I would imagine that it's possible that some of the people reading these comments might already know a lot more detail about Twitter/Facebook bots than this story goes into.<p>Although you fine folks might remember me from a few other projects, I'm currently a Graphics Editor at The New York Times. We're actively curious and interested in pursuing lines of inquiry about this sort of behavior, and if any HN'ers have any interesting leads or tips, I'd encourage you to get in touch. You can reach me at my username @nytimes.com, or email me and I'll send you my Signal number.
No kidding.<p>I remember when I first moved to the Bay Area in 2009, I met a smart guy at a startup meetup, a serial entrepreneur with a previous exit. I asked him what he was working on, and he said "Oh, I've been writing a Twitter bot that will follow people of your choosing, engage them in simple conversation, and retweet their tweets. It's building a network of followers - I don't know yet what I'd do with that, but it's an asset that's likely valuable to somebody." I ran into him again a few weeks later, and he'd sold the company.<p>I think I'd heard from either him or someone here on Hacker News, around that time, that 75% of Twitter traffic was bots and automated accounts. Note that that was <i>early 2009</i>, before Twitter went mainstream and when they still had a really easy-to-use developer API.
Huh.<p>The landline telephone business has, for a long time now, been compromised by spammers and bots (telemarketing calls and robocalls). I canceled my land line about four years ago after going for three months without receiving a single call I wanted.<p>It seems the commercial social networks are headed for the same fate. And, they're headed for hardnosed and unpleasant regulation by governments. They probably need to clean up their acts.
This seems like a pretty straightforward thing to identify. I remember when the NYT's story, "The Agency", was published [0], some of the fake accounts it had mentioned were still up. Even though the accounts in that story were actually populated by real people, the sockpuppetry was pretty easy to identify. One: the accounts' past tweets right up until they started spreading the news about the fake U.S. disaster were in Russian. Two, all of the tweets of the fake news had almost exactly the same number of favorites and retweets (around 300), and you could see that everyone in that cluster was just retweeting each other.<p>I'm more fascinated by the spam by Facebook accounts. These show up <i>all the time</i> in relatively popular comment sections, and yet apparently FB doesn't care, or the problem is trickier to automatically flag. For example, this comment [1] is clearly spam...but if you click through to the account, it seems to be a real person [1], with a normal-seemingly friend network, mundane photos of life that aren't obviously stock photography. There are a few junk comments (a bunch of "hi's", but as an outsider, this is what makes FB a lot trickier to analyze, because you don't know how much privacy that user has enabled on their own account.<p>[0] <a href="https://www.nytimes.com/2015/06/07/magazine/the-agency.html" rel="nofollow">https://www.nytimes.com/2015/06/07/magazine/the-agency.html</a><p>[1] <a href="http://imgur.com/a/Rr8d3" rel="nofollow">http://imgur.com/a/Rr8d3</a><p>[2] <a href="https://www.facebook.com/gulfam.raj" rel="nofollow">https://www.facebook.com/gulfam.raj</a>
I'm just amazed at this entire series of threads. "Gosh, who knew violating social network ToS by proxy and creating massive false consensus networks was so easy?"<p>Everyone. Everyone should know.<p>We're watching this terrible trend rip apart the entire social proposition of the internet after spending 2 decades trying and finally achieving buy in. And here y'all are, hopefulls for a digital economy cheerfully defrauding the very networks that will probably bet the monetization strategy for many startups that pass through HN's doors.<p>The total lack of any personal responsibility here, or notion of consequence... It stuns me.
