Talk:FeatureList

(Pardon my editing. If anyone is more familiar with wiki editing than I, maybe they would care to format this more appropriately. However, I would not appreciate any editing of my comments other than for format.)

MODDING

First: I recommend against allowing mods in any discussion in which you have taken part. Preventing mods only to "your comment chain" would be far too complicated to implement and there are too many potential holes to exploit if someone wants to cheat.

Technobabble removed --- I don't see why it's really a desirable feature. You're opening the door to people gaming the system by modding other comments in the topic that they don't like, in order to support the comments in that topic that they did make. If you mod all your enemies down, and your own and friends' comments up, you distort the whole picture. My opinion is no. It's just a bad idea to allow mods in the same forum you comment in. (Please understand that my intention here is to be constructive. I used to think this should be allowed on the original site too, but after giving it more thought I changed my mind.)

I understand where you're coming from; what I and others have experienced is that we don't bank mod points until such time they want to use them since they expire. Most of the time when I have mod points, I find myself annoyed that I can't contribute to a conversation that I'm now paying close attention to. Are you specifically concerned about an issue-poster who will hunt through articles once they have mod points to game the conversation?

Second: I recommend against giving users the ability to mod ACs up. Why bother? They're ACs. Allow downmodding but not upmodding. That gives them incentive to create an account. (I realize this would penalize the occasional brilliant comment from an AC... still, I say: let them create an account.)

ACs do sometimes post very good content. Sometimes individuals mentioned in an article feel the need to post with some plausible deniability. I'd advise against this. Perhaps a feature might be to simply hide AC posts as a display option (but not the default). ::Add a second vote here. AC comments should be moderatable in the same way as other comments, but I do like the idea of offering an easy way to filter out AC posts (Heck, posts from any Nick should be blockable, at the users descretion)

"Sometimes individuals mentioned in an article feel the need to post with some plausible deniability." Regardless of the occasional "good" purpose this can be used for, this is called sock-puppet activity and is pretty universally despised in online forums. Encouraging this kind of behavior will not do your site any favors.

I disagree. I feel that being able to contribute or add an additional detail anonymously (plausible deniability) is vital, especially because of the sensitive nature of some articles with respect to where people work. Remember that a moderator still has to moderate the anonymous post up, and moderators are aware that an anon reply to an authenticated post might be a sock-puppet if it looks strange or like someone is deliberately trying to sound like a second voice.

Slashdot was not a site optimized for Facebook arguments about politics or DND. Even if you can "win" an argument with sock-puppetry; if you won an argument on Slashdot you have already lost. It's a site optimized for technical discussions and each post should be able to stand on its own merits and there are no winners or losers, just informative or bullshit; the only amplification (speaking from authority) should come from registered posts and reputation. That's why Slashdot allows some registered or high-karma accounts to post with more points by default. If there was some real weird conversation steering by a bad actor with sock puppetry combined with mod activity you're already talking about needing multiple accounts in good standing, and that takes resources. If it can be detected by the clustering analysis and investigated, then we could combat it. If it can't, then it's a sophisticated attack by someone with a lot of patience or money, and in which case I don't think we can prevent that.

I think that a targeted attack that abuses the ability to post anonymously could just as easily be done without anonymous posting (and yet another account), so I think the value of being able to post something very valuable without fear of repercussion outweighs the ease with which someone might try to make their own argument seem popular. I'd rather fish the depths for a gem and take that "go ahead and post the master HDCP matrix, free speech is paramount" approach that Slashdot had than tell you to use a pastebin.

Third: I do not think fractional mod points is a good idea. It adds unnecessary complication, decimal numbers use up more space in the database and add performance penalties, etc. If you want to make some mods worth more than others, make them worth more than one point. E.g., 2 points and 1 point rather than 1 point and 0.5 points.

You wouldn't store decimals, just the traditional modpoints times 2. The fractions are a UI detail and their conversion (and truncation) are a function of the code that prepares fetched data to be templated and rendered to HTML. The reason for adding this is so that the default rendering option still shows a -1 to 5 scale as before, but allows for Funny and Flamebait to be de-emphasized. It's really a matter of empowering the default, non-logged in view, which we can't ignore.

