Play length edits

Posted in

#1 by siotechua
2021-04-27 at 01:38
< report >Can somebody please fucking stop changing the play length of this game?!

I mean seriously people keep on changing the Play length from long to very long then another changes it from very long to long. Can anybody lock this editing of the game so that there are not changes anymore of this game.
#2 by mutsuki
2021-04-27 at 02:22
< report >if you skip all the lines and all the voices the game is actually very short so could someone get to changing the page to reflect that?
#3 by phantomjs
2021-04-27 at 06:01
< report >#2 Skipping the lines and voices in ANY Visual Novels make the length considerable shorter, so what are you implying? ( What's the point of reading a Visual Novel like this anyway?)

And to the person (and anyone else) using EGS to argue as fact on a VN's length, please stop. EGS has a voting system for a VN's length unlike here in VNDB where we are trying to give an objective answer to recorded title length and that = subjective length of time / medium time to complete shown (ESG also have a ton of, imo, troll votes; Rewrite's and Hoshi Ori Yume Mirai and look at the amount of people voting <30 hours as proof of what I'm saying).

To the admins:
To avoid the above situation (above - just us readers subjective views on how long we think a VN's length should be), I purpose implementing an objective definition to a Visual Novel's length here in the database. I personally like - 'A Visual Novel's length (in VNDB) is the amount of time it takes to fully read it without skipping any text or any scene (including 18+ ones if any), letting the voices finish on all text scenes without skipping it, and reading all available routes fully without skipping any'. That. or maybe change the current system in a vote based one like EGS.

Edit: FWIW, I think the length of Sanoba Witch should be <50hours even by using my definitionLast modified on 2021-04-27 at 08:42
#4 by kiru
2021-04-27 at 06:18
< report >^We already have a definition. d2#2

Sanoba Witch is, and always was, long, not very long. Scriptsize is roughly 2.7mb. You need 3mb+ of a typically voiced VN to become very long. Scriptsize is a handy indicator that wasn't really used to determine initial length, but rather it happened to be an easy way to make comparisons with games. Per every ~300kb it's 5 hours, or 1 minute per kb, that's a surprisingly fitting figure that's also easy to calculate. Unvoiced is usually less than that.

Yes, sure, it might not be 100% perfect, but it IS providing a way of comparison without keeping every single VN in a subjective way. In other words, Sanoba Witch is shorter than Rewrite, Fate Stay Night and the likes. Thus long. That the idea of categories. If you've played a bunch of games, you also know instinctively, that Sanoba Witch doesn't compare to the really long beasts out there.

Also, you can use EGS just fine, if you know how to use it. Everything lower than 20 hours on there is USUALLY fairly accurate. Everything above needs to be looked at with experience in consideration with scriptsize. Basically, 24hours + is usually equaling our long, while very long needs to be around 40. EGS is votes, and you have to consider that people may drop games or not read all routes. This gets more and more common with long works, which is why this discrepancy happens, as with longer works it also has a mathematically bigger impact.
i.e. 10 hour long, 1 person drops it after 5 hours, 1 person finishes it: 7.5 hours average.
i.e. 50 hours long, 1 person drops it after 5 hours, 1 person finishes it: 27.5 hours average.

Also do note that scriptsize isn't everything. If the game has actual gameplay, or the engine is utter crap, so the game is slow as heck, and requires hours of skipping read text (i.e. Chaos Child), then even though the script isn't too big (~2mb in this case), it can still end up a category higher thanks to that. EGS also kinda showcases this extreme slowness of the game.Last modified on 2021-04-27 at 06:28
#5 by tester
2021-04-27 at 06:48
< report >Scriptsize is actually quite bad comparsion criteria. Script formats may be very different, even between the versions of scripts of the same engine.

For example, if Kusarihime script of codeX RScript has a few technical instructions and is very compact itself, Alpha Nighthawk script has a lot technical instructions. To the extend typical Alhpa Nighthawk script may be ten times bigger than typical Kusarihime scripts (with exact number of lines).

Some scripts may have a few technical instructions and bigger concentration of strings and otherwise.
Not to mention JRPG engines scripts.

Once more, script may be compressed too.

