DISQUS

Scobleizer · 7 months ago

Wow, now I know why I've been seeing tons of "bot like" accounts being created on Twitter. Just look at TechCrunch's followers. They are NOT human. I bet there are people creating bot networks to just spam your @replies. Really nasty stuff. This is going to be a lot harder to block than you think. They can just keep reusing the same accounts for a few times until Twitter figures it out. Now I am even more happy I've switched almost my entire life over to friendfeed. You should join me there, the DM feature there is a real pleasure to use compared to Twitter's and the search engine is a lot better too.

Scobleizer · 7 months ago

Why is friendfeed more resistant to spam and jerks? For one, if I block you you disappear EVERYWHERE on friendfeed, even on search or DMs. For two, if I start a node I can delete any comment underneath that node. So, the crowd can keep the community clean of spam (friendfeed has very nice decentralized moderation). On Twitter no one can delete anyone else's stuff, even if it's obviously spam or nasty stuff.

This will end up being a competitive advantage for friendfeed and it will prove to be difficult for third-party applications to get rid of this stuff in time.

Mihai Secasiu · 7 months ago

Sure you can delete any comment, but if you'll have hundreds of spam comments under a node how will you do that?
Also anyone deleting comments on a node created by someone else doesn't seem normal. It could be abused.
Friendfeed is better only because it doesn't make ( financial ) sense for spammers to be there at the moment, just like it didn't make sense for them to be on twitter until now ( or a few months ago )
If friendfeed doesn't use the time it still has to improve on this it will be in the same place as twitter

Scobleizer · 7 months ago

Mihai: I delete tons of spam on my blog. I can keep up with a very high level, believe me. And it's far easier to delete spam on friendfeed than it is on my blog (it takes fewer clicks). Plus, my blog gets a lot of spam because it's easy to post a comment anonymously. That's impossible to do on friendfeed (you need a verified email account).

You can't delete comments on someone else's node, but I can delete comments underneath MY node (and you can delete comments under yours). This actually works very effectively because most of us are self interested and want to make sure we present a clean room, which is why I delete spam off of my blog.

If you actually used friendfeed (your statements here make me believe you don't) you'd see that friendfeed is 100x more resistant to spam than Twitter is.

Mihai Secasiu · 7 months ago

sure you can keep up with tons of spam if you want but not everyone is willing to do that.
Doesn't twitter also require a verified email address ? Anyway even if it does I don't think that's working. Why? because there are bots out there that can create thousands of email account on free email sites just like they can create as many twitter or friendfeed accounts with these email addresses.

Are you saying friendfeed is more resistant just because you can delete comments on your node? You can do that on the blog too but that doesn't solve the spam problem. How many spam comments do you get in a day? I bet you don't get many , but that's not because they know you can delete them that's most likely because you use some antispam system like akisment that identifies spam and throws it away. So maybe what you see is just what akismet misses? Does friendfeed have that ?

I use friendfeed too , of course not as much as you do, and I like it but I think you're overestimating it's capabilities.
You get 200 comments on some of your nodes on FF and those are all from real people, what will you do when ff reaches 25 million users and spammers start to get in? can you still keep up? good luck with that!

Nick Toumpelis · 8 months ago

Will try to implement a solution in my Twitter client along the lines of dm-ing the Twitter spam account.

Jeff Harbert · 8 months ago

A 'report as spam' feature would be great, but what I very much want keyword and @username blocking ability. The recent @aplusk and @oprah ridiculousness drove me nuts. Being able to block those names (and other words) from appearing anywhere in my Twitter feed - not from any particular user, but being mentioned by people I follow - would make me very happy.

Loic Lemeur · 8 months ago

Jeff, that's another interesting one, we can also do it at the client level.

Jeff Harbert · 8 months ago

That would be great, Loic. Thanks!

Susan Beebe · 8 months ago

please add @TweetStack to your list of collaborators :)

norsehorse · 8 months ago

this definitely would be a welcome feature Loic. in fact, I've already been spammed this manner at least twice thus far and it really peeves me off, particularly due to there being no readily available as well as easy to use recourse to address such abuses.

webjay · 8 months ago

Track http://search.twitter.com/search?q=to:spam and crowdsource your way out of the problem.
I.e. if X have marked a tweet as spam, your clients should put it in a spam folder.

Jalada · 8 months ago

Pretty simple to do, just @spam or d spam. I could probably integrate that into Twitterfall too :-)

RBL · 8 months ago

Great idea.

@Jeff Harbert, I agree also need a BLOCK button. Twitter already supports blocking. Just surface the feature in Twhirl and Desktop. Also like your suggestion for KEYWORD blocking. That would be really useful.

I also think color-coding @ replies would help. Green = they're a friend, blue = they're a "good" follower but not a friend, red = they're a follower and they just signed up and/or have other attributes common to spammers.

Another idea: When trying to @ reply someone, allow users to optionally require captcha on first @ message, or all @ messages.

This would be minimally intrusive. If a user required captcha on first @ reply, they could then mark as spam if it ends up being a spammer (unlikely anyway, right?). Entering the captcha code validates the user as not a bot.

If a user is paranoid, they could require captcha on all @ replies. Frankly, I see this as a necessary feature anyway. Right now, with Twitter, it's all or nothing. Either anyone can @reply to you and follow you, or your updates are blocked. The service does need a middle ground.

