SoylentNews is people
SoylentNews
[Update 20190127_200249 UTC: corrected number of downmods to qualify for mod bomb from 4 to 5. Clarified that no mod bans have been handed out in a long while. --martyb]
Our primary goal at SoylentNews is to provide a forum for the community; In as much as is reasonably possible, we try to take a hands-off approach.
The infrastructure provides a means by which the community can (among other things) vote on polls, publish journal articles, submit comments, and perform moderations.
There are, however, some things that require an active role by the admins.
One of these is dealing with moderation abuse, something which can come in different forms. See the FAQ for some background. Addressed there are "mod bombs" and "spam mods". A mod bomb is deemed to have happened when one user (user1) has performed 4 5 or more downmods against comments by another user (user2). Upon review, if a mod bomb has been found to occur, then the moderator (user1) gets a 1-month mod ban on the first occasion; 6 months on the second and subsequent times. Mod bans have not been issued in a LONG while; extra mods are reversed.
Sockpuppets: And now we come to the focus of this article: there is another form of moderation abuse: sockpuppet accounts. Wikipedia has a suitable description:
A sockpuppet is an online identity used for purposes of deception. The term, a reference to the manipulation of a simple hand puppet made from a sock, originally referred to a false identity assumed by a member of an Internet community who spoke to, or about, themselves while pretending to be another person.[1]
The term now includes other misleading uses of online identities, such as those created to praise, defend or support a person or organization,[2] to manipulate public opinion,[3] or to circumvent a suspension or ban from a website. A significant difference between the use of a pseudonym[4] and the creation of a sockpuppet is that the sockpuppet poses as an independent third-party unaffiliated with the puppeteer. Sockpuppets are unwelcome in many online communities and may be blocked.
Right here I'll admit that I was sorely tempted to take unilateral action. Name names. Apply mod bans. And... you get the idea. Instead, I'm trying to take the high road. So, instead, I chose to present what I found to the community, solicit input, and then see what, if anything, needs to be done.
There may well be other cases, but the one I have discovered shows this history of upmods. Out of the 100 most recent moderations performed by "user1", 80 of those have been upmods of the same user "user2". And of these, there have been 10 upmods on January 21, 10 more on January 22, and yet 10 more on January 23. (For those keeping score that is 30 points in 3 days).
I cannot imagine in any way that 30 upmods in three days by "user1" on "user2" is reasonable or desirable.
This would be purely academic except that comment moderation affects a user's karma. All registered users start with a karma of 0. Submitting a story that is accepted on the site earns 3 points. Each upmod to a comment of yours earns a point. Similarly, each downmod deducts a point from your karma. Get enough karma and when posting a comment you can give it extra visibility so that it starts at a score of 2 instead of at 1. (Comments posted anonymously or by ACs start at 0.) Get a low enough karma and you earn a "timeout" against posting comments for a month.
Inasmuch as "user1" was able to perform 80 upmods of "user2" in 19 days ("user2" had hovered near the karma cap of 50 when this all started), that means that "user2" received approximately 80 downmods from the community. Excluding the actions of our sockpuppet ("user1"), "user2" should have been in negative karma and thus in a month-long "timeout".
What I see is that the community has spoken (the comments posted by "user2" are not of the kind the community wants to see on the site) and that has been intentionally countered by the sockpuppet activity of "user1".
Rather than the admins taking a unilateral action, I am asking the community what should be done in this case (and any others like it that may come up)?
I offer a proposal that is analogous to our handling of a "mod bomb."
What is a mod bomb? Four (4) or more downmods in 24 hours by "user1" against comments posted by "user2". qualifies as a mod bomb and earns "user1" a 1-month moderation ban (initially; subsequent mod bombs earn a 6-month mod ban) It's been a long time since mod bans have been issued..
Proposed: Four (4) or more upmods in 24 hours should also be considered a mod bomb (sock bomb?) and should receive the same treatment.
The point of moderation is not to bestow karma points, it is to help improve the visibility of well-written comments and reduce the visibility of the lesser ones. The karma is simply an incentive to actually perform the moderations.
I've toyed with various values for number of upmods per unit of time (4 per day? 20 per week?) I keep coming back to the same metric we use for our existing "mod bomb" definition: 4 down mods in one 24-hour span that commences when mod points are handed out at 00:10 UTC.