I know how it works. You are still adding needless complication. Regardless of whether it's "just display" or not, you are adding floating-point arithmetic that simply isn't necessary.

We could change the range. But I like the idea that the score is in a 0-5 "stars" range. Familiarity and all. I'd like it not to go into double digits (0-9) at the worst. And you're making a bigger deal out of it than it is considering PHP and Perl have no issue with floating point vs. integer scalars (and the database wouldn't store it). This isn't something to get worked up about. What I'd like to hear is a solid argument against finer weighting options.

"but allows for Funny and Flamebait to be de-emphasized" Why would you want to de-emphasize them? I think Funny is well-deserving of a full point, and Flamebait, while sometimes misused, is still a worthy mod.

As proposed, the fractional weights can be changed on an individual basis. If you think "funny" should be modded with 0.5, 1.0, or 1.5 points, it is your call. That said, each of those points still costs you. If you have 2 mod points, then you can mod 2 comments at +1 or you can mod 4 comments at +0.5. As a moderator, it is your call.

Slashdot already allows for user-defined weights for rating types. I have seen some user journals earlier on from slashdot devs where they talked statistics on how top-level comments end up getting overmodded (they receive a disproportionate number of different ratings to value the unauthenticated user score). Seeing how some comments quickly shoot up in score, I think this is still happening. I feel this would allow for a bit more fine-grained control for users who don't value Funny as much as you might. (I'm one of them, I like Funny, but I think it gets steam-roll mod applied sometimes, like the Reddit effect you mentioned earlier)

Fourth: I would give a lot of strong thought to some kind of means to limit sock-puppet accounts and especially sock-puppet mods. I doubt there is any perfect way to do it, but having many times been victim of sock-puppet mods, I think it should be harder than in the original version. It should also be made clear to users that sock-puppets are considered the lowest of the low by the community. (They already are, but it should be made explicitly clear.)

Perhaps a new requirement is a feature in the backend that uses cookies and/or IP/browser fingerprinting to cluster accounts, and then further watch for correlated activity among them to identify potential sockpuppet accounts. A manual adjudication process would be necessary. While multiple accounts would not be specifically against the ToS, the site would reserve the right to ban accounts believed to be acting in bad faith (trolling, spam, sockpuppetry).

I agree. I think a form in the user area could be used to register a "complaint" about suspected sock-puppet mods. This could run an automatic routine to find correlations to cookies or IPs (cookies is probably more reliable). If that is above a certain threshold, it could be forwarded to moderators for a review. Just an idea, but that would give users a way to complain, without incessantly bothering moderators, yet allow moderators to act if anything is real.

Fifth: I recommend adding an "I disagree" mod, which would be worth +1 point.  :)

Only if you change it to be "Devil's Advocate" and conversely a "Groupthink" downmod at -1. These are more descriptive rather than opinion-based tags (even though they really all are) which are easier to metamod and force you to think about your perception of the community attitude, not just your own.

It was just a joke. I like the groupthink idea, though (a "Kool-Aid" mod). I think it might be abused though.

Some others have already expanded on this list apparently :)

Adding more moderation tags to the list is okay, to a point. It is way too easy to overwhelm users and moderators alike. We should probably collect a list of desired tags, and vote for maybe the top 10, just to keep it reasonable.

Sixth: "What about not-posting in the thread where you moderated? Good or bad rule? Maybe if you really must post, allow cryptocoin paying for post?" HELL NO. People should not have more say just because they have more money. This is a very bad idea.

Agreed. This is a dangerous precedence. Also, I don't think it's a good idea to incentive-ize anything beyond low UIDs, more browsing options (site scraping, CPU-taxing searches), maybe auctions for popular channel names to be owner/editor on.

Show multiple mod reasons. One thing that always bugged me about Slashdot's modding is that you could get a +1 funny and +1 insightful, but only one of them shows as the reason for the upmod. Why not "+2, funny, insightful" ? <== vote this one up! Mfnickster (talk)

KARMA (CHOPS)

I really don't think mod points or karma should depend on user's customized point systems. Leaving it open to customized scoring could make your karma system vulnerable to cabals of users, unintended consequences, or a form of "elitism" (Good Old Boy Syndrome). It's been known to happen. There should be a clearly articulable standard that cannot be modified by users. Anything else is subject to chaos. If you don't give users a clearly defined standard, they won't know how to behave.