Better criteria is, for instance, to check the number of symbols and to divide it to average Japanese reading speed. But there may be effects in the game, voices etc, so it's needed to correct it some via, for example, multiplying on some empirical coefficient.

So I think it's actually possible for Sanoba Witch to be 50+.
To resolve the problem it is needed to be compared more thoroughly, via number of symbols, not via some unreliable criteria like the "script size".Last modified on 2021-04-27 at 06:58
#6 by phantomjs
2021-04-27 at 07:24
< report >@kiru

Yes we do have a definition for the length section of a VN and I'm proposing we change it. The way it is phased is not objectively enough in my opinion. For instance, if I 'skip all the lines and all the voices' in a VN, does the amount of time i took to read and complete all the routes really counts as per VNDB's current definitionb(or is it any useful at all)?

Also, in reference to your scriptsize point, I'm not actually referring to it. I just think adding the that point on voices into a revised length definition will fit better than the current one

For EGS, I will admit I am not a frequent user of the site so I have no clue on how accurate your statement on the '20 hours USUALLY accurate' thing is. I usuallly look at it when I happen across arguments on length of a VN here happens, and most of these are for mid - long range. 1 major problem with using EGS now imo (etc the last person who edited the length for Sanoba Witch) is people likes to use the Med length (and the votes itself) as absolute fact on a VN, not to mention the fact that they will spawn threads and use EGS data to argue about the length. By changing the length definition to a more objective one, I think we can at least circumvent that somewhat
#7 by kiru
2021-04-27 at 08:07
< report >@5: Script size means a VNs text in sjis encoding. Aka one moji=2byte. Control code and the likes are not part of that.

In other words, it's not a bad comparison. It's the only comparison that actually works on a neutral basis. And the 2.7mb are exactly that. Just moji. (~1.3m)

@6: The thing is, for the most part vndb is actually very accurate, until some new people come in and think they know everything better.
Like I said, it's maybe not the perfect solution. But the important thing is that a game with a specific length category is comparable with other games in that category. If you change Sanoba Witch to very long, pretty much every game by this company would need to be changed, and a ton of other similarily long games all under "long" as well.

There's no point to that. Even if YOU think they are a bit longer for yourself, they are comparable in that category. Simply add a few hours on top for yourself.
The numbers aren't what you should be too focused on. The goal is, that two long VNs are somewhat comparable in terms of how long they are, and very long is reserved to the really big beasts out there, which Yuzusoft games do not belong to.Last modified on 2021-04-27 at 08:10
#8 by phantomjs
2021-04-27 at 08:40
< report >@7
You are probably right on the current state of things in VNDB, but why have this type of argument arise when we can avoid it?

Also, I think the current definition is rather ambiguous and unclear anyway. Changing it should bring only benefits in the long run. I myself knows the length of a VN in the database is a rough idea only, that I shouldn't be too focused on it, etc. Problem is quite a number of people don't, and they change it based on their opinion and use stuff like EGS median voted results to argue / forced their way of thinking down other's throats. Changing the definition or changing to a vote based EGS-like system should bring this down to a min in the long run.

As for the mass changes of VNs's length if my system gets adopted (I'm assuming that is what you are implying), well, we users will rectified the situation ourselves with time right?Last modified on 2021-04-27 at 08:50
#9 by zakashi
2021-04-28 at 03:19
< report >I like using link
It's a nice source.
#10 by kiru
2021-04-28 at 07:07
< report >@9: That's unfortunately not helping the case either, because you aren't supposed to take any time like that, but compare the game to already existing games in the category. You can do that with scriptsize, you can do that with time votes, provided there are enough. (i.e. egs times) But those don't translate 1 to 1 to the category.

Average reading times by people who use a specific platform only aren't really reliable nor is the intention of the categories to 100% reflect accurate average times, as that's impossible to begin with. The goal is, that games in a certain category are roughly comparable and give you and idea of how long it is.

