[X] My Assistant

This is something I've been working on infrequently.
Please let me know if you want to know anything.

Some disclaimers:
1. "Gallery" is more accurately described as gallery chain. It possible to go down the chain, but it would significantly increase time taken, even going breadth-first until 2018.
2. Galleries covered consist almost entirely of galleries visible with default settings to unregistered users, for various reasons. Any statements made reflect the galleries I've gathered data on in that subset.

2019 vs. 2020

In comparison to last year, we have...

+81.56% total file size from last year
+66.06% files from last year
+46.30% galleries from last year
+28.08% more unique uploaders
+7.39% hits per day
+1.62% visits per day

By percentage of total galleries,

Doujinshi (36.07%) to (26.71%) = -25.93%
Western (18.85%) to (16.46%) = -12.68%
Non-H (11.62%) to (15.06%) = +29.60%
Misc (8.26%) to (10.27%) = +24.39%
Manga (10.49%) to (10.11%) = -3.59%
Image Set (4.66%) to (9.06%) = +94.39%
Artist CG (6.30%) to (8.22%) = +30.51%
Cosplay (1.48%) to (2.76%) = +85.59%
Game CG (2.27%) to (1.35%) = -40.69%

By percentage of total hits,

Doujinshi (31.53%) to (29.86%) = -5.33%
Manga (22.62%) to (19.15%) = -15.36%
Artist CG (12.25%) to (13.90%) = +13.43%
Misc (10.56%) to (12.90%) = -15.36%
Western (9.29%) to (8.76%) = -5.68%
Game CG (4.69%) to (2.81%) = -40%
Image Set (3.49%) to (6.33%) = +81.66%
Cosplay (3.27%) to (4.04%) = +23.70%
Non-H (2.29%) to (2.24%) = -2.14%

Image Sets experienced the most growth, both in terms of count and hits, of all categories.
Game CG has been hit the hardest.

Traffic
For the most part, the number of galleries uploaded has increased across the board.


However, the average hits and visits per gallery has suffered an overall decrease.
This can also be inferred from the low increase in total hits and visits compared to the number of galleries.


An interesting observation can be made by looking at the hits distributions by each year.


In comparison to 2019, 2020 is "flatter".
It appears that the site receives more even traffic throughout the day compared to the previous year.
This is an obvious effect of COVID. People are staying in, giving them more opportunity to enjoy themselves throughout the day.
Sleep schedules are also likely affected, bringing in more traffic when people would otherwise be asleep.
Chinese galleries especially have the most visible flattening, presumably from their stricter quarantine protocols.

2019

Overview
2019 was an important year for us. Servers were changed. UI underwent a big revamp.
Some major concerns and scares were raised.
(And yet, 2020 manages to be worse, haha.)

And in that year, we contributed...

7,549,005,259,652 bytes (7030 Gigabytes or 6.8 Terabytes) of files
6,192,143 pages of content
747,694 tags
70,935 galleries
26,377 torrents
7,603 uploaders
1,007 unique fetish tags
3.95 average stars

Of these galleries, I've collected stats from 69,000 of them. These are the galleries used from this point.
This amounts to…

23,252,870,164 total hits - An average of 63,706,496 per day
1,206,755,956 total visits - An average of 3,306,181 per day

Categories
Here's a breakdown of the categories

01. Doujinshi (25,588)
02. Western (13,372)
03. Non-H (8,245)
04. Manga (7,438)
05. Misc (5,858)
06. Artist CG (4,466)
07. Image Set (3,305)
08. Game CG (1,610)
09. Cosplay (1,053)
10. Asian Porn (0)

Languages
Some errors in this part, mostly from cases like dual-language galleries.
In those cases, it goes with the first encountered, alphabetically.
Like the gallery system, Japanese is the default language.
Here's the top 10:

01. Japanese (32,933)
02. English (12,070)
03. Chinese (8,233)
04. Korean (5,257)
05. Spanish (2,629)
06. Portuguese (2,403)
07. Dutch (2,036)
08. Speechless (1,525)
09. Russian (1,114)
10. French (1,108)

Other languages are <400.
We have Croatian, Tagalog, Norwegian, and Swedish at 1 gallery each.
Of the 70,935 galleries, 26,463 are translated.
That's 37.3% of the galleries! Thanks for your efforts, translators!


