I’m curious about this statement. Is malicious voting a frequent problem? How is it identified?
It’s only happened a few times, and only for something obvious like one-starring every single game by someone in a short time. Usually the malicious voter makes things easier by publicly posting that they’re going to downvote someone’s games or upvote each other’s games without having played them, thus making it clear that they have bad intentions. In cases where it’s uncertain we contact one of more of the involved parties.
But it’s pretty rare. In four years it usually comes up less than twice a year.
6 Likes
I continue to be curious about this. Is there some sort of formal process with oversight involved? Are statistics published? Are there defined criteria to be met before action is taken and a defined process for handling such a matter?
2 Likes
Any serious action like that requires discussion with other IFDB mods, usually trying to get three people in agreement. There is also an IFDB board slightly larger than the group of mods who are occasionally consulted on questions of policy.
The defined criteria are these:
It’s an ‘innocent until proven guilty’ system. There are several accounts that are, I believe, probably in bad faith or not helpful which we won’t take action on because there is plausible deniability. People are allowed to dislike games, even large classes of games; what’s against the rules is ‘bad faith’ voting, so downvoting someone for reasons outside of your personal opinion of the game (e.g. because you hate them).
The defined process is to alert the other mods, gather facts, propose a course of action, and get consensus.
There aren’t any statistics published, most likely because it is extremely rare. I don’t think any malicious votes have been removed in the last year. Each mod focuses on different areas though and do more work than me so some may have more to say than I do.
8 Likes
You can’t have hard-line defined criteria for a moderation system. That just leads to infinite rounds of abusers gaming the rules.
“Gather facts, propose a course of action, and get consensus” is the only thing you can say.
12 Likes
Very good policy.
4 Likes
I’m speaking in the context of this forum - I know you’re asking about IFDB which is different since users don’t personally interact as much on there, and rules are more about actual physical site actions than social interactions.
We always confer before major disciplinary action like banning or silencing, and usually it will be in the context of a Code of Conduct rule that is being broken, but if someone invents a new way to cause problems, we’ll still moderate that. It’s always easier to be able to point at a pre-listed rule that a user agreed to instead of making it up as we go along and risk being accused of making it personal. A common tactic is “Well, you don’t have a rule for what I did; how was I suppose to know not to do it?” This is covered by blanket rules; ours is “Always be Civil.”
Long annoying Hanon tangent about this forum and not IFDB
Example
We used to have a rule that we would “only moderate what happens here” but it turned out people were exploiting that to harass people off-site and then behaving perfectly here as a way to purposely make people feel uncomfortable. (It’s the thing where mom says to stop poking your brother and you dance around him pointing but not poking saying “I’m not touching you!!” over and over as a method of trolling.) We had to change the rule to say if we can verify off-site harassment that is egregious such as threatening or doxxing regarding anyone in the IF community, we can consider de-platforming them here to make sure it doesn’t carry over.
What that doesn’t cover is a situation I’ve witnessed where fights will start on another forum or community, and someone will run here to continue the fight where they believe we’ll protect them like they’re declaring Sanctuary. Sure, you’re protected inside the castle walls, but that doesn’t mean you can still continue to hurl insults over the walls in an attempt to just venue change an existing flamewar to your advantage. We don’t want those fire-arrows either.
Our preference is always education and behavior modification. We want to keep our users and have them all get along. The behavior isn’t a problem if the user understands why we don’t want them to keep doing it and agrees to avoid repeating the behavior. We’re less concerned with “right” or “wrong” or people’s personal beliefs so much as preventing them from repeatedly stepping on each others’ toes when they might not agree.
People who get moderated often like to claim we are “infringing on free speech”. This forum is opt-in and users must agree to the Code of Conduct to participate. We’re all for free speech, however there are specific subjects we have learned cause too many problems and aren’t relevant enough to IF to foster those conversations. General politics is a specific one.
The biggest mistake people make during the moderation process is in almost every case we’re not interested in arguing about who is “right” or “wrong.” Moderation is not a forum to justify the position that got someone moderated. Rarely are we going to go “Oh, we didn’t understand. Please continue posting in this manner that’s making people mad…” The person under moderation needn’t change their opinion, they just need to know how to either express it in a way that doesn’t cause friction with other people, or just agree to move on and not discuss that topic.
8 Likes
IMO, the difference here is that each malicious forum post itself contains malicious content, and can each be moderated just by reviewing the content.
