Mark spam works

I've added a feature to mark works as possible spam, and call on other members to vote on them.

Example work: http://www.librarything.com/work/7474425
All proposed spam works: http://www.librarything.com/spam_works.php

As works are voted as spam, I will approved the votes and the works will start vanishing from places where they appear.

Details:

1. Works may be proposed as spam at the bottom of any work's "Editions" page. For now at least, I wanted to hide it.

2. There is a page that lists all open spam-work proposals, here: http://www.librarything.com/spam_works.php

3. The topic is voted on.

4. The ultimate decision to close the vote and effect the results rests with me, at least until I know the system is working well.

There are some wrinkles to the voting:

1. Voting is not private. When you vote, it lists your note. Willfully marking something as spam that isn't, or voting for its removal as spam, is a violation of the Terms of Service.

2. Like tag combination there's a "threshold" number, which is reached when one side is winning by more than four times and fifteen votes.

Thanks:

This feature was a long time coming. I'd like to thank the members who've been reporting spam works, and cordoning them off in various ways—mostly by combining them into aggregated works. To some extent that backfired as the aggregated works are now rather "popular." So this aims to provide a better way.

Thank you so much for this.

One question -- how much diligence is enough, to avoid charges of TOS violations? If I look at the mega-work "Spam 1" that you have proposed as spam, can I trust in your good judgment, that of the spam combiners, and the fact that the top few screens, and what I could see in a quick glance down, all look spammy? If there's a non-spam edition that's been caught up in there, is the proposer, and anyone who votes "yes", running afoul of the TOS?

No, and I can tone that part down. I'm just very wary of people overdoing it here, and marking music as spam, or whatever.

Generally, you can never violate the TOS unless you're doing so willfully.

Do the 195 works in this series listing spam works need to be flagged individually?

http://www.librarything.com/series/spam

Ugh. Yes. And the series needs to be killed.

I think being able to vote from http://www.librarything.com/spam_works.php is a bad idea. The tag voting gets quite a few votes from people who don't look at the tag page, and tags don't have canonical titles that can be set. The two up there right now are bad examples of this; "Spam One" and "Spam Two" just tell us that that someone didn't like them enough to set the Canonical Title. Without going to the editions page, that's useless.

That series (and others, now reduced to that one) is the way people were flagging spam books before, and then putting a placeholder to combine them with other spam works.

Another question: would it make sense to explode the current mega spam works to flag things individually in case there's anything legit there? The problem there is that there are hundreds and thousands of things combined in there. I've done most of spam 2 and can more or less vouch for it; I've tried to remove legit works from spam 1 when I could find them, but I can't guarantee I got them all.

Another question, a book I just listed, Bulgaria Ski Holiday | An affordable Bulgaria Ski Holiday is held by one member, bulgariaholidays, who has been removed. So why didn't the work go with him?

Before deleting, can you check whether the book in question is in any catalogues that don't seem spammy, by whatever criterion? Paid members, at least?

Actually, Spam 1, Spam 2, and Spam 3 are mega-spam works.

Spam 1 was renamed from the previous mega-spam work, which is a real mess, since it has a bunch of things that had their canonical names & authors already changed to spam then combined together, so it ended up with a lot of false positives mixed up there, and it was hard to track back mistakes. I tried to remove legit works that I could find. But there are or were LOTS of books where people ahd changed the canonical title to spam, most of which had been combined into that work.

Spam 2 is a new mega-spam work that I started, and have been using to combine spam works into, to keep them separate from the problems of spam 1.

Spam 3 is another mega-spam that someone else started, mostly for a certain type of book.

Tim, with these mega works, I don't think there's any good alternative to checking the members list before deletion just to make sure there's no legitimate people there, with their works accidentally mixed in (esp. private members).

When these get deleted, will that delete the CK too?

One problem I've noticed with trying to track down and combine spam works before is that there are layers and layers of ck information that got hidden when others had combined the works with other works. I've been removing all CK information before combining them, but I can't "get to" some of the CK information that's in there. Try searching CK for spam, and clicking on works, and you'll see the problem.

Message 4: infiniteletters
Do the 195 works in this series listing spam works need to be flagged individually?

http://www.librarything.com/series/spam
flag abuse Post a message
read Today, 10:54pm (top) Message 5: timspalding
Ugh. Yes. And the series needs to be killed.

--------

One proposal: How about whenever someone flags a book in that "series," that person deletes the information currently in the CK series field? That'll get gradually rid of the series, but let us use it as a way of identifying works to flag in the meantime.

I agree with prosfilaes in #6. Allowing votes directly from the "all proposed works" page doesn't provide enough information, and encourages voters to just run down the list without checking anything.

I started to scroll through Spam 1 to see if anything legit was buried in there. MEGO. When spam works are deleted, will the zeitgeist numbers adjust? Number of unique works and so on?

Before deleting, can you check whether the book in question is in any catalogues that don't seem spammy, by whatever criterion? Paid members, at least?

Well, nothing will be deleted. The world code is a global level. I could delete every work in the system and nobody's catalog would be deleted. But I'm not even doing that. What this will do is remove spam works from Zeitgeist pages, tag pages, author pages and so forth.

I think it's a good idea to take a look if any paid members have any of these works.