Tags
The tag system is a core part of what makes E-Hentai so great.
Let's have a look at exactly what type of content the gallery system has to offer.
We'll start off with the most frequently tagged, uh, tags.

01. female:big breasts (22,679)
02. female:sole female (20,626)
03. male:sole male (16,396)
04. female:stockings (11,112)
05. female:nakadashi (10,490)
06. misc:full color (9,979)
07. female:blowjob (9,767)
08. misc:group (9,235)
09. misc:mosaic censorship (7,872)
10. misc:comic (7,390)

Nothing too surprising here! Most people, and most content is relatively "vanilla"!


Traffic
Now this is something interesting!
It's time to find out exactly when things get fappening.
Breaking down galleries by upload time, this is what we get…

If we further separate it into the top languages…

And look at the hits rather than gallery counts...


It should be noted that the Chinese are on a different scale entirely!
This is a hint of that "Asia peak" Tenboro talks about.


Trends
I'll try to remember as much as I can.
I'm probably missing quite a bit.
Here's the 2019 year in terms of hentai trends:

Fate/stay night: Heaven's Feel II was released on January 12, and the trailer for III was revealed August 2.


Gotoubun no Hanayome aired from January 10 to March 28, 2019.


Kaguya-Sama: Love is War aired from January 12 to March 30, 2019.


The Rising of the Shield Hero aired from January 9 to June 26.


Pokemon Sword and Shield was revealed at E3 (Jun 11, 2019).


Tenki no Ko (Weathering with You) aired July 19.


Shingeki no Kyojin Season 3 aired from April 29 to July.


Do You Love Your Mom and Her Two-Hit Multi-Target Attacks? aired from July 13 to September 28.


Kimetsu no Yaiba aired from April 6 to September 28, 2019.

2020

Overview
Well, it's been a hectic year.
A couple more days and we'll hopefully move on to a better time.
Not too much went on with the site itself. Some backend updates. A big expunge system revamp.
However, a dear staff member left us.
Rest in peace, Maximum_Joe.

This data should account up to the first week of December.
Well, this year we contributed...

13,705,930,839,559 bytes (12764 Gigabytes or 12.5 Terabytes) of files
10,282,711 pages of content
1,053,536 tags
103,776 galleries
66,966 torrents
9,738 uploaders
1,043 unique fetish tags
3.89 average stars

Of these galleries, I've collected stats from 96,879 of them. These are the galleries used from this point.
Keeping in mind that I've collected data up to Monday, December 7 (342 days), this amounts to...

23,397,382,038 total hits - An average of 68,413,397 per day
1,149,060,499 total visits - An average of 3,359,826 per day

Categories
Here's a breakdown of the categories

01. Doujinshi (27,726)
02. Western (17,083)
03. Non-H (15,633)
04. Misc (10,660)
05. Manga (10,492)
06. Image Set (9,399)
07. Artist CG (8,527)
08. Cosplay (2,859)
09. Game CG (1,397)
10. Asian Porn (0)

Languages
Here's the top 10 for 2020:

01. Japanese (52,641)
02. English (17,679)
03. Chinese (13,805)
04. Korean (6,709)
05. Spanish (4,227)
06. Speechless (2,126)
07. Portuguese (1,598)
08. Russian (1,208)
09. French (1,153)
10. Text Cleaned (720)

Other languages are <700.
Languages with only 1 gallery: Greek, Armenian, Finnish, Bengali, Hebrew, Norwegian, Albanian, Danish, and Bulgarian.
Of the 103,776 galleries, 32,846 are translated.
That's 31.7% of the galleries and a 24% growth from last year.
Thanks for your continued efforts, translators!


Tags
Here's the most frequently tagged for the COVID year:

01. female:big breasts (33,081)
02. female:sole female (25,594)
03. male:sole male (19,156)
04. female:stockings (15,517)
05. female:nakadashi (13,359)
06. misc:full color (13,176)
07. female:blowjob (13,138)
08. misc:group (10,487)
09. misc:mosaic censorship (9,576)
10. female:schoolgirl uniform (9,080)

Trends

Helltaker was released May 11, 2020.


