So Puzzling.SE seems to be the only site (that I've noticed) that has had a continuous pattern in the usernames of the spammers (that I blacklisted a while ago). Look at this. Is there any other site where such a pattern has been noticed?
@JF I'm think it might be better to have two watches, one specifically for the username, ^filmabox$, and one for the domain filmabox\.net. The primary reason would be so that people looking at them in the future know exactly what was intended to be caught, particularly because username searches are separate from body/title searches on MS, as are blacklisted usernames.
@SmokeDetector Please see: Are job offers spam?. What's discussed in that Meta post is more inclusive than just job offers for payment. The example given doesn't show an actual offer of money, just the solicitation of having someone work on the project. IMO, the differentiation is that the Q is primarily looking for a person to help off-site, rather than looking for an answer to an on-site question. IMO, the original version of this Q clearly qualifies.
There's a lot of potential grey area here. I feel we should be careful; judging, as usual, on a case-by-case basis, but this question was clearly looking for bodies to help off-site (even offering payment).
@EriktheOutgolfer More people flagging, is always beneficial. While I can understand being more comfortable staying with the sites where you're more comfortable, as you say, spam is spam. Spam is often very clearly spam. OTOH, there are, of course, a considerable number of edge cases. More of them are reported into this room, as it receives those posts detected by the experimental reasons.
@EriktheOutgolfer While signing up for autoflagging is helpful, it's not necessary. It's certainly possible, and probably more helpful, to manually flag posts, when you feel they're spam or rude/abusive. For autoflagging, you can set the criteria that's used for your flags. By default, the criteria are that there is a very low chance that the post is not spam. You choose to set them such that no post that we've ever encountered which human feedback has indicated wasn't spam, would be flagged.
@EriktheOutgolfer Some people do. In addition, there's a reasonable number of edge cases where the post could be considered spam, or might just be NAA. For those cases, it does take human judgement and usually remembering to assume good faith.
Yep, that's your Stack Exchange global ID. A different, much lower, number is created for you on metasmoke, which is your metasmoke ID (similar to having a different user ID on each Stack Exchange site).
Well, it depends on which privileges you want. For some, it's being able to write to the SmokeDetector GitHub repository. For other's it needs changes on metasmoke (i.e. an admin).
@EriktheOutgolfer That's the minimum total weight which the post must have prior to it qualifying to be flagged by that condition. Each detection reason caries a "weight", which is the accuracy % * 100 (i.e. between 0 and 100). Each reason which detects the post has it's weight added to the post's total weight.
@EriktheOutgolfer Well, for autoflagging, you should need no additional privileges. If you want to give feedback to SmokeDetector (SD), then that's given on a chat room by chat room basis (i.e. I'm not sure if there's one specific to PCG) & requires writing to the SD GitHub repository. I can create a PR for this room, but we have to wait for it to be merged. Other rooms may have specific requirements (some require approval of a RO for that room).
If you want to use the userscripts (very helpful, if you're going to be watching chat and giving feedback), you need the Reviewer role, which requires a metasmoke (MS) admin.
@JF I am aware of 5 admins: tripleee, angussidney, Andy, ArtOfCode, Undo. None of them are currently in this room. Other users in this room might be able to help you.
@EriktheOutgolfer You have to set the flagging condition to something that yields a certain percentage (click Preview). The default is weight >= 280, user reputation <= 1, and reasons >= 1 (although, weight of 280 effectively makes this 3, because any single reason can't have more than 100 weight). Historically there are 43,545 posts that qualify which are spam (True Positive, TP) and 1 which is not spam (False Positive, FP), which is 99.998%. You can increase the weight to get to 100% (weight >= 300).
@EriktheOutgolfer Then the users who autoflagged the post are @ pinged in here, and in other rooms. They can then evaluate the post and determine if they want to retract their flag. Unfortunately, retracting a flag is one of the things that can't be done through the SE API, so it's not possible to have metasmoke automatically retract the flags for people.
Yes. You do risk having a declined flag. However, that's really not that bad of a thing, unless you have already had issues with flag bans on some site.
Do be aware, that there are enough people participating in autoflagging such that your account being used for flagging will be relatively rare. I don't have actual numbers, but a person or two has complained that the number of flags that are cast with their account is too low. IMO, those people are in it to get their flagging numbers up rather than to help out. If you participate in providing feedback/flagging in the room, then the number of flags cast manually can be significant.
@EriktheOutgolfer There are those that claim that. :-(
Please read car rental tips HERE.
