)、無職転生 (31)

Those contain roughly 1500-1600 unique kanji per volume (jpdb) so lots of N2, N1, and beyond vocab/kanji. Likewise lots of N2 grammar points and N1 stuff here and there.

I don’t read manga much these days, but I seriously doubt common 少年 manga is at a similar level. It’s hard to compare, without extracting the text, but: HunterXHunter at 31, Jojo 33?

That makes little sense to me… when comparable fantasy 少年 manga is around 23-26. OnePiece 23, Dragonball 23, etc. Not too dissimilar too books aimed children at of the same age.

I’d just expect a level 30 manga to contain similar dialogue to level 30 LN. I am sure there are plenty of manga like that.

3 Likes

Would it be frowned upon to have a separate account for manga? I do the same on bookmeter.

Obviously, I would avoid double grading.

2 Likes

Right, that’s certainly true… and it’s why I prioritize prompting manga to manga comparisons over manga to novels or other things. In general, I’m more concerned with making sure that the hardest manga are labelled as the hardest manga, rather than necessarily making sure everyone agrees a L33 manga is more difficult than a L31 novel, as people have different perspectives on what they find difficult… which are exacerbated when comparing across book types.

But you’re right, the system does imply a cross comparison, which isn’t perfect. I’m not sure what exactly the right answer is yet. My general theory right now has more been about the end user, rather than focusing on a book has x number of kanji, x number of N1 grammar points … etc like jpdb. It’s more about finding the aggregate opinion on a book. If you can read JoJo and want to read a novel, would konbini ningen be comfortable for you? Would it feel more challenging?

If the rating system succeeds in that sense, then it’s a good recommender. I personally believe there are more subjective things when it comes to difficulty than simple kanji metrics, which is why I like the subjective system on learnnatively.

Getting back to your point though, if direct cross comparisons between novels and manga messes up the best difficulty ordering of the manga themselves (remember, these are deprioritized so hopefully fewer), than those cross comparisons are simply a hinderance and should be prevented. So certainly welcome your input if you do end up reading any of those :slight_smile:

No issues! But yes please don’t try to double grade :joy:

And I really appreciate your direct examples you mentioned, I’ll keep those in mind and may send an email or two to users who have read those things, to see their thoughts. The L30s manga been a concern of mine as well.

Edit: And yes, I’ve also thought about perhaps have a separate level system for manga, but I think it’s a bit more confusing, even if it is more accurate. I agree though, it’s something to think about as people do have very different opinions when it comes to difficulty between novels / manga.

5 Likes

What about the L40s ones? :stuck_out_tongue: I do want to read 攻殻機動隊 some day, so I’ll see if it feels like it belongs to that level eventually, anyway.

5 Likes