Another Fucking Anime Database

Thread Description
Come help out

Halo

varishangout.com
Regular
That could be quite expensive on the DB
Yeah, that it would be. But that's what we call a "future me" problem.
I haven't even gotten anywhere near adding in ratings yet so I haven't thought much about it besides what would be nice to have, I'm still fucking around with the editor for writing comments.
1723624250425.png
1723624305302.png
 

YakuInTheFlesh

varishangout.com
Regular
Yeah, that it would be. But that's what we call a "future me" problem.
At my first company we used clickhaus (we had roughly 3mil new data points a day), but it's probably way out of scope.

As for the editor. Just bbcode (and maybe css) is enough. HTML is just too much of a pain imo.
 

JX475

varishangout.com
Regular
To comment on the rating:
In an app my previous company developed, we implemented a basic honor system. We had a formula for it, but the basic 3 points were: Age of account, number of reviews, avg. rating of reviews.
All users started at 0 "points" and would get smth like +1 per month, +1 per review, +2 per positive rating and -3 per negative rating. Also the company we made the app for could give/take points.

There were upper and lower limits from -25 to 75 and based on that a review would contribute less or more to the rating (a -25 was practically shadow banned, only adding 0.01 to the avg).

This system can still be abused (which is why you won't tell to the users, that it exists), but it's relatively simple to implement.
I agree with you and @Halo as a whole. This is a better idea, because on Anilist, it's still better than MAL with interface, features and even on app, but there is a lot of bad actors trying to influence overatted anime like Full Metal Alchemist and Kimitsu No Yaiba to the top and messing up overall ratings still sadly even on there. Implementing measures will prevent them from doing anything like this for VN, Anime, and ensure that actual good content makes it the top and be much more fair than just random accounts and popular fandoms using their influence to put themselves to the top.
 

Lusty Daicon

varishangout.com
All users started at 0 "points" and would get smth like +1 per month, +1 per review, +2 per positive rating and -3 per negative rating. Also the company we made the app for could give/take points.
This system can still be abused (which is why you won't tell to the users, that it exists), but it's relatively simple to implement.
This sounds very corporate, for lack of a better term. There's a bias towards positive ratings and activity for its own sake. It's also dependent on implementation obfuscation. Perfect for a commercial service, but not necessarily something more communal. Honesty shouldn't be de-incentivized, even if secretly.
 

Halo

varishangout.com
Regular
This sounds very corporate, for lack of a better term. There's a bias towards positive ratings and activity for its own sake. It's also dependent on implementation obfuscation. Perfect for a commercial service, but not necessarily something more communal. Honesty shouldn't be de-incentivized, even if secretly.
I think you guys are putting a little too much speculation into a system I'm not even sure will make it in yet.
I appreciate the enthusiasm though.
Although I'd like you to clarify why you think this would create a bias for positive ratings?
 

YakuInTheFlesh

varishangout.com
Regular
There's a bias towards positive ratings and activity for its own sake
If you base this on my example values, then it might be. These are however just that examples. The actual application I was referencing used far more fine grained values, since it takes some testing and calculating to create the right kind of bias.
Perfect for a commercial service, but not necessarily something more communal
I mean, I did pull that example from a commercially developed application to begin with.

But I guess Halo is right. There is no need to fully plan a feature that isn't even being necessarily implemented.
 

eldr

varishangout.com
hmm, maybe something could be put together by using third party software, there's stashbox which has a lot of the things being talked about already implemented (trust system and all that) here's an outline of how it works, the issue is that it's made for hard data only (so no reviews, scores or any of that) and it's not really designed with multi chapter stuff in mind, although there are features that would allow it to work like tags and studios

it also has a fairly robust API so if anything it could at least serve as a base to build upon or at least as a source of inspiration since it's an example of an open source, fairly successful community driven database
 
Top