It might also be good to send out notices whenever members have a book so tagged.

and encourages voters to just run down the list without checking anything

Well, the previous system was to have members combine or series-mark things without any checks or controls at all—and that mostly worked.

"The world code is a global level." Yes, I really hope so. Please don't delete the world :-)
Did you mean "Work code"?

Thoughts:
•  Make sure votes from removed spammers are removed. (In case the spammers find this, and start voting no on their spam. It doesn't take many No votes to kill a vote.)
•  Most spam works (with a few obvious exceptions) are not combined with any other works. Perhaps disallow marking as spam if there is more than one edition listed? (To avoid accidentally marking as spam a work that has non-spam editions included in it.) I think it would make more sense mark at the book or edition level, but gather that would be quite difficult. I do realize that this would require splitting and re-voting the big spam works, would it be worth it?
•  Since you're allowing voting on the spam_works.php page, could you also list the author of the work?
•  Can you prevent a work that has been flagged and is currently being voted on from being combined with another work? (I'm thinking of ways a spammer may try to derail the process.)
•  When a work is voted to be spam, will LibraryThing staff follow through and check if any reviews are spam, and anyone who listed it in their catalog is a spammer? Or should we be going through and flagging these as well?

Thanks, this should help quite a bit.

This looks good. In the longer term, I wonder if the spam message could be made less prominent on the editions page?

v

#19: The spam message when the book is marked as spam? That strikes me as a bad idea; it should be as loud as possible, so anyone seeing the page knows the book has been suggested as spam.

Again, Tim, is there any reason to keep works whose only copies are held by users removed for spamming?

20: No. I'm talking about the message that gives the opportunity to alert as spam.

v

Again, Tim, is there any reason to keep works whose only copies are held by users removed for spamming?

I can't parse that, could you rephrase?

#22: Bulgaria Ski Holiday | An affordable Bulgaria Ski Holiday http://www.librarything.com/work/6305887 is held by one user, bulgariaholidays. That user has been removed. Once a user has been removed, what's the point in keeping the work around?

23 - The way the system has been set up, these works hung around in the system even once all copies/editions were gone. That's the whole point of this new flagging system, to finally be able to delete them.

False positive:
http://www.librarything.com/work/7640397

This one has been marked as spam, presumably because the author is listed simply as "Kevin," which is also a name that spammers have used for some spam works. If you look at the catalog that holds that book, however, you'll see it's a legitimate member and book. This member has put a lot of books in with simply a first name as the author.

We can't conclude from the fact that an author page has spam on it that all books on that page are spam.

#24: But why? It's somewhat improbable that a work with no holders will be of any use to anyone, whether or not it's spam. Why manually remove a bunch of books after fussing over them, that can be deleted automatically without fuss?

In any case, the work in question does have a copy in the system, one held by the user who has been removed.

25: The voting system Tim introduced seems to be working in the case of this book, no?

v

26 - They *aren't* automatically deleted. That was the problem. Spam works were staying around in the system LONG after the spammers were nuked. (The problem before was that works could not be deleted from the system, even if they had no copies. It's been long discussed in the spam groups. Now, LT has come up with a solution that will actually delete them, more or less.)

Yeah, but it would be nice to have all the works from members who have been purged automatically removed too.

Well, the problem is that purged members—spam, abuse, etc.—don't always have spammy works. For example, we've had spammers add a few legit works in order to pass as legit. Do we delete Harry Potter because a spammer added it to their account?

So, it's a little more complicated.

30> Do we delete Harry Potter because a spammer added it to their account?

Is the spammer the only one with the copy of Harry Potter?

Doesn't seem as complicated as you make it in this example. I'm sure there are others where it is, though.

If you can't make it so only their copies get deleted (not a clue): obviously not.

30>

It's a little more complicated

Only if you deliberately misconstrue the situation. The question is not "Do spammers have copies of this work?", but "Do legitimate members have copies of this work?". If the answer to the latter question is no -- that is, if only spammers have the work -- then it can be construed as a spam work. (It is, remotely, possible that a spammer might add a legitimate but obscure work to their library, such that nobody else on LT has it. This would indeed be a false positive, but one with essentially no cost; the first legit user to add the work would see no difference.)

>25 rsterling:

Speaking of Kevin, why are you and others undecided about http://www.librarything.com/work/1168311 ? The official guidelines say: "What isn't a spam work? Any work that actually exists as a book in the real world." Isn't this sufficiently convincing?

I agree with 31.

If the members have been purged for spam and they're the only ones who have a particular work, then the work should go too.

Tim, is there a way to drop a proposed spam work if you've changed your mind after proposing it?

36> I'm going to guess not, considering similar situations on tag combining.

I wonder if both tag combining and this could benefit from the original proposer being able to click a button that would make a line show something like "Soandso no longer supports this proposal." You can always change your vote, but this might help to indicate that the proposer realized it was a mistake later.

I like the idea of having a (quarantined) archive of spam examples around.

34 - I put undecided because the person who has the book on LT is private, so I couldn't check the catalog to see if it was a real book or not. I hadn't had the time to search the internet to see if a book existed. Will change my vote to no.

Tim - will this eventually remove books from the search results as well? That's one of the places that spam works are really annoying. (Spam authors too... hint, hint!)

If the members have been purged for spam and they're the only ones who have a particular work, then the work should go too.