So, now it's your turn. I'd appreciate your feedback and thoughts on this. If we should choose to implement it, it would probably have a soft launch with any "violations" being met with a warning.
Ultimately, it's your site. How do you want us to deal with sockpuppets?
(Score: 5, Interesting) by AthanasiusKircher on Friday January 25 2019, @03:46PM (28 children)
I'm going to violate my general rule here and actually reply to Arik, but I agree with most of this post. I'm very surprised that 4 downmods in one day is a sufficient threshold to flag someone as "mod bombing." Heck, I've seen a single poster frequently make 4 downmod-worthy posts on a single article!
As some of you know, I took a break from this site for a while because I believe moderation is not working here. Not that I have a much better solution -- I just feel like we have several users who take advantage of other users' goodwill (as well as quirks of the moderation system). But that's for a separate post.
What I do think is important to consider is that moderation serves at least two separate (though related) functions:
(1) Individual post moderation helps us flag useful posts in a discussion, as well as spam/trolls/flamebait and other bad actors.
(2) As a whole, moderation affects karma, which helps establish the reputation of users and thereby identifies consistently good and bad voices in the community.
The first of these is necessarily an individual judgment call. The second should only emerge with the consensus of a larger portion of the community.
Thus, while no system is perfect, one method I might propose could go something like this:
-- First, the "mod bombing" definition is far too broad. Martyb suggested 20 mods/week as an alternate to 4 mods/day in his post regarding sockpuppetry, but I think that's a reasonable threshold for mod-bombing too. If user1 is downmodding user2 at least 20 times in a single week, that seems like it should be targeted for review. (NOTE: before you have a kneejerk reaction about how this might affect karma, note I have an important qualification coming below.) If we were wanted to implement some other triggers for review, perhaps consistent targeting could also trigger a review even if it doesn't rise to 20 downmods in a single week (e.g., >50% of user1's mods over a given week or longer period are downmods of user2, or perhaps a different percentage). Some will note here that I am assuming downmodding should be fairly common, which I think it should be. I think we're far too lenient with bad behavior here. And I think a "4 downmods triggers review or ban on moderation" is a horrific policy that discourages downmodding of consistent bad actors.
-- Second, whatever policy we adopt for downmodding should be (as Martyb suggests) roughly mirrored for upmodding and sockpuppetry too. 20 upmods from user1 to user2 in a single week should probably be an unusual event. Though here I do think it's perhaps more important to take into account overall modding patterns, as we tend to encourage upmodding here. If someone uses all 70 possible moderation points in a single week, I don't even know if 20 upmods for one user should flag it as a problem. If X% of all of user1's mods over that period are simply upmodding user2 though, maybe that's a better flag. I don't know. (Again, note below regarding effects on karma.)
-- Third, the above comments also presuppose modification to karma calculation. I don't know the details of the current karma calculation, but I do think a single user shouldn't be able to affect another user's karma by a huge amount (particularly over a short time). Someone else here proposed a limit of user1 affecting user2's karma by a maximum of 1 point per day. I think that's reasonable, and will cut down on any possible effects of modbombing or sockpuppetry without discouraging moderation of posts. (Sometimes on a particularly active story, you get one user posting 10-20 posts on a single story, and if they get into an argument with another poster who is consistently acting like a jerk, a moderator may reasonably want to flag a bunch of those as good or bad.)
-- Fourth, regarding karma, I also think (as I said above) that karma should be a reflection of community consensus. I don't think a single user1 should ever be responsible for more than, say, 20% of user2's reputation. (If we had a larger community, I would lower that expectation to maybe 10% or 5% or even lower.) My off-the-cuff proposal would be to implement something in the karma calculation to the effect that user1's mods of user2 cannot exceed 20% of the overall karma calculation. Thus, if user1 consistently upmods or consistently downmods user2, even over a longer period of time, user1 will not be sufficient alone to raise or tank the reputation of user2. That ensures that even borderline modbombing or sockpuppetry won't have a huge affect from a single user to another individual user. (Yes, any system can be gamed, but this combined with my third restriction above would make sockpuppetry a lot more cumbersome to maintain.)