Medium: The usual average full price release, 1.5mb scriptsize+- normally voiced. Those are the most common games released, and are like 90% of the moege you can find
Long: Longer than normal games, 2-3mb scriptsize normally voiced. A lot more rare.
Very Long: Absolute beasts that usually aren't made and might be compilations, have gameplay or are just huge exceptions. If a pure VN, usually above 3mb normally voiced

The time is just there to give you an idea of how long it might take. Quoting the guidelines: "To determine the length of a game, it's often better to ignore this time indication and instead compare it with other games you've played. It's all relative, after all. " And doing this very broadly, leads to the above.Last modified on 2021-04-28 at 07:13
#11 by phantomjs
2021-04-28 at 09:49
< report >@10

For the purposes and intentions of this thread, that howlongtobeat site provides data which is at least as relevant as EGS, and seeing how many people here likes to reference EGS's data in their argument, I think the data in howlongtobeat is extremely useful (at least in the case of this thread)
#12 by essword
2021-05-31 at 01:22
< report >Is that guy seriously trying to claim that 15 votes on an obscure english website is somehow more valid than 260 votes by people who played the game in its original language?
#13 by acereishiki
2021-05-31 at 04:35
< report >@12 People who believe that they're more legitimate than the majority opinions. They're everywhere, and dear vn god do they hard to argue with.
#14 by phantomjs
2021-05-31 at 05:25
< report >#12 an#13

Doesn't necessarily invalidates the data on that site unless you can proof they submitted fake votes for fun or something. And I don't find the playtime data on your beloved EGS necessarily more trustworthy; I refused to believe such a large % of players finished Sabona in <20 hours, and the likes of Hoshi Ori Yume Mirai, Rewrite, and the 1st Grisaia (just to name a few) in >25 hours without skipping any content.

Be more opened-minded people ~~
#15 by kiru
2021-05-31 at 06:49
< report >^How about you get a little bit more open minded? I mean, I already wrote how categories work. There's not much more to add. You can read slowly, you can read fast, the point is, that games of similar length are in the same category. And this game is simply not part of the really long games out there.

Get your mind away from silly playtime numbers. That literally doesn't matter. The time is just an example. What you are doing is more likely to change the time references of categories than the game's category.
#16 by phantomjs
2021-05-31 at 07:55
< report >^
How are I not open minded?

I never said EGS playtimes aren't at all trustworthy and to totally disregard the playtime votes. I was replying to #13 which seems to suggest that EGS times are far more worthy than a 'obscure' English website's data with comparatively far less votes.

I'm also not advocating that we disregard EGS playtime votes altogether, just that people here like to use them as evidence of a VN's length. I'm just against that. You even read what I typed previously?

And, what do you mean by this?
What you are doing is more likely to change the time references of categories than the game's category


Take a look at what you typed, it's absolutely true that some read a VN slowly, some read it fast, some read the entire thing without skipping any scenes or voices (like myself), some skip the h-scenes entirely coz most of the time they aren't relevant to the main story, and so go. Going by VNDB's current definition of defining length of a VN, it's arguably all of us are right, and thus it leads to arguments such as the one in this thread, with people using subjective data as objective evidence in their side of the argument. That's why I'm advocating for a more objectively definitive definition of the playtime.Last modified on 2021-05-31 at 08:16
#17 by mrkew
2021-05-31 at 08:32
< report >
it's arguably all of us are right
lolno. The rule says to finish all endings. You don't finish something by skipping through parts of it. Otherwise, you can do what #2 said. I mentioned in the how to edit thread that SW has 36hrs of voices, which kiru previously said should indicate a very long game. I was thinking the game is around 45-48 hours, but I'll put it on auto-read with 500 moji per minute auto-read speed and see how long it takes.
#18 by phantomjs
2021-05-31 at 08:44
< report >#17

Did the current definition say you must get all endings by sitting through everything (no ctrl though anything at all), or you must listen through all the voices on your way to getting all endings? Nope, and that's a big problem which leads to threads like this.
#19 by mrkew
2021-06-04 at 20:34
< report >Like I said, I spent the last few days having the whole VN play out on auto-read to find out the actual length instead of this guesswork, as should probably be done in any length dispute instead of locking the entry for half a year until everyone gives up.
Results first before everyone loses attention, it took 63 hours and 17 minutes, which is way over the 50 hour mark. Now the notes which answer the questions people surely have.