What if it was a legitimate work? What if someone legitimate adds the book again later? The idea of works is to have a global layer that is separate from the individual catalogs. That is, if everyone with Harry Potter deleted it, the work would continue to exist, with 0 copies, until someone added it again.

Anyway, I think there's every reason to remove books of spammers, and spam books, and to de-increment works with spam books' copies. But user, book, work, etc. are different level of conceptual reality.

41: Okies, remove the books from visibility then. :)

>40 rsterling:

Thanks. I'm going to need to add this one by one where it's needed. I'll add it to search tonight.

41:
If when you remove the user, you remove the book, then the work has 0 copies, right?

Please, please, please hide all 0 copy editions/works wherever they may appear.

Problem solved.

41> Personally, I don't care much either way. But from a theoretical point of view, I'd be more concerned with a book getting deleted if we could actually add books to LT from LT, rather than always having to either enter them manually or go to a third-party source. I know you're working on a way to tackle that so when that comes into the picture I can see how it would be undesirable. But right now, 0 copy works aren't really all that relevant, in my opinion. They're books that no one has, no one reads and the data in LT is probably just as likely to be wrong because it hasn't been vetted like it has for books where many members have a copy.

From my experience, I'd say the main two sources of 0-copy works are
1) Books that were entered incorrectly (either manually, or from an inaccurate source) and were edited, leaving a copy of the original, incorrect, data.
2) spam
Neither of which I would miss if they were hidden forever.

In a distant third place (or probably fourth or fifth) would be real, correct, books that were entered, and then deleted. These may automatically get combined in when someone else enters the book again. Though I don't see too much value in that.

#41: What if it was a legitimate work?

It's vaguely possible, but if you want to have all legitimate books, why not create works for every book the Library of Congress has? Given the design goals of LibraryThing, this seems like an unimportant concern.

Not only that, there's a lot of junk works that aren't spam. Poorly named works that may or not match any real-world edition, but if they do, can't be combined them for the sole reason we don't know what they are. Stuff like Robots by Asimov. Letting them hang around even if their creator has disavowed them is not a gain to anyone.

What if someone legitimate adds the book again later?

A new work will be created. So? I don't see what value the zero member work would have.

It's depressing how many of the spam works that are being voted on have "0 members" already. They don't exist in anyone's catalog, so why are they even here?

I hate to think how much time is being wasted by helpful and conscientious users who are going to each individual work's page to be sure they're spam and then voting on them, when they could just be deleted by the system when the user is deleted.

Ok, it's not a lot of time per person, but how many people have to check each one out and vote on it before it goes away? When it's already been deleted from the catalog of the spamming user that created it?

If it's not in anyone's catalog, just get rid of it. Why are we wasting our time?

>48 r.orrison:

You know, I was being a deligent spam works checker and voter for a while this afternoon, but during that time, developed the same thoughts as you. I became further disheartened in light of another time sink - the author and venue image flagging system. Images are duly investigated, flagged, and voted upon, reach threshold, but aren't ever removed. Sure, they disappear from the "up for a vote" page and go to the "removed" page, but that's a misnomer, because the images are still there on author pages. That's a different system and goal, of course, but the primary issue still comes down to the very word that you ended your post with - time.

#48
I hate to think how much time is being wasted by helpful and conscientious users who are going to each individual work's page to be sure they're spam and then voting on them, when they could just be deleted by the system when the user is deleted.

Amen!

48, etc.;

I think you're missing the point.

Yes, the solution is sub-optimal. But it's the best we're going to get. Despite much pleading to delete spam works, Tim has chosen not to implement any such automated solution -- apparently because he's concerned that very obscure books might get deleted from LT when no users hold them and they have zero copies. I don't agree with his decision, but this is the best we're going to get, and it's a long sight better than what we've had up to now, which is spam works persisting forever and ever, on Zeitgeist and Search and everywhere.

Spammers will get cleverer. They'll probably start posting their stuff as reviews of popular books, next, if they haven't already. But this at least makes the spam-book approach less attractive.

I can't see any reason for books owned by 0 members to continue to clutter up the site.

Often on an author combine/separate page you find a list of 20+ editions when there is only 1 copy in a member's catalog. That can make the author c/s pages ridiculously long & difficult to work with.

I can't see any reason for books owned by 0 members to continue to clutter up the site.

Often on an author combine/separate page you find a list of 20+ editions when there is only 1 copy in a member's catalog. That can make the author c/s pages ridiculously long & difficult to work with.

ETA:

concerned that very obscure books might get deleted from LT when no users hold them

Now I'm really confused. Isn't LT supposed to be for works held (listed) by members, not a source for all works ever written?

51> Yes, the solution is sub-optimal. But it's the best we're going to get. Despite much pleading to delete spam works, Tim has chosen not to implement any such automated solution...

Well, he hasn't said that he absolutely won't ever do it yet, so there's still the chance that he might be persuaded by the unassailable logic of our position.

53>

Now I'm really confused. Isn't LT supposed to be for works held (listed) by members, not a source for all works ever written?

That's what I'd always thought too, right up until Tim said in #41:

The idea of works is to have a global layer that is separate from the individual catalogs. That is, if everyone with Harry Potter deleted it, the work would continue to exist, with 0 copies, until someone added it again

I'm confused right along with you. Well, actually, I just think Tim's wrong on this one. Works are something that emerges from books, that's the only way it makes sense, and zero-copy works should be a nonsensical concept.