Note about the above: if what was meant was that it shows differently to users based on their custom scorings, fine. But the internal base scoring should not be modifiable by users.

Eeep edit conflict ... YES that's what I meant and I agree

A Karma (or Chops) system needs to go above what the original system uses (only a few levels, with "Excellent" being the top). The current system allows people to receive massive numbers of positive mod points, but still remain at "Excellent" karma, which can be lost in a single day; they can find their karma at 0 if they are ganged up on by others and receive only a few downmods in a single conversation. Yes, it does happen. Karma should be allowed to accumulate (negative OR positive) beyond just a few points on either side of zero. It need not go on forever, but people should be allowed to develop some kind of persistence of positive or negative karma. That way somebody with a few years of good karma won't find themselves below 0 due to a single flamewar. And vice versa.

This already happens from what I understand. Karma isn't totally capped right now on slashdot; there's a functional limit for display purposes and mod perks. I think there's a weighting system where karma for everything older than a few months or so "sum from back then", capped at the functional limit, and then all the subsequent karma-affecting activity since then allows it to go above or below that number, and it's again capped at the limit for checks. And then they summarize the oldest month on some periodic basis into the new "sum from back then", but they again are capping it when stored so all the extra upmods in the past can't outweigh current downmods now.

I don't think they really do this. I say that because I have had many years of good /. karma, with consistently many more upmods than down, only to see it literally disappear in a single day, over a single discussion in which I took a "politically incorrect" position (about 5 downmods). I had to re-build it from scratch. I thought that was really pretty terrible: losing years of reputation in a single discussion. That should not be allowed, but was.

They do do this. But it relies on an external script to do some stuff (it could be more nuanced on the slashdot in production vs. what's in the CVS). And I've done this too (got downmodded a lot), but have not experienced what you experienced. Or perhaps I posted enough other stuff that got upmodded subsequently that it didn't matter the next time I checked.

I'd like to continue using such a system so that you're given the ability to carry Chops above the "Excellent" threshold value, but summarization means that you can't "bank" good behavior limitlessly and it also cuts down on expensive database queries for long-lived accounts.

I don't think limitless would be a good idea either. But I also don't think big queries should be necessary. Just accumulate points in a field of the main user record. Allow the accumulation of a certain amount of points above "Excellent", or below "Terrible" (or whatever the low limit is... I've never been there). I don't see the utility of maintaining complex running averages, when a simple accumulator achieves most of the same goals, as long as you let it go somewhat above the top visual rating and below the bottom visual rating. (And as I mentioned before, I don't think the original site actually does averages, either.)

What I'm hearing is that the summary-ization threshold is too low. Are you okay with it turning everything beyond "4 weeks ago" into a past total and starting point for karma (capped at 150% the display?). Let's say the display limit is "10" for excellent. That means you could have 15 upmods banked from a month ago, plus any recent upmods, and then need that many downmods to happen to you all at once to put you at zero.

Care should be taken to avoid multiplication of karma points, either positive or negative. I.e., you should not receive more positive points for having positive points.

I can't think of a situation where this occurs as exists or proposed (except with sockpuppetry... )

I agree. I just thought it was worth mention because I had at first misunderstood your description of how reputation is scored. The "Reddit effect" is somewhat minimimized on Slashdot because comments don't move to the top based on their score. But on Reddit (if they still do it the same way) it is a serious positive feedback effect, in which mediocre comments can get shoved to the top if they get a few upmods early, while more insightful comments get buried in the noise. That's not directly related to karma (chops), but the effect should be avoided in any scoring system.

Agreed.

UTF-8

Just to play devil's advocate here, I thought Unicode support (and lack thereof) was one of the things people commonly bitched about on Slashdot. So if it is possible to be smart about Unicode from the beginning, that could help us take off. I say this as someone without the relevant technical knowledge of the back-end so I don't know why UTF-8 is especially important to you guys. --Sir Garlon 16:34 EST 2014-02-10