On IFDB, we distinguish between written reviews and star ratings (votes). Users can rate games without writing a review; the ratings have no content except the number of stars you assign.
No individual no-text 1-star rating (or 5-star rating) can be a violation of IFDB’s Code of Conduct. You’re allowed to give a game 1-star! It’s in the pattern of behavior that we see sockpuppets, brigades, and other bad-faith activity.
13 Likes
Following up on this, a curiosity of the state of the Top 100 rankings on IFDB over the last couple of months: The game Excalibur did not appear on the June 10 2024 edition of the Top 100 list, but it reappeared sometime between then and a few weeks later. Were ratings for it deleted?
1 Like
Archive.org shows it has the same number of ratings in May 12 as it does now.
I’ve noticed my game Never Gives Up Her Dead doesn’t show up in the ifdb top 100 despite being ranked higher by ifdb’s own algorithm than my other games.
I suspect it’s based on number of ratings or something like that. It’s possible that the cutoff for number of ratings to make it into the list changed for one month for some reason, since Excalibur has 19 and my game has 18, so it’s right on the cutoff.
1 Like
That’s very interesting. I agree that per the stated formula Never Gives Up Her Dead should be on the Top 100 list now, somewhere around position #7.
Using Worldsmith as an example, the stated current weighted average (4.51) corresponds exactly with the formula result for a “minimum votes” value of 13. If it were 18, then the game’s weighted average would be 4.45.
The current shown weighted averages for The Wizard Sniffer, Toby’s Nose, Alias ‘The Magpie’, and Worlds Apart all also check out per the stated formula when assuming 13 as the minimum number of votes, so it seems that the Top 100 list is not operating as advertised. (Perhaps a bug? Does anyone know how to get in touch with IFDB user Pegbiter?)
… but that doesn’t answer the question about possible removed ratings for Excalibur. I guess my question at this point is: Can you even tell whether votes have been removed after having been deemed malicious? Is there a database record of their cancellation, or are they simply deleted? And if the latter, is there any log of the action anywhere?
2 Likes
IFDB is open source, so you can poke around to see if it keeps track of individual moderator actions; I’m not aware of any such feature, but I wouldn’t mind seeing one.
IFDB is backed up fairly regularly and imaged on archive.org, neither of which moderators have direct control over (both are automated processes), so you can check any shenanigans yourself through those features if desired.
We do keep a log of our own actions through the Intfiction DM system:
(The screenshot is a joke but we do in all honesty report to each other. And malicious vote removal has happened maybe once a year or less. Searching the records I find nothing recorded in June).
Whenever we act we communicate what we’ve done to the other mods, which keeps us all honest.
12 Likes
It’s a filed bug that we don’t have a formal log of moderator actions. Add history and log for admin operations on users profiles · Issue #178 · iftechfoundation/ifdb-suggestion-tracker · GitHub
I don’t think we’ve deleted any number of ratings for Excalibur in that time frame, but it’s hard to prove that we didn’t.
I agree with @mathbrush that @Pegbiter’s script has very likely updated his values for c and/or m without updating the list’s description. But, then again, it could just be a bug in Pegbiter’s script.
I think Pegbiter is crawling IFDB using our API. Maybe there was an error crawling results for Excalibur? It’s impossible to say.
I’ve filed a bug to re-implement his IFDB Top 100 algorithm. I don’t expect to work on it any time soon (perhaps never?) but I’d be willing to review a PR if someone filed one. Re-implement Pegbiter's IFDB Top 100 · Issue #481 · iftechfoundation/ifdb-suggestion-tracker · GitHub
3 Likes
… How did you ever manage to get -15 replies???
1 Like
Oh, right. Yeah, I suspected it was that but didn’t know…
1 Like
We must have deleted malicious replies! 
2 Likes
Oh, and one thing I have to say is you have 308 notifications on the hearts bit. You know there’s a dismiss button, right? 
2 Likes
I actually didn’t! Thanks for the tip!
2 Likes
All joking aside, it really does seem that there should be a formal log of moderator actions affecting the data. Periodic dumps of the database do nothing to address transparency here if rating removals are effected through simple deletion of records, because deletions between backups would leave no trace. Personally, I would think it would be appropriate to set up the database such that canceled ratings remain in the data (marked as canceled so that they are not displayed), each attached to a timestamp and the ID(s) of moderator(s) approving the cancellation.
It looks like the bug that dfabulich mentioned has been open for just about 3 1/2 years now. Its scope as currently laid out doesn’t cover rating removals at all.
2 Likes