Loic Lemeur · 8 months ago

Good ideas RBL

Fred H Schlegel · 7 months ago

I like the color coding idea. Has use beyond spam.

sumyunguy · 8 months ago

This method could be easily abused. Imagine if all of X desktop client users decided to report @loic or @askseesmic as an abusive user. Or what is Ashton decided he didn't like someone as as a joke 1Million twitterers reported someone as spam!

Why should twitter be the arbiter of what is and isn't spam?

Let us setup keyword user blocking on our own and let twitter handle it how they want.

Loic Lemeur · 8 months ago

of course it can be easily abused, I am starting a discussion here, not
saying it's perfect, we just need a solution

Pete Austin · 8 months ago

In the email world, managing a spam blocklist is a lot of work because you need human admins to correct mistakes. There will be similar issues for Twitter. Also I expect lots of political/religious/4chan campaigns to get controversial users banned.

Kristi C. · 8 months ago

Loic, I love the initiative. I have an app about to come out called Twitterface and we will do whatever we can get Twitter to agree to on this issue. For power users of over 2000 people, this is an especial hindrance to ENJOYMENT of the tool and I have wasted tons of time dealing with spamming creatures over the last year. We need the equivalent of "rules" like we all have in email so we can repel unwanted advances. Follow management is one of Twitter's most urgently needed feature additions, I think, but I have no idea if they are doing anything to address that problem or not. If you can lead a charge to get something done, Twitterface will gladly jump on board.

MayankDhingra · 8 months ago

Loic

Honestly speaking, how much time do you think will Twitter take to add, report as spam button ?

Gary Rust · 7 months ago

Spammers, Trolls, Idiots and Thieves. Sounds like a Clash song. It's just crazy to think how much people go through in not learning the basics of social media. I guess it's like the guy who farts in church and thinks nothing is wrong? It's like my dad said to me 1000 over the years, its all past of growing up. Some people get it and some people don't!

AppHacker · 7 months ago

I created a twitter spam blocker, which still needs a lot of work, but it's available here: http://gtfokthxbi.appspot.com/ It doesn't help with reply spam, but it helps with follow spam. If you have a chance please let me know what you think.

Ryan Graves · 7 months ago

You knew this would happen. The spammers will always be ahead of the game and it's up to apps like Seesmic desktop and companies like Twitter to stay ahead of the curve and preserve the user experience.

I look forward to seeing (and supporting) what you guys come up with!
Best of luck.

Jesse Stay · 7 months ago

Loic, if you implement a report spam feature, let's coordinate along these lines. I'd love to integrate that feature with SocialToo's auto-follow. We already have a blacklist users can specify on a per-user level, and if we can allow users to say "I only want to follow those with a less than 5 spam reports (or some dynamic number)" people will follow fewer spammers that way. In fact, we can have it auto-unfollow or even auto-block users that meet certain criteria like that if the user opts to do so. I can provide you an API to add users to the blacklist (on a per-user level), as well as read from it if you want.

Jeff Harbert · 7 months ago

Another idea - Split the @ replies into two sections, one with replies from people I follow and another from people I don't follow. In the second column, add a Block button by each Twitter username. If I get an @ spam, I just click Block. With this right inside the client it would be very easy to manage.

rafaeltech · 7 months ago

I've received a reply spam too. From Br3ndaBot. Did we deserve this?

"@rafaeltech: All international orders must be accompanied by payment in U. S. funds."

So what? I'm just a be concerned about twitter future and solutions for spammers

Diego Sana · 7 months ago

a "report as spam" button is a simple and necessary step, and twitter clients may help providing some filters for their users, but in order to really prevent spam to reach users in the whole platform, the twitter devs needs to come up with smart algorithms and some thresholds. For instance, one obvious thing is, when someone submits a reply, twitter would check the previous replys from that user and compare if they are equal or match a pattern. Or, assuming they know how many replies the average user sends daily, they could set a threshold to log the usernames of whoever, let's say, sends more than 10 replies in a day. Such log could get inspected by smart scripts, by humans (they could build an app to crowdsource spam filtering, like facebook did to translate from english to a lot of languages) or even get cross-match this log to find users who are also in the "reported as spammmers" list. There's a lot of small things like this that could be implemented by twitter and would block most of the spam.

danmactough · 7 months ago

I think the real issue is that Twitter changed what the setting "Show me all @replies" means. That used to still be restricted to people you follow -- and the reason was to prevent spam. See: http://help.twitter.com/forums/23786/entries/14595

In any event, "Block" is your friend -- If enough people block someone, the Twitter team picks up on that and eventually the spam account gets disabled. Same outcome as "report spam." YMMV

Sharon McPherson · 7 months ago

I agree with danmactough. If more people started blocking the spammers and bots, instead of leaving them in their follow list to "pad their numbers", we might start seeing a decrease in this type of activity. As Dan said, the more blocks an account gets, the sooner it raises a red flag to Twitter.

sull · 7 months ago

go to twitter settings / notices.
you can adjust whether you want to see ALL @replies or only @replies form people you follow.
just was not sure if you were aware of that.
still, better measures need to be taken so that we can enjoy seeing ALL @replies from good people and flag spammy @replies.

Anthony Mitchell · 7 months ago

For the 10% of American men with some sort of colorblindness, color coding would cause real problems. Colors can still be used, but need to be accompanied by a high-contrast symbol/number/letter. There are captcha-entering services using manual inputs offshore, paid $8 per hour by their spammer clients.