1: So Tim, we now seem to have a number of proposed spam works above your threshold ((4*0)+15) but nothing listed as "above". What happens next? There seems to have been a loss of interest.

There seems to have been a loss of interest.

I wonder if this will be a victim of its own success: it will allow people to see how many thousands of spam works there are, but at the same time that might make most people less likely to go through the page and vote.

I wonder if the threshold should be lower for spam works - since Tim's going to check them all anyway (that's quite a job!) - or the time limit for proposals very long.

Are more than 15 people going to flag over 3,000 books (and probably more soon if people don't get competely demotivated)? I don't think so Tim and I think you might consider rethinking the voting system so that it isn't such a waste of many people's time. I also think that a sign of activity/interest from you on this new, wonderful in its peculiar way, yet not fully implemented feature, would be most encouraging to many.

v

No, I agree. Shall we lower it to 5? One paid member?

(I meant 2,000 - can't count)

Thanks for quick reaction. I don't know about the paid member thing but I don't tend to distinguish. Others would probably feel more strongly. The 5 figure sounds OK and could possibly go lower but I think the real brake should be any form of indecision/dissention because it is so rare that it is worth noting, even if it only turns out to be a voter clicking the wrong option. I've noticed that I'm only coming across roughly 1 non-spam or doubtful item out of every 200.

v

I said this before about the tag combination voting, but I still think that the older proposals should come first. I think it would prevent the proposals with fewer votes than necessary getting "lost" on page 20. It would take longer for new proposals to make their way to the front, yes, but I think it would be more likely that all proposals would get the votes they need.

61: agreed

v

Changing the vote required for removal would be good (five uncontested votes sounds about right), but even better would be to automatically remove any works that were only owned by members that were removed for spamming. Most of the works I'm voting on are at 0 members, or 1 member that has been removed (e.g. http://www.librarything.com/work/6854495).

I know the guidelines say that if someone has included non-book items in their account those items are not considered spam, but are non-book items spam if they include a web address (presumably so that friends/family members know where to buy the wishlist item)?

http://www.librarything.com/catalog/wildfell_wishlist&tag=notabook&colle...

So, I did an author search for "http" and came up with 2078 results. Out of the 1600 I have checked so far, only 4 or so* have turned out to be legitimate. Over 1000 have had one or more spam works attached to them.

*One of those I accidentally flagged before I realized the member who entered it was "real."

>59 timspalding:: To me, one paid member sounds a little low. Anyone can make a mistake. Five voters total with no dissenting votes sounds pretty good though. I think that 15+ is a bit much to ask. Most of these suckers are extremely obviously spam and there are a *lot* of them for voters to wade through.

I would also like to register agreement with post #61 about older posts coming first in the queue and with post #63 about automatically removing works only owned by a single member who has already been bounced for spamming. I'd say 90+% of the spam works would be taken care of by this single step.

66: I've made a couple of mistakes myself, but then changed my vote. It would be nice if the proposing member could drop the proposal in that case (if no one else has voted on it).

Five "yes" votes with no dissensions sounds good to me. I do wonder, though, if some of the spammers might find out what's going on and counter-vote. Guess we can cross that bridge when we get to it.

I like 66's other suggestions as well. And automatically removing "0 owner" items would help a lot.

eta: Tim, when are you going to start removing them? Lots have over 20 "yes" votes at this point.

Is it possible to get a list that includes only proposed spam works that you have not yet voted upon?

68: Yes, on the Spam Works page, on the "Voting" line, click on "Unvoted".

69: Thank you.

Has anyone noticed the glorious absence of spam works in the "top reviewed" on the zeitgeist?

(They're still sticking around in "highest rated" though.)

How delightful to see the number rise for "Closed." Does that mean they are gone?

I'm working on it right now. So, don't correct/praise until I write again.

Okay, so:

1. Closing is working
2. Voting is now > 3 votes or more than three times yes to no.
3. Search results omit spam.
4. Spam omitted from Zeitgeist, but there is some caching.
5. Spam omitted from tag pages.
6. Spam work pages look like this http://www.librarything.com/work/9709572

More to do, of course.

Wonderful!

I've woken up this morning and it's Christmas:
1) The sun is shining when the weather forecast was bad
2) A spam miracle has occurred
3) My first ever attempt at getting an ER book in all the time I've been on LT has hit gold

v

#74 A glorious new dawn for LT.

There's got to be a more appropriate word than 'suppressed' though. That word, used about a work on a bibliophile website... think of the Googlebots.

I just don't picture myself in the 'shaking of pitchforks and waving of burning torches' group when helping to remove spam from LT.

Love the new measures :)

Since they've been "suppressed" (or at least some of them), I assume that the answer to my question in #64 is yes?

Beautiful! Thanks Tim.

Were there any false positives, or spam books in legitimate members' catalogs?

ETA Like this one: http://www.librarything.com/work/1168311&spam=1
How do we mark something that's already been suppressed as "not spam"? Weren't there going to be some checks to see if actual members had the book?

81> I agree that work is not spam, although a cursory look would have been deceiving. Just goes to show no system will be perfect, even if it's crowd-sourced.

Do we get a petition for reversal?

57>...but at the same time that might make most people less likely to go through the page and vote.