I don't claim any of this is a final answer. I just offer it with the following beliefs in mind: (1) downmodding here should not be discouraged; quite a few users behave with a "Jekyll and Hyde" mentality, maintaining relatively high karma which seems to allow them to act like jerks or trolls, and they should be able to be downmodded when they do so, (2) karma should be hard to earn, should take some time, and should require community consensus. (And I don't mean to highlight downmodding so much -- upmodding is also very good, but it also shouldn't be discouraged when someone is consistently posting good things... it shouldn't trigger an automatic suspicion of sockpuppetry.)
Obviously I also agree with Arik that any major action should be reviewed by an admin, rather than just an automated rule. Or perhaps some sort of warning displayed for a first offense, a minor punishment and more severe warning for a second offense, and then admin review before more severe action against a user. Just some thoughts.
(Score: 2) by AthanasiusKircher on Friday January 25 2019, @04:05PM (3 children)
By the way, I also realize that my suggestions would result in more "moderation wars." While I don't look forward to it, I do think it's healthy on controversial threads. Also, note that I am not encouraging people to downmod posts purely for disagreement (other than the "disagree" mod, which of course does little). However, I don't think anyone should be afraid to downmod someone who is acting like a jerk, posts in bad faith, or threatens civil discussion here. Others may disagree with my stance there, but if our goal is good discussion and not just shouting and insults, I think it's important to think about.
(Score: 2) by Arik on Friday January 25 2019, @04:33PM
Of course it's relatively easy to set an objective standard like 4 posts in 24 hours then enforce that impartially. I have to understand why the admins want a rule like that. I just know from experience they never work, that just turns it into a game and whoever exploits the rule more effectively wins. I've seen that happen too many times.
I never mod people down *merely* for disagreement but I'm all too aware of how subjective downmods are, even when I'm doing it. Is posterX really trolling, or mentally ill/incompetent or do we simply lack sufficient shared background to communicate? At some level that's impossible for me to know, or for anyone else. But at some point you just have to make an honest pick and hit the button and move on.
If laughter is the best medicine, who are the best doctors?
(Score: 2, Funny) by Anonymous Coward on Friday January 25 2019, @05:06PM (1 child)
Veteran of the Mod Wars Blue Oyster Cult
(Score: 0) by Anonymous Coward on Saturday January 26 2019, @06:57AM
Let's dance, city cop!
(Score: 3, Disagree) by SunTzuWarmaster on Friday January 25 2019, @04:46PM (10 children)
Check out some of my posting below. I think the lions' share of the solution is actually to simply limit modding. A limit on modding to "just a little, only to people with positive karma" makes a sock puppet network difficult to maintain, as the sock puppets have to contribute. 70 mods/week is just way too much - it is enough to significantly influence a discussion or to trash a user. Are people actually using 70/week?! I'm probably closer to *2*.
(Score: 4, Insightful) by Arik on Friday January 25 2019, @05:08PM
If laughter is the best medicine, who are the best doctors?
(Score: 4, Insightful) by http on Friday January 25 2019, @05:42PM (1 child)
...if you browse at -1. You'll see some that deserve to be higher, and plenty that deserves to be lower.
I browse at -1 when I have mod points. It's unsettling.
(Score: 2) by JoeMerchant on Friday January 25 2019, @09:19PM
Thank you for shaping the content that I see, I don't generally read anything that hasn't achieved at least a 3.
Of course, the problem is achieving community consensus about - well - anything, but especially which posts "deserve" to be modded up or down. As you say, some are clear, but many will fall into a grey area where some think they are good and some think they are bad, and which way those fall has a big impact on the content that floats up above a 2.
🌻🌻 [google.com]
(Score: 2) by RandomFactor on Friday January 25 2019, @11:17PM
Someone got peeved at me a month or so ago (probably because there are NHL and Olympic Ice Hockey Rink dimensions, and i was using the wrong one for measurement). It was idly interesting to see all the notices of old posts getting downmodded. They apparently got bored after a few days though.
I vaguely wonder if they hit the limit or were using multiple accounts.
Curse you Red Downmod Bomber!!
В «Правде» нет известий, в «Известиях» нет правды
(Score: 2) by AthanasiusKircher on Friday January 25 2019, @11:25PM (1 child)
I'd have to see moderation statistics. I'm not opposed to limiting the number of mod points, but my perception for why we have so many here is to ensure adequate moderation within a relatively small community. Some days I have used all 10 of my mod points (though VERY rarely), most days I probably use 0. I should probably do more.
