Adding corrupt files

[Andemon]

We already had some discussion about this in the DB Change Requests, but I think that further clarification about the issue would be useful.

On the other hand I've been told that it's ok to add corrupt files that are widespread, but on the other hand I've seen such entries get deleted quite frequently these days, evidently just because someone happens to complain about them.

I've been very reluctant to add them, simply because the rules seem so unclear and it seems more than likely that the entry would just get deleted -- so why bother? -_-;

Certainly nothing wrong in removing self-corrupted files that only one user has -- but if the entry was just added, how could more than one user have it? I think the mods (by that I mean *all* mods) definitely should either contact whoever added the file and ask about it, or if that's not possible, then at least wait a month or so and see if more users add that file.

Furthermore, what is a definition of widespread file? Something that has been downloaded several thousand times from fe. boxtorrents certainly qualifies, but what about a corrupt file that has fe. eight sources in DC++? What about five sources? Three? What's the limit? While back, I happened to search for Metropolis. Found over dozen of different versions of the InsidiousDivx release, many of them with at least three sources. Should all those be added? (...what *is* up with that release, anyway? It's mindboggling that there are so many corrupt versions of it.)

The corrupt file rules should probably also be included in Wiki...

(btw, I was trying to find out exactly how many corrupt files I have, but noticed that the 'Corrupted files in MyList'-report doesn't show files that have been added as the regular file, but with 'corrupted version/invalid crc'-parameter. ...perhaps it should?)

[DonGato]

Just use Razorback' Statistics to see if the corrupted file is widespread or not (tons of sources) by its hash or be informed about the source of the file.

If the mod gets a response from the user who added the file stating that it's a self corrupted file (i.e. disk problems) then why wait? In case he doesn't now maybe is good to wait and if it's from a massive source (BT, ed2k release, IRC, etc) then I think it should be left because if not at some point in time it will be added again.

[pelican]

`tons of sources' aren't needed; if you can find it on any p2p network, fserve, etc. that's spread enough for it to be in the db.

There are perfectly whole files that are harder to find than many corrupted files that would be excluded by that rule.

[egg]

Here is what I put in the other topic:

Razorback stats really isn't a valid method for determining if anime has no sources; it can only confirm the presence of sources, not the lack of them. The userbases of various P2P programs don't actually overlap too much -- just because the file isn't available on eMule doesn't mean that it can't be found on IRC or DC++ or as torrent... This is especially true for unpopular anime and rare files.

I agree, Razorback can confirm that a file is appropriate to be listed in the system, but if it is not in Razorback that is not conclusive that it should not be. I have gone through a number of older CRC verified, ARC Forum posted links that are not in Razorback.

My suggestions for a procedure would be:
1) PM the user to verify the source
2) Check Razorback to see if there are sources, If there are, then leave the file, put in a comment that it was verified.
3) If the user responds it is self corrupted tell the user how to correctly handle it and delete the file.
4) If the user responds they got it from a particular source, verify the source (or trust the user) and put in a comment that it was verified.
5) If there is no response after a period of time, check Razorback again and if there are sources, then leave the file, put in a comment that it was verified.
5) If there is no response and after a particular period of time there are no other users, PM the user to tell them how to correctly handle it and then delete the file.
6) If there is no response for a period of time, but a couple other users have added the file to their mylists, PM them and see how they got the file... (Many people do not add files by ed2k links and somebody may have addeed by mistake.)
7) If there is no response for a period of time, but there are a number of other users who have added the file to their mylists, then the file is widely spread and that's your answer...

I didn't put in specific number of users or time periods, but it is just a procedure that I think is fair... It won't be fast enough for some users, but it limits the chances of removing something that really shouldn't be removed.

[Andemon]

Sounds good. Of course, how practical that is largely depends on whether the mods feel like it's worth the extra trouble to confirm the validity of such files, instead of just deleting them...

In any case, from now on I shall add any corrupt file that I get from p2p. Just like that Psychic Academy episode ten that I just added -- can't be found on Razorback stats, but quick search showed three other sources on DC++...

[pelican]

Sounds good. Of course, how practical that is largely depends on whether the mods feel like it's worth the extra trouble to confirm the validity of such files, instead of just deleting them...

...or, more sensibly: ...instead of just leaving them.