That's exactly what happened with me. I saw how many spam works there are (and perhaps not even all are tagged yet). The sheer number of spam works has scared me off from voting. Well, that and the time factor. It would take me WEEKS to thoroughly check everything...

83: It's now between 20-60, so much easier to look at. :)

Edit: Make that 160 or so. It's going up again...

81: This is a perfect example of why I said in 60 that ANY book which has even a single vote 'no' or 'undecided' should be examined manually by Tim - the proportion is so small that it is no major trial. I know I and others voted against this as spam and it even came up earlier in discussion on this thread so it should never have gone through as spam. I have no idea what the vote totals ended up for it.

v

Current tally: Yes 3, No 28, Undecided 6

So why did it get suppressed? Even Tim voted no on it.

Summat wrong with t'system there ;)

v

85: Yes, I agree.

Yeah, agreed. Will re-calculate later.

Glad that will get fixed.

What happens, though, when well-meaning people make a mistake, and vote something as spam that isn't? Is there any way to switch it back? Will proposing a new vote on it - if the nays have it the second time around - make it no longer suppressed? Is there some other way to indicate mistakes and false positives, once they're "closed"?

Fixed. Thanks for the heads up.

I hope it's ok, but I started a thread in the Spam Fighters group for discussion of spam work proposals (e.g. borderline or unclear cases, etc.)
http://www.librarything.com/topic/90187

The Highest Rated on Zeitgeist shows the heading, but no books. Is that related?

There's something a bit odd happening: the occasional spam work which I know I've already marked as spam previously is coming back to the vote. It's as if somewhere along the line they've got unmarked as spam or undeleted and have been picked up by members again as spam. They are not items where people have voted no/undecided. Some I simply remember because they are so distinctive like this: http://www.librarything.com/work/9106365, otherwise I know they are repeats because the text is purple, ie I've already opened that link in the recent past, like this one: http://www.librarything.com/work/9518160

It's not that I've opened them and forgotten to vote because I voted on them all until that section was empty and there were no more to vote on.

I wonder if Tim has time to zap another lot that is over the threshold. It would take us to over 5,000 dead spam works which would be satisfying.

v

>94 vaneska:

Both of the users referenced above are multiple offenders. It would be good to remove these members, all of their "works", and all associated "authors".

#95

One of them has been spamming since Jul 2, 2009, the other since Sep 5, 2009. How much is too much?

I've found worse/older cases. Just flag their profiles and the works, and they'll get removed.

One of them hasn't added any spam works since October of last year. The other one added one last summer, then came back for more this Jan and Feb.

As for why the works aren't already/still flagged... maybe it's a bug. Maybe the first time you went through the flag didn't stick or you didn't go all the way through (I know sometimes when I've gone through a bunch at a time, I click on the "editions" page but forget to flag one or two). Maybe it got no votes (unlikely). Maybe it got separated from another work. If it's a bug, it should definitely get fixed.

A few points on spam works:

Some ways I've found helpful to find spam:
-- search for tags on certain topics (e.g. life insurance quotes)
-- search for works with certain words (seo, make money online)

Things to watch out for -- false positive:
-- it's a good idea to check out the member. If it's a legitimate member and the other books in the catalog are fine, it's probably not spam.
-- if there's a web address in the title, author, or elsewhere, that doesn't necessarily mean it's spam. It could be an online resource, an online book, an online article, or something else. Unless it's commercial spam (see the guidelines), it shouldn't be spam.
-- Non-book works are not spam. See the guidelines.
-- There are real books that might look a lot like spam but aren't, e.g. books on topics that are often the subject of spam (online marketing, weight loss, insurance, colon cleansing, etc.). Be careful not to get real books caught up in the net when flagging based on key words. It can help to search Amazon or similar sites to make sure it's not a real book, especially if the member looks legit, etc.

I've noticed that the spam threads are pretty dead these days, which seems like a good sign: that we have ways of dealing with and removing spam more efficiently by ourselves now!

Tim, earlier you mentioned that flagging spam works would delete them, sort of, or at least hide them, from various places on the site:

What this will do is remove spam works from Zeitgeist pages, tag pages, author pages and so forth.

I think this is working in some places but not others, and for some works and not others.

In particular, works do not seem to be disappearing from authors' pages at all. On tag pages, some works are disappearing, and others are not.
Example with works invisible:
http://www.librarything.com/tag/start+seo+training
Example where works are still visible, despite flagging and suppression:
http://www.librarything.com/tag/antispyware

re #99:

I was wondering about this myself. There is a Spam work that doesn't want to disappear from the author page "John (http://www.librarything.com/author/john)

http://www.librarything.com/work/9864460

One way to really make this useful it really easy to police & find spam works would be to have a page of "new books added to LT". I'm thinking of Wikipedia's "recent changes" list, which is how people police spam & vandalism there; here, I think, it's just by happening upon something, or else by actively searching.

or am I missing something?

In my dream, it works like this:
- visit "new books added to LT" which is a reverse chronological listing of books / member-added / date-added. This could be available under Zeitgeist.
- there is also a "mark as Not Spam" option on book pages (available only to named members with catalogs and/or long-term active memberships and/or other activity) which
- the "new books added to LT" has an option to turn on "spam policing mode" which highlights the "mark as spam" on the book page, and deprecates items marked as not-spam more than 5 times ... and maybe shows an abbreviated version of the book data on the "New books added to LT" page.