Anyhow, if we can get adequate moderation of posts with a lot number of points granted per individual, that's fine by me. Or, as I think you (and others) have noted, only give large numbers of mod points to those who are consistently maintaining high karma or good posts or whatever.
(Score: 2) by Hyper on Saturday January 26 2019, @07:03AM
This is a good idea!
Publish anonymized moderation stats.
In graph form of de-identification is a concern.
Who mods the most? Daily? Weekly?
Are there positive and negative modding patterns between specific accounts? (Stripped of user names etc of course)?
Who up votes or down votes the most?
Do we get badges for hitting limits? "Modded positively 300 times" You get the Community Member award.
Modded 30,000 times! You get the "Part of the furniture" award!
(Score: 3, Insightful) by acid andy on Friday January 25 2019, @11:56PM (1 child)
Here's why the mod points got increased to 10: Too much trolling [soylentnews.org]. At the time, I agreed with the decision and it seemed to help a lot at fighting some of the off-topic spam and trolling whilst still leaving enough mod points to upvote genuinely positive contributions.
If a cat has kittens, does a rat have rittens, a bat bittens and a mat mittens?
(Score: 3, Interesting) by SunTzuWarmaster on Saturday January 26 2019, @12:47AM
Thanks for referring me to this. Maybe a solution is what was suggested there? Separate modding pools for AC/realUsers? That would seem to solve both the two classes of (traditional and sockPuppet).
(Score: 3, Interesting) by stretch611 on Monday January 28 2019, @06:06AM (1 child)
Personally, I agree with the 10 moderations a day.
I admit that I do not always use them all and often enough I do not use any. But I am more likely to mod knowing that I have points left if I see another worthy post.
However, maybe we should make a change to base the number of moderations you get on your current karma. Instead of everyone getting 10 points when they have a karma of 20+ (I don't remember what the minimum is any more.) Maybe you get 5 mod points at 20+, and 10 mod points at a karma of 45+. I would expect that most of the regular SN community that people listen too have a 50 karma, (I would not require 50 only because anyone can have a bad post from time to time.)
A similar possibility would be to assign 2 mod points to anyone with a karma of 20, and add 2 more mod points per day for every 5 more karma. (e.g. 20karma = 2mods/day, 25 karma = 4 mods/day, 30 karma = 6 mods/day, ... )
This idea will not eliminate mod bombing, but it will make it take longer before a sock puppet account can become effective.
Now with 5 covid vaccine shots/boosters altering my DNA :P
(Score: 2) by The Mighty Buzzard on Monday January 28 2019, @03:09PM
Nah, if you comment often and don't do a bad job of trolling on every single post, you're going to stay at max karma. It's pretty useless as a metric except for in very limited circumstances.
My rights don't end where your fear begins.
(Score: 2, Interesting) by Gault.Drakkor on Friday January 25 2019, @08:25PM (1 child)
This perhaps over time could add up to some chunk of admin/moderator time. Perhaps maybe a system where first tier of review is other logged in members. Probably would need a settings flag set indicating: 'willing to adjudicate'.
Proposed procedure.
An event of suspected mod bombing/sock puppetry(or other event) occurs and review is desired.
The relevant information is displayed to say five users(or more). The users make a decision on a sliding scale 0..100.
If all users agree (low standard deviation) then appropriate automated action is taken.
If users disagree more users can be asked or it is bumped up for moderator/admin review.
Benefits:
Allow some semi automated salable review, with reviewers being community members.
Keep workload of dedicated moderators/admins lower.
Concerns:
Will selected users be fair and impartial?
Will enough users be willing and actually do adjudication when asked?
How much would it cost to create relevant information display? I would suppose a compact view would be good idea even for admins/moderators.
Would revealing account X is doing mods y be too much privacy dropping?
Is this happening often enough for this to be worth while?
(Score: 2) by AthanasiusKircher on Friday January 25 2019, @11:31PM
The key word in my post was "MAJOR action." To me, a "major" action would be something like banning a user from moderation for a month or more, banning a user from posting (temporarily or permanently), and such things. I don't imagine these would be triggered very often. Warnings and a temporary ban on moderation for a day or even a week seems like it could be automated -- most users actually engaging in bad behavior will likely get the hint, and if the system is flagging false positives, they can contact admins about it.