In their recent SEC filings, Twitter estimates around 5% of the MAUs are spammers or bots. They also estimate they have 317 million MAUs, which when you work it out gives around 16.7 million monthly spammers or bots.<p>Numbers come from their 2016 Q3 filing
<a href="https://investor.twitterinc.com/secfiling.cfm?filingID=1564590-16-26749&CIK=1418091" rel="nofollow">https://investor.twitterinc.com/secfiling.cfm?filingID=15645...</a>
The question this sort of thing leaves me with is "Why does {Twitter/Facebook/Google} permit this? What are they getting out of it?"<p>Its pretty clear what the bot guys get out of it, pay to promote services, pay for followers, etc. They can monetize "fame" through the robotic horde. But as this article and ones before it point out, these networks are generally quite easy to spot. So why not take them out?<p>It probably isn't because they can pad 'subscriber growth' or 'MAU' numbers, they appear to be only small components of that number. And while I could imagine it may be hard to purge them at the moment, its been a problem long enough that <i>someone</i> in engineering must have figured out a system for taking down large numbers of accounts.<p>The only thing I can come up with, and it is way too tin-hattish to really count, is that it creates an "observable" for the underside of the Internet. By watching what people are asking the twitter bots to do you can observe other objectives that are perhaps less observable. There are some obvious customers for that but I don't think they actually pay for that (except perhaps by buying access to the Firehose)
I'm not surprised. This has been a phenomenon for a very, very long time [1][2][3][4][5][6][7][8].<p>[1] <a href="https://news.ycombinator.com/item?id=11905266#11906591" rel="nofollow">https://news.ycombinator.com/item?id=11905266#11906591</a>
[2] <a href="https://news.ycombinator.com/item?id=9170433" rel="nofollow">https://news.ycombinator.com/item?id=9170433</a>
[3] <a href="https://news.ycombinator.com/item?id=485659" rel="nofollow">https://news.ycombinator.com/item?id=485659</a>
[4] <a href="https://news.ycombinator.com/item?id=5525638" rel="nofollow">https://news.ycombinator.com/item?id=5525638</a>
[5] <a href="https://news.ycombinator.com/item?id=5501654" rel="nofollow">https://news.ycombinator.com/item?id=5501654</a>
[6] <a href="https://news.ycombinator.com/item?id=5996790" rel="nofollow">https://news.ycombinator.com/item?id=5996790</a>
[7] <a href="https://news.ycombinator.com/item?id=833188" rel="nofollow">https://news.ycombinator.com/item?id=833188</a>
[8] <a href="https://news.ycombinator.com/item?id=4346386" rel="nofollow">https://news.ycombinator.com/item?id=4346386</a>
I think this story is getting more coverage than the one from a few months ago that showed that, for instance, both Clinton and Trump had <i>millions</i> of fake accounts following them:<p>> <i>Per eZanga, 4.3 million—or 39 percent—of Trump's more than 11 million Twitter followers as of August came from fake accounts while the other 6.7 million are actually real users. And for Clinton, 3.1 million—or 37 percent—of her more than 8 million followers were fake while 5.3 million come from real accounts.</i><p><a href="http://www.adweek.com/news/technology/more-third-presidential-candidates-twitter-followers-are-reportedly-fake-173628" rel="nofollow">http://www.adweek.com/news/technology/more-third-presidentia...</a><p>350,000 is about 0.1% of Twitter's user base. Does anyone here think the number of fake accounts isn't orders of magnitude higher than that?
Anyone on twitter could've told you that. When you're followed by "@hotttladie3" & "hotttladie453892", it's only reasonable to assume there is also an @hotttladie4 through @hotttladie452891.<p>The problem with having a user reporting based plan for acting against fake accounts in an environment where the psychological motivations of using the service, if not to disseminate news, or maintain a closet standup comedian habit, is affirmation. In almost every motivation for using the service, the user has an incentive to keep their numbers up, whether they are real or not. A huge part of the game is the number of followers.<p>Personally I'm just surprised that they've moved from advertising cam sites (which could conceivably act as a secondary, almost passive income) to quoting star wars novels while inflating numbers for people who pay for followers. That's the aspect that confuses me, and makes me feel vulnerable.
There was a "Ask HN:" post about Twitter bots the other day: <a href="https://news.ycombinator.com/item?id=13497235" rel="nofollow">https://news.ycombinator.com/item?id=13497235</a><p>Strange that these bots aren't spammy but are posting every minute or two. I wonder what they're for...<p><a href="https://twitter.com/superpolice001" rel="nofollow">https://twitter.com/superpolice001</a><p><a href="https://twitter.com/superpolice002" rel="nofollow">https://twitter.com/superpolice002</a><p><a href="https://twitter.com/superpolice003" rel="nofollow">https://twitter.com/superpolice003</a>
> A Twitter spokesman said the social network had clear policy on automation that was "strictly enforced".<p>> Users were barred from writing programs that automatically followed or unfollowed accounts or which "favourited" tweets in bulk, he said.<p>I am constantly getting followed by accounts with tens or hundreds of thousands of follows and followers, usually checkmarked accounts though I've never heard of them. It's painfully obvious these verified users are using bots to randomly follow people, both to spam my inbox with "you have a new follower" messages and to encourage people to "follow back".<p>But Twitter does nothing about it. It's not "strictly enforced" at all.