Princess Connect! Re:Dive aired from April 6 to June 29


Kaguya-sama: Love Is War Season 2 aired from April 11 to June 27.


Re:Zero Season 2 began July 8.


Uzaki-chan wa Asobitai! aired from July 10 to September 25.


Love Live! Nijigasaki High School Idol Club aired from October 3 to December 26.


Genshin Impact publicly released on September 28, 2020.



Some of you may have seen quite a lot of this already (and more besides). For the moment, this is what I'm (somewhat) confident in sharing with everyone.
It's a taste of what's to come. Next up is the 2019 trends. By the time that's posted, the first layer of 2020 data should be complete.

I'm hoping to receive some feedback and whatnot before anything else, though.
Get a feeling if this format is fine or whatever, fix the most obvious errors, the works.

This post has been edited by saltcutlet: Jan 3 2021, 11:09

>01. female:big breasts (22,679)
As expected.

which timezone used for the chart? If it's ehentai time, then it surprises me that the peak hour for Chinese is actually from about 3-7 pm, instead of the 9-11 pm as I have always assumed..
more importantly though, I want to see the stats for 2020 compared to 2019

which timezone used for the chart?

UTC, same as the gallery system.

If it's ehentai time, then it surprises me that the peak hour for Chinese is actually from about 3-7 pm, instead of the 9-11 pm as I have always assumed..

I intend to cover this. 2020 data should be relatively complete after today.

more importantly though, I want to see the stats for 2020 compared to 2019

I've updated with some of the basic ones.

quite an increase compared with last year. +32,841 galleries (+46% from 70,935), +5,734 GB (+81% from 7,030), +2,135 uploaders (+28% from 7,603).. it seems that our uploaders worked overtime last year huh

btw, could you increase the tag list to top 20 or 30 instead? the current top 10 is just the easiest one to tag (like big breast, sole female/male, blowjob) without telling anything about the actual fetish..

quite an increase compared with last year. +32,841 galleries (+46% from 70,935), +5,734 GB (+81% from 7,030), +2,135 uploaders (+28% from 7,603).. it seems that our uploaders worked overtime last year huh

Yep! However, traffic has not grown to match the increased number of galleries.

btw, could you increase the tag list to top 20 or 30 instead? the current top 10 is just the easiest one to tag (like big breast, sole female/male, blowjob) without telling anything about the actual fetish..

I've included an image of most of the top tags.

I've also updated with some data for individual parodies, a light comparison of 2019 vs 2020, and some formatting changes.

This post has been edited by saltcutlet: Jan 3 2021, 10:39

+81.56% total file size from last year
+66.06% files from last year

Does this account for exact duplicate images which are not duplicated in storage? I'd have assumed larger files from imagesets would account for a lot of this bulk.

Great stats thank you very much! Can you share the data please? (IMG:[invalid] style_emoticons/default/smile.gif)

Excellent data poking as usual.

Excellent data poking as usual.

porking*

Does this account for exact duplicate images which are not duplicated in storage? I'd have assumed larger files from imagesets would account for a lot of this bulk.

Nope.
To account for that, each gallery would have to be visited and hashes grabbed from the individual image links (multiple times per page, if there are many images).
Some issues with this method as well, since thumbnails have collisions so not every duplicate thumbnail represents a duplicate image. To be 100% accurate, it would have to go through each original image, which is strenuous on both sides.

Although imagesets experienced the largest growth, they are still only 9% of 2020's content. As a whole, gallery filesize has increased. This has been consistent through the years.



It may seem that 2020 received a large growth, but file size growth appears to be exponential.
Probably due to multiple factors such as better software, better computers, less necessity for file size capping, and so on.
Here it is on a logarithmic scale.

Great stats thank you very much! Can you share the data please? (IMG:[invalid] style_emoticons/default/smile.gif)

Sure. I have a 395,203-line JSONL file of API calls back to 2014.

[mega.nz] https://mega.nz/file/HiwSSSiS#ytUOHXOnlFWun...L3_aTnJA5AxaqyM

Excellent data poking as usual.

porking*

(IMG:[invalid] style_emoticons/default/wink.gif)

This post has been edited by saltcutlet: Jan 4 2021, 01:41