(Again, note that my post suggested a relatively high bar for warnings and bans -- like 20 fairly consistent mods from user1 to user2 within a week, perhaps an even higher threshold for upmods. And warnings should be clearly given before any banning takes place.)
(Score: 2) by Joe Desertrat on Friday January 25 2019, @11:02PM (5 children)
Perhaps I am misunderstanding but I took the concept of mod bombs to refer to moderating a single post multiple times, not just modding for or against a particular user across the board.
(Score: 2) by Arik on Saturday January 26 2019, @01:47AM (4 children)
If someone triggers you so badly you want to make a real impact on their karma, you have to go back and look at their previous posts and hand out mod points on those also.
If laughter is the best medicine, who are the best doctors?
(Score: 0) by Anonymous Coward on Saturday January 26 2019, @07:06AM (1 child)
Some people are sick in the head.
No, I don't mean MDC. Actually ill. In need of therapy. Or a hug.
(Score: 2) by The Mighty Buzzard on Saturday January 26 2019, @02:54PM
Most people could benefit from at least a bit.
My rights don't end where your fear begins.
(Score: 2) by Joe Desertrat on Monday January 28 2019, @09:51PM (1 child)
Ah, I see. Wow. That seems like a whole lot more effort than I'm interested in putting forth. I can't say I haven't up or downmodded any particular user more than once in any given session, but I at least think I base any such decisions on the content of a post, I don't really pay too much attention to the user. It's usually more fun to argue the stupid and/or trollish posts, and watch them dig themselves deeper.
(Score: 2) by The Mighty Buzzard on Tuesday January 29 2019, @04:10AM
If that's all anyone ever did, there wouldn't be a limit. It's the douchebags who think getting their vendetta on is reasonable human behavior that ruin it for everyone.
My rights don't end where your fear begins.
(Score: 2, Interesting) by dvader on Friday January 25 2019, @11:13PM
+1 to Third and Fourth.
Modding based on relation and not content should be discouraged. To avoid bias, a single but active user should ideally not have as much influence as many less active ones.
Modding serves one main purpose: it makes the site readable. Karma etc is secondary (imho) and only necessary for distribution of mod points.
If I had a lot of time for implementation, I would try different measures of suspect behavior and perhaps make automated ban rules based on a total score. Things like:
1. Distribution of mods. Score based on how many percent of the up/down-mods go to the same person
2. Timing. How long does the user spend "reading" a thread before modding. Rapid modding would be more suspicious than slow.
3. Account activity. How many articles are read, comments are written, mods are applied etc.
4. Controversial mods. Is the mod "aligned" with the general consensus about the comment/user? This would of course never be enough on its own but could indicate if someone is actively trying to inflate a comment/user.
5. ...
On their own 2-4 mean nothing but a medium score on 1 and high on 2-4 would be suspicious. The best would be to implement and gather data for six months, then analyze and see what combination of factors indicates suspicious activity.
However, I am painfully aware that time is limited for most people so a hard rule of max N mods / day / user is better than nothing. The kind of behavior the summary described is detrimental to the site and should be dealt with. In fact, on top of banning, I would argue for flagging and Karma removal from the target after a manual review.
(Score: 3, Interesting) by requerdanos on Saturday January 26 2019, @12:39AM
Rather than moderating sensibly, which is apparently no longer desired, I am going to take this opportunity to post instead and say that the above statement is all three of interesting, informative, and insightful, if one-sided--because there are regular users who've made four *upmod*-worthy posts on a single article. Takyon comes to mind as doing this on a regular basis, but there are others.
In any case, this site does not have a large enough userbase of regular posters for four (or 4*4) of *anything* matching between two given users being any sort of "bomb," good or bad, and if someone is looking at such trivial stats and worrying over them, then I salute their dedication but disagree with their conclusions.
It's the same people story in and story out that make comments. Some of the same people make regular insightful posts and get the love frequently, and some of the same people make regular insipid posts and get the -1. "Dealing with" *four* of something (as "abuse"!) has nothing to do with "mod bombs" or "sock puppets." There's going to be a significant overlap between these "bomb groups" and "people who actively moderate." By all means, automatically remove moderation points from people who consistently mod up the good and mod down the junk, but only if moderation isn't important to you, because, again, anyone doing that is applying moderation to the same small group over and over and is going to match at least "four same mods" over and over. I am not a super-regular moderator, but I have certainly matched the both four-down-mods and four-up-mods in 24 hours because when I set about moderating, I moderate posts made by the same people posting over and over.