>101 lquilter:
There's a recently added books module for the Home page.

102 - That moves too fast, though.

I agree it would be useful to have a Zeitgeist of recently added books, with a decently long list.

"Recently added works" would be what would really be useful; that would point out both spam and orphans in need of combination, rather than just the N+1st copy of something.

>104 lorax:
That would be great.

104: occasionally I've suggested something like this for tracking favorite authors.

106>

It would work less well for that (most of what it turned up would, in fact, be orphans in need of combinations), but I suppose there would be a few legitimate new works by popular authors -- somebody has to be the first to wishlist a rumored new novel, after all!

Time for Tim to have another killing session: there's over 3000 above the threshold. I do wonder whether the current figure of something over 8000 spam works is the tip of the iceberg, or a sizeable chunk.

v

Killed.

There's an issue where if someone flagged something as a spam work, and that flagged work gets combined with another work, the spam notice is no longer displayed on the work page. You can see it on the list of spam work proposals, but since there's so many, it can be hard to find an individual proposal there.

Is there any way to change this? Or, alternatively, should we have a moratorium on combining works that are flagged as spam until after their vote is done? (There's a problem of enforcement there, though.)

I've seen this happen both with spam works, where several had gotten combined together (or perhaps were automatically so because of similarities in the name?), and with legitimate works, where someone had flagged a legit work as spam, and someone later combined it with the rest of that legit work.

I see this is as a philosophical problem more than anything. What should be the desired result of combining a spam work with a non-spam work? Why would someone do that? What are they trying to accomplish?

My worry is that someone will thoughtlessly combine a spam work into a very much NOT spam work and doom it.

I think answer number one is to remove spam works from suggestions to combine. Anyone seen one I can use?

Note: The basic problem here is that the "I am spam" field is not used universally where I list or show works. I need to add the code "AND spam 1" (or whatever) wherever that's needed. So I need pointers on where spam works are showing up that might be a problem.

See Message 35, 37, and 38 in the discussion thread.

A work with one copy was incorrectly suggested as spam, I voted no then combined it in with the work where it belonged.

What should have happened? I don't know. In this particular case, I think it was right for it to disappear from the work's page, because no-one would have suggested that the main work was spam. On the other hand, it makes sense for it to remain on the spam works list, so that spammers can't hide things by combining them with legitimate works.

So maybe it worked right...

so that spammers can't hide things by combining them with legitimate works

I don't see that happening. It's too complicated. Spammers show an incredible inability to understand what LibraryThing is, let alone how it works.

109: Hmmmm.... What else can I get Tim to kill? :evil:

v

In Message 111 Tim said: I need to add the code "AND spam 1" (or whatever) wherever that's needed. So I need pointers on where spam works are showing up that might be a problem.

Now that the CK clouds have updated, "SPAM!! SPAM!! SPAM!!" is appearing as a popular series, though it only appears to contain hidden spam works. (You can see the entries on this list: http://www.librarything.com/commonknowledge/search.php?q=spam&f=23)

Well, in fairness I only see it in one cloud...

How many clouds were you expecting to see a series in?

(hides head)

Something I've noticed, by making a mistake: If there's something that's already been flagged as spam, and I carelessly click "flag as spam" instead of clicking on "vote yes", it puts the work on the spam voting page again, with 0 total votes, and says that I proposed it. The message and voting total on the work page itself is unchanged.

Maybe this isn't enough of a problem to do anything about, but I thought I'd mention it. It could also be one source of the repeated suggestions that someone else noted.

Thanks. Minor but annoying, I agree.

Why did this work get suppressed? There were more "no" than "yes" votes:
http://www.librarything.com/work/9542649

MikeBriggs has proposed marking BICB 0 Copies as spam.
Vote: Yes | No | Undecided Current tally: Yes 1, No 5 (threshold met)

Yes: danielx

No: lorax, fleela, mihess, cbl_tn, rybie2

Are there any other bugs like that in the recent closed batch: other ones where the consensus was "not spam" but that got suppressed? How do we undo something that was wrongly marked and suppressed as spam?

Let me look at it. Don't separate editions out or anything.

122 - I'm finding other cases as well, where the votes were either overwhelmingly "no" or were at least not overwelmingly "yes," but where when they were closed, they were suppressed as spam. One thing all these have in common is "undecided" votes; other works with only "no" votes but no undecided votes reverted back to normal works and didn't get suppressed. Maybe there's something gumming up the code there, with undecided votes? I think this may only be for the most recent round of "closed" votes, but I don't have any way of knowing for use. ETA: hmm, but the case I mentioned above in 121 doesn't have undecided votes; there goes that theory...

