The flag weight system could use some improvements

Question

Problems I see:

1. How disputed flags work: as I argued here, disputed flags are useless and might do more harm than good.

   Really, what is the incentive to flag a flag invalid? The way I see it, I'm spending one of my own flags to ensure the user who flagged incorrectly doesn't lose any flag weight as they should, while I don't earn any FW.

2. The penalty for incorrect flags is way too high as your flag weight improves. I'm capped at 750 right now. If I get one invalid flag, it will take 92 of them for me to get back to where I was. This strongly discourages folks with high FW to flag questionable content (at least that's what I think).

   Granted, since your flags have more weight, you should be more responsible while flagging. But the risk-reward ratio is way too high IMO. In fact, if you have maxed out FW there is no reward at all.

3. Users with close privileges (ie: > 3k rep.) can't flag questions as "doesn't belong here," since that makes them cast a close vote instead.

   If we encourage users without close privileges to flag questions that don't belong, why don't we encourage users who cast (correct) close votes as well?

4. Flags found in /tools/flagged (10k only) are an easy way to "farm" flag weight. What I mean is, it takes much less effort to go through flagged posts and re-flag them than to use /review or flag posts "in the wild".

   About 90% of the flagged posts have "correct" flags, while less than 5% of posts found in review are flag worthy (these numbers are guesstimates based on my own experience, not hard facts).

5. Lack of a neutral resolution (with the possibility to add a custom message) that doesn't impact the flagger's flag weight. This would be helpful in guiding flaggers without punishing them. Currently, moderators can't provide custom feedback unless they decline a flag.

6. Flag dismissal decisions (eg: helpful or declined) can't be reversed. If a mistake was made, we can't fix it. This results in a ton of flag weight grievance posts on meta.

Answer 1

The penalty for incorrect flags is way too high as your flag weight improves. I'm capped at 750 right now. If I get one invalid flag, it will take 92 of them for me to get back to where I was. This strongly discourages folks with high FW to flag questionable content (at least that's what I think).

Flag weight is now dead, you should decline flags that are not helpful and teach people how to flag in a helpful way.

Granted, since your flags have more weight, you should be more responsible while flagging. But the risk-reward ratio is way too high IMO. In fact, if you have maxed out FW there is no reward at all.

Flag weight is now dead.

Users with close privileges (ie: > 3k rep.) can't flag questions as "doesn't belong here," since that makes them cast a close vote instead. If we encourage users without close privileges to flag questions that don't belong, why don't we encourage users who cast (correct) close votes as well?

A badge for "correct" closing is a whole other kettle of fish, please raise a separate issue on this.

Flags found in /tools/flagged (10k only) are an easy way to "farm" flag weight. What I mean is, it takes much less effort to go through flagged posts and re-flag them than to use /review or flag posts "in the wild".

Sure, and it is also a way to give mods an extra hand, users can add context to the original flag and self-serve.

Lack of a neutral resolution (with the possibility to add a custom message) that doesn't impact the flagger's flag weight. This would be helpful in guiding flaggers without punishing them. Currently, moderators can't provide custom feedback unless they decline a flag.

Flag weight is dead, decline unhelpful flags.

Flag dismissal decisions (eg: helpful or declined) can't be reversed. If a mistake was made, we can't fix it. This results in a ton of flag weight grievance posts on meta.

Flag weight is dead, a mistake you make has only a tiny cost, not worth the effort needed to reverse it ...

---

Also, have I mentioned, flag weight is dead.

Answer 2

First, let me caveat that I have no idea about disputed flags and the like. I have nearly 2000 flags, have had FW 750 since March, and still flag. I don't think I've had a disputed flag, and I've only had one marked invalid since we started reporting valid/invalid counts (I know I had at least one invalid before then, maybe as many as 5.)

There are a number of good reasons to have invalid flags cost 10 flag points. First, since high FW users have their flags at the top, they really should be sure they're right. A mod having a tough day might rely on who the flag is from and act on it, whereas a flag from a stranger might get more checking. Having more "skin in the game" is a good way to ensure that the "names" (I understand less than 90 people out of 350,000 have ever achieved FW 750) are sure before they click the button. Second, it increases the adrenalin factor and might be one of the things that keeps high flaggers flagging. Third, and probably most important, it's simple to code. Why does everything have to be super complicated, special cases, check this first, it matters what order you do it in, don't forget to subtract 3 on Friday? To be fair? FW doesn't mean anything! Fair is just same-for-everyone, not same-all-the-time! Once you drop from 750 you've dropped from 750, whether by 0.1 or by 10. It doesn't matter to me whether it takes 1 flag or 100 to earn it back. Those of us who still flag at 750 are, as far as I can tell, folks who just flag. So what you do to the score doesn't matter much, and simple wins.

Answer 3

1. Agreed. In addition, the entire "Invalid Flag" interface is initially difficult to understand and seems counterproductive. Listing as "Disputed" flags which I've invalid-flagged in my flag weight display doesn't make it at all clear that I am helping anything. (Even trying to talk about these flags is awkward. I had to rewrite that sentence three times.) I also think it's sometimes unclear when a flag should be considered "Invalid" by 10k users. It should be noted, however, that the moderators have occasionally indicated that they find the "Invalid flags" very helpful.

2. I have to disagree with this. I've given my view previously, and I think Brian Reichle's answer to that same question is pretty much irrefutable. Granted that I would have been dismayed to get a flag returned as "Declined" while I was gunning for Marshal recently, but I still think that flag weight is not and should not be an end in itself.

3. Definitely agree (suggested reading: Give an incentive for finding duplicate questions), although in a way, the flag-to-close system serves the same purpose as the suggested edit system: a way to allow newer members to help out while being shepherded so they can learn what is considered appropriate. The only reward for editing after hitting 2k is the satisfaction of a cleaner site; likewise for close voting.

4. I don't really see a problem with this. First of all, moderators have indicated that piling on flags in borderline cases is helpful to them. Second, you've got limited numbers of flags in a day, so even if you're working for that sweet Marshal star and camping out in /Tools, there's only so much farming you can do. Then, once you have the badge, there's really no point to farming.

Answer 4

I think flag-weight as a displayed number is not a good idea at all.

If we want to incentive people cleaning up, use another number like clean-up reputation (better name needed).

You would get points for

• edits (which are not rolled back later)
• close votes which lead to closing (and not reopening later without relevant edits)
• reopen votes which do not lead to closing again
• flags which are considered helpful (but different amounts for "flagging in the wild", "flagging from review page" and "flagging again from the flag queue", according to difficulty of those)
• reviewing anything on the /review tabs
• As moderators: handling flags on the mod flag queue (so mods have also some thing to work for)
• Anything else which the devs consider useful, but does not directly relate to domain knowledge (as reputation does).

The clean-up reputation could be used (instead or in combination to normal reputation) for some of the privileges.