1. I mentioned the speed for unvoiced text is 500 moji per minute, which is documented as the average reading speed of a normal Japanese adult. However, the speed fluctuates between 500 and 800+ depending on how long the line is. Anyone can make sure of this by themselves by watching the recording at the end of the post, but on average it is even faster than 500/min. The first narrated line is - それは昼休み、トイレから教室に戻る途中だった。 - it takes 155 frames (at 60 fps) from the time the first moji (そ) of this line appears, until this line disappears and is replaced by the next line. That would be 500 moji in 61.5 or 56.2 seconds depending on whether it counts punctuation.
Now, that's pretty fast for an average JSL. But it gets a lot faster on longer lines. If you check 19:18, the line there 昼休みも終わり、授業が始まったというのに、いまだにオレには嫉妬の視線が向けられていた。振り返らずともわかる。 has 49 moji without punctuation if my eyes serve right. How long does take from the first moji until the line disappears? 200 frames. 3.33 periodic seconds. That is 500 moji every 34 seconds (30.8 seconds if it counts punctuation). There's about 2 minutes of narration from the point I linked above. I urge anyone who has doubts about the speed of the auto-read to follow and read along for those 2 minutes.

2. It ran on auto in the background, which means I couldn't have reacted immediately to the choices. I dealt with this by learning how to use a video editing software, and cutting out every pause in the recording. At worst, this adds like 3 minutes over the whole game, so it's not an issue at all.

3. This covers all routes and their epilogues. The route are in this order and the lengths are very roughly:
Common 11.5 hours counting in all the various brief branches from choices
Meguru 9.5 hours
Touko 11 hours
Tsumugi 9.5 hours
Wakana 5.5 hours
Nene 16 hours

You can find the whole recording on youtube here.
#20 by diabloryuzaki
2021-06-04 at 20:42
< report >#19
i vote this result because it is more accurate rather than people who calculate it by ignoring average reading speed or just skip some scenario
#21 by mutsuki
2021-06-04 at 22:01
< report >nope still a very short in my books can't convince me otherwise
#22 by phantomjs
2021-06-07 at 10:27
< report >#19
Wow, thanks for all the effort it took to make this recording
If the mods ever change, or if a consensus is ever reached among VNDB members that a objective definition similar to what I purposed will supersede the current one, then I think your findings will prove to be very useful and definitive. Otherwise, I think you will find a lot of objections if you, or anyone else, tries to use this method as objective evidence of a VN's length.

Saw about 3mins of the first 5 parts. I myself reads the non voiced parts faster, and I *usually* but not *always* listens to voiced lines fully (also, while I usually read every line and doesn't skip any routes, I tends to skip hearing voiced H-Scenes lines entirely a lot). All of the these = I finished it a lot faster than 63 hours.

I personally do still support this method of defining a Visual Novel's length as I think a Database should provide objective data.

Doesn't matter. No one should flame you for it when one can supposedly 'finish' such a Visual Novel like this in about 6 hours
#23 by cubky
2021-06-07 at 12:00
< report >I agree, clearly spending a few tens of hours recording an automated playthrough of the game is something we should strive for as a golden standard for every future "Length" edit.
#24 by bcirno
2021-06-07 at 12:19
< report >no one reads at speed of autoplay, this is not average length at all
if some cuck spent 60 hours reading this vn, this doesn't mean average should be 50+ hours on vndb page
this is just beyond stupid
otherwise, every fucking moege going to be marked as 50+ hours when in reality most people complete them in ~30 hoursLast modified on 2021-06-07 at 12:23
#25 by mrkew
2021-06-07 at 12:21
< report >I just did it to solve the dispute in an objective manner. I did not believe it to be 50+, but one can simply not ignore facts. Such evidence is not needed in most cases, but the entry is locked right now. I don't know if it's possible to search for locked entries via SQL, but I bet there are dozens of entries which have been locked for months, because the current solution to length edit wars is to lock it indefinitely and ignore all arguments from both sides.
Alas, not even an objective proof is sufficient, as I reported it and it stays locked even now, instead of fixing the entry and removing edit rights from wannabe rebels. Maybe we'll meet here again on the 1st anniversary of the VN's lock date.