http://www.librarything.com/work/9239190
bergs47 has proposed marking 50 Blogs - Sex, Love, Dating and Life as spam.
Vote: Yes | No | Undecided Current tally: Yes 0, No 4, Undecided 1 (threshold met)
No: mihess, StormRaven, rankamateur, bergs47
Undecided: fleela

http://www.librarything.com/work/2899679
rybie2 has proposed marking The Poetical Works of Mark Akenside: With Memoir and Critical Dissertation. as spam.
Vote: Yes | No | Undecided Current tally: Yes 1, No 5, Undecided 2 (threshold met)
Yes: ty1997
No: bw42, danielx, cbl_tn, bergs47, rybie2
Undecided: JulesJones, bibliorex
rybie2 has proposed marking The Poetical Works of Mark Akenside: With Memoir and Critical Dissertation. as spam.
Vote: Yes | No | Undecided Current tally: Yes 1, No 5, Undecided 2 (threshold met)
Yes: ty1997
No: oregonobsessionz, danielx, cbl_tn, bergs47, rybie2
Undecided: JulesJones, bibliorex
(that one's in the votes twice)

http://www.librarything.com/work/181067
bergs47 has proposed marking Great gifts to make and wrap as spam.
Vote: Yes | No | Undecided Current tally: Yes 1, No 5, Undecided 1 (threshold met)
Yes: danielx
No: JulesJones, oregonobsessionz, cbl_tn, bergs47, rybie2
Undecided: ty1997

http://www.librarything.com/work/5892488
bergs47 has proposed marking Bones Season 3 Episode 1 - The Widow`s Son In The Windshield as spam.
Vote: Yes | No | Undecided Current tally: Yes 4, No 0, Undecided 3 (threshold met)
Yes: JulesJones, bibliorex, danielx, rybie2
Undecided: oregonobsessionz, ty1997, cbl_tn
(this one may well be spam, but the votes weren't definitive)

http://www.librarything.com/work/9997269
Barka has proposed marking Dolce vita Magyarországon as spam.
Vote: Yes | No | Undecided Current tally: Yes 1, No 6, Undecided 1 You voted: No (Cancel) (threshold met)
Yes: Barka
No: sorsha, JulesJones, rsterling, ty1997, cbl_tn, rybie2
Undecided: danielx

121-123: Any luck figuring out what's going on with these false positives? It would be good to fix the bug before any other votes are "closed."

Got it. Two categories of voting were being counted together for this purpose. I redid all the "results" data. All fixed now, and another batch done.

What is the best way to unflag a work that looks as though it may have been improperly suppressed?

Should I try reflagging it but voting no?

On the way home: West Virginia soldiers on the Sultana seems (as far as I can tell from a bit of Googling) to be a work with a legitimate ISBN and only legitimate members holding it. I don't see any obvious reason why it ever got suppressed.

NOTE: Even though it's flagged as spam, it turns up on the recommendations lists for other works (the generated lists from the other works, not anything that required human intervention to recommend it). So apparently the spam flag isn't suppressing it from everything.

I'm no longer flagging spam until the issue of improperly suppressed books is sorted out.

v

126: Wow. Good catch on that work. It's definitely not a work of spam. Boy, I sure hope I didn't accidentally vote to suppress it as spam whenever it got flagged, I'd feel pretty bad about that. It's an obscure book, but it's real.

I've found several case like this. I'm not sure if there's a bug in the system somewhere, or if they were really voted (wrongly) as spam.

I started a thread for cases like these here:
http://www.librarything.com/topic/93339

126 - Ok, I've looked through ALL the closed spam votes - yes, all of them - and I cannot find that book (On the way home: West Virginia soldiers on the Sultana).

So either there is or has been some funny combination/separation stuff going on, or there's a bug that's catching legitimate and non-voted works somehow and marking them spam. Or I missed it. But I don't think so.

Tim, can you say more about the back-end (non-vote-related) processes for suppressing spam? I ask because there are a bunch of things getting suppressed as spam that were not voted on (I checked), and that are in real members' libraries. I know that there's now a nightly process (as discussed here) that suppresses books only held by deleted members. Is it possible that something's going wrong in the code there or somewhere else? How's it possible that things held in legitimate members catalogs (and in no spam catalogs) are getting caught up and suppressed as spam?

In a few cases, the authors of these legitimate works (though sometimes non-book works) had been combined with the Evil Spammer author. (That was the case for the author of these works among others: 9121059, 5646976.) Is the system automatically suppressing anything attached to the Evil Spam author? I'm not sure that's a good idea, because sometimes people make mistaken combinations.

In other cases, though, the works suppressed don't even seem to have any connection to any spam author or anything about them that could have even been mistaken for spam. For instance here: http://www.librarything.com/work/907277&spam=1

What's going on here?

I believe Tim is at the ALA conference now and he may not be keeping up with Talk as much. A direct e-mail or post to his profile might work better.

I suspected as much, but he's probably also not checking profile comments now either. I'll give it a few days, and if there's no follow-up, I'll send a comment then.

ETA: Actually, why wait. I just posted a profile message for Tim.

Getting to it soon. Sorry people.

User http://www.librarything.com/profile/yogaequipmentstore is still around, and just added another spam work.

This is a knotty one:

http://www.librarything.com/work/9548596

It's clearly a spam work, but it's easy to imagine someone writing a book with that title - though nobody's listed one so far - and a member subsequently running into problems when trying to catalogue it.

This is what voting is for, of course, and I've flagged it.

I've stopped voting on spam works. Inevitably, the 'author' is already combine with Evil Spammer, rendering my opinion moot as the decision has been made for me by the combiner.

>136 alaudacorax: it's easy to imagine someone writing a book with that title - though nobody's listed one so far - and a member subsequently running into problems when trying to catalogue it

A real book would have a different author, so that wouldn't be a problem.