The tldr in case you missed it: This site does not have a large enough userbase of regular posters for four (or 4*4) of *anything* matching between two given users being any sort of "bomb."
I believe that your proposal to limit any interaction between two users (including upmod or downmod) to such a low threshold will produce results that group actual up or down bombs together with expected moderation behavior given the active and posting size of the userbase on a given day.
Given the "four mod limit" it's hard to take seriously the sockpuppet question, since four upmods do not a sockpuppet make. But I'd say that setting reasonable automatic thresholds such as ten upmods per day more than one day in a (some short period) + a total of (some reasonably high number >50 such as the 80 you mentioned) over a slightly longer period, with (some)% total mods applied to same user. Once someone meets a threshold, a live person (preferably more than one) should look and see whether the data indeed have revealed sockpuppetry or modbombery, with mod points withdrawn in clear cases where that's so.
Four upmods for user3 means user3 is making good, helpful posts that should rise to the top. Four downmods for user4 means that user4 is making insipid, offtopic, redundant, or whatever posts that should not rise to the top. Neither of these people is being "bombed." We're just discussing the news a couple days after it happened here, because we don't go to the general store.
Karma, as you point out, is not a "reward", and a consequence of that is that twenty sock puppets upmodding mr_egotistical just gets his karma to max value--with no detriment to the site other than the maybe-not-upmodworthy posts that he has made being a few points higher than they would otherwise be.
Thanks & Peace.
(Score: 2) by coolgopher on Saturday January 26 2019, @03:18AM
Good points, and I'm largely in agreement I think. One thing I do see as an obvious weakness here is that it would encourage large-scale sock-puppetry where before a single sock could give +20 karma, now 20 socks will be used. Maybe they'll circle-jerk for good measure. At that point it becomes a matter of distinguishing modding groups, which again can be interfered with by the socks also randomly modding others. Continue the race for a while longer and it'll soon be algorithmically implausible to distinguish between frequent socks groups and frequent genuine groups.
One option to delay this progression might be to a) only allow any one user to affect another's karma by +/- 1 at any given time, and b) only allow that effect provided the user is already in good karma standing. I think SN has the critical mass to sustain such a scheme now. Something along those veins would make it quite tiresome to get even a single sock account up to the level where it can affect karma.
Another option which strikes me as somewhat more dangerous, is to give subscribers some sort of extra boost (e.g. more mod points, a +/- 2 karma cap, etc). The rationale would be that at least they've put some money where their mouth is, but the downsides are obvious, and I could easily argue that SN should not become so easily influenced by doshing around. I'm also reminded of the Blizzard Real Id saga [cad-comic.com] here...
(Score: 2) by gringer on Saturday January 26 2019, @07:31AM
Then make every unique upmod (or downmod) count towards karma, making it impossible for a single user to [significantly] bump up another user's karma, no matter how much they are upmodded.
Ask me about Sequencing DNA in front of Linus Torvalds [youtube.com]
(Score: 3, Interesting) by stretch611 on Monday January 28 2019, @05:37AM
Actually, this sounds like a good idea to me. You should be allowed to moderate any number of posts, but only allow one point of karma change between users per day.
I do not see this as being a big moderation problem. Most individual comments will be corrected by the rest of the community... if you post crap, it will be modded down. If you are unjustifiably modded down, usually the community will mod you back up. (This will fail if the mod bomber goes back a week to an old story to change moderation, but then again, how many people read something older than a few days.)
As long as the rest of the community continues to properly moderate, Adding one point of Karma per day from a sock puppet account should be overpowered by down mods and negative karma from the rest of the community.
As for the sock puppet account, when caught, in addition to losing mod privileges for 30 days, maybe they can lose 10-20 karma in that account (or even all karma) This will require a sock puppet account to rebuild their karma in order to mod bomb again. (And unless you are in tune with the rest of the community, I would think this would be a long time to rebuild a reputation.)
Now with 5 covid vaccine shots/boosters altering my DNA :P