#137 Ah. That's never occurred to me before. For what it's worth, I, personally, only combine obvious spam authors - 'stuffedbobcats4sale.com' or such. Any doubt at all and I don't combine it. Also, a large proportion - perhaps a majority - don't have authors' names attached.

#137 Thinking over it some more, I think you're right. I think I should be going back after the vote to do the combining, if only for courtesy to voters.

137 Author combining is easy to undo, though. I've separated many authors out post facto. I don't see that as a barrier to voting. For me what matters more is the catalog it's in, not the author. People can put spam works in with real authors' names, so the author in itself is not very decisive. I do think that we shouldn't combine real names with evil spammer, though.

136 The issue with things that could share a name with a real book is trickier. My impression is that most people voting don't investigate further or look at the spam proposal discussion thread, but instead just rely on the proposer's judgment and vote yes if it appears at all spammy. It's been hard to overcome cases like these (or false positive cases) with "no" votes, and the proposal doesn't seem to go away on works with split votes.

Quick query - I see an alarming amount of works called sound recording (with the square brackets attached - there are about 16,000 works listed) that mostly list CDs of music (just do the search in the search box, then click on 'works'). One of the first items is Björk's music. Are music CDs allowed, then? I mean, the guidelines for spam works clearly states that anything not pushing members to buy or go to dodgy websites is not spam. So they do not fall into this category. BUT this is not a case of audio books or videos of poetry and literary readings I am talking about! :-)

Anything that is not listed with intent to sell or drive members to a commercial website that the lister benefits from is OK.

So music CDs, perfumes, dresses, sports equipment, lists of URLs so long as they aren't obviously and inarguably spam are okay. The definition of spam is quite different here than almost anywhere else; if anyone anywhere could imaginably list something for other than spammy purposes, it's allowed.

Music cds, video recordings (dvds, etc.) are allowed. Perhaps not encouraged, nor is LT ideally designed for listing them, but many of us have included them in our catalogues. They are not spam.

Ok then, I'll leave them be!

@soniaandree,

BTW, the fact that the owner added "sound recording" is a very good thing; it will prevent the records being combined with similarly named book records.

I hasten to add that I don't agree with the anything-goes attitude here, but Tim has specifically said that's how he wants it, so that's how it is.

By the way, there is some spam I have put to submission to voting members, and I think that people might be confused, especially since it is in *French* - the title clearly indicates that the owner of old coins (i.e. 'monnaies') wants to 'sell' (as in 'vente'/in English 'sale') to people interested in 'bidding'= 'offres' (just look for works including 'vente sur offres', you'll see what I mean).

What's the status when reporting spam in other languages, then? In this case, I am certain LT is being used by the member for advertising the fact that he is selling parts of his coins collection, which, I assume, is not ok under the terms and conditions.

No. A long time member with over 2000 books has a few French auction catalogues of coins. Nothing wrong with that. Or would you also consider this to be spam? http://www.librarything.com/work/323835/book/6127116 Googling the title would have shown that such catalogues really exists, and that the ones listed are for sales that happened some time ago.

Assuming this is to do with auctions (i.e. he should have written 'vente aux enchères' in French, and not 'vente sur offres', which is something else entirely, about sale and bids), he should write the title with those correct *auction* words and indicate the year of the auction, as most auction catalogues do. This is why I haven't included Sotheby's catalogues or other ones. Those were legit.

He uses either the wrong French words or the correct ones for what he wants, but, like most spam, I tend to go with what I read and check, as double or triple-guessing the owner's French is going beyond what is required. And anyone can be spammy, this is not relevant to how long we've been in LT or the rest of one's library.

Anyway, most people voted that this was not spam and, after all, vox populi rules in such matters! :-)

I think what MarthaJeanne was saying was that there is an actual catalog, with the words that are in the title. That makes it a book, not spam. As you've seen with Sotheby's, many of us collect auction catalogs for one reason or another.

You can check whether it is a legitimate auction catalog by looking the title up on large bibliographic databases, such as WorldCat.

If there is an actual catalogue, then the year should be entered. The French words are misleading.

He used the words on the cover of the book as the title. He entered the number of the catalogue from the cover in the title. He entered the year in the publication data. He tagged it as an auction catalogue.

It is possible that the catalogue should have had a different title, but he can hardly be blamed for using the title that is on the book.

https://www.librarything.com/work/16702777/reviews/122967094

http://www.sunrisepool.com
Sunrise Pool Services is a swimming pool cleaning company in Atlanta for over 30 years. We have pool maintenance professionals for residential and commercial. Call us today!

https://www.librarything.com/profile/sunrispool

>154 PitcherBooks:
This is a dormant thread. You can report spam here: https://www.librarything.com/topic/200802

>154 PitcherBooks:
For a minute there, I thought someone spammed the new features thread. LOL

>156 gilroy:

Yeah, I thought it was spam rather than spam reporting too.

I think we need a Max Headroom-like mascot called Mark Spamworks...

Is there a way to undo marking something as spam? Some combination of browser lag and an oversensitive touchpad resulted in me accidentally clicking the spam link for The Winds of Folly and there's no obvious way to fix it, other than asking people to vote down my mis-click.

>159 fyrefly98: No, but if you post in the Spam Fighters! group, people will vote it down real quick. It happens sometimes.
https://www.librarything.com/groups/spamfighters