This has become obvious to me after getting a few blog posts at the top of HN.<p>I just search for my domain on Twitter, and there are dozens of "people" who do nothing but retweet hacker news articles. They are presumably doing this for some kind of "reverse" reputation.<p>I'm interested if anyone has any more insight on this phenomenon. Maybe it's as simple as convincing some naive users to follow them with links vetted as high quality.<p>Examples:<p><a href="https://twitter.com/bartezzini" rel="nofollow">https://twitter.com/bartezzini</a> (123K tweets, nothing but HN-type links and comments)<p><a href="https://twitter.com/EggmanOrWalrus" rel="nofollow">https://twitter.com/EggmanOrWalrus</a> (15.7k tweets, ditto)<p><a href="https://twitter.com/MarkBeacham" rel="nofollow">https://twitter.com/MarkBeacham</a>
UCL has dropped the ball on marketing itself.
If this had been a student at MIT or Harvard, the article headline would have started with "MIT/Harvard Researcher ..." instead of burying the school's name in tiny print in the middle of the article.
Bot accounts, not fake accounts. I don't even know what the latter means—plenty of people don't associate twitter with their real name. And why would you!<p>Secondly, of course there are this many. There are probably many more. I run several bots myself; there's nothing wrong with this.<p>Twitter's TOS is only as good as its enforcement, and if there's anything twitter is terrible at, it's having any control over its community.
Twitter has become unbearable to use for news. Every single submission from news stations to President Trump is filled with combo-replies from people. 1 dude will bombard the tweet, then a chick, then a different guy.<p>Example: <a href="http://i.imgur.com/zbyM7YG.png" rel="nofollow">http://i.imgur.com/zbyM7YG.png</a><p>You need to scroll down at least 5 load-more's to see regular people tweets. It's a really terrible user experience that Twitter needs to solve.
I remember hearing about this radio host who was competing with another host to get more followers. Then over about a week one of his fans who owned a bot net gave him 100k followers. He sent a DM saying "you're welcome, enjoy".<p>He already had about 250k followers so it wasn't a huge spike in that context but it was interesting to think of the implications of that when you come across a random account with 100k followers and 100 following... they might not be as influential as it seems.<p>Also this was about a year ago and last I checked his follower account was roughly the same.<p>Wasn't there a story on HN about a guy who created a fake identity and twitter account with 20k followers and got invited (and paid?) to speak at a tech conference?
Twitter says 5% of their MAUs are bots but this surely an underestimate. Facebook estimates nearly 9% fake users which is also likely too conservative. Bot traffic in ad fraud is even higher (30+%? every study is different, seems difficult to measure).<p>So we have 5-9% as a lower bound and perhaps we can look at e-mail for an upper bound with nearly 60% spam by volume.
Is this really a surprise?<p>I don't even use Twitter, and I have 4 different Twitter accounts. The amount of fake accounts must be staggering, but there's no way Twitter will cull them otherwise their MAU numbers will tank, and along with it, their ad rates, etc.
- messages being posted only from Windows phones<p>The smoking gun. What real person still has one of those.<p>Disclaimer: I owned a Nokia Lumia 920 for > 2 years.
It is trivial to create a BOT type twitter account.<p>1. Get burner email and phone number<p>2. Post bot to DO, AWS or run it on a raspi<p>3. ???<p>4. Profit from all those sweet followers.<p>Many of them look entirely "real" or they can be hilariously obvious. I would bet it is happening on Facebook, Instagram, Snap and any other social network where "value" is derived from followers/eyeballs.
if youve not read it already, here is an excellent interview with Andrés Sepúlveda who has helped to "rig elections throughout Latin America for almost a decade" partly through using twitter bots and similar techniques. <a href="https://www.bloomberg.com/features/2016-how-to-hack-an-election/" rel="nofollow">https://www.bloomberg.com/features/2016-how-to-hack-an-elect...</a>
Anecdotally I've seen a huge uptick in spam/bots following my main account on Twitter since the beginning of the year. Probably 50%+ of my new followers have been accounts with no profile picture, no tweets, and close to zero followers.
Everybody talks about bots but let's not forget that there are plenty of bot-like accounts controlled by humans from poor countries for (fractions of) pennies. You don't need a "hacker army", just a couple of bucks and a third-world nation with good enough infrastructure. Look up "click farms".<p>The problem with this is that while bots can be detected (even if doing so is an arms race) it's much harder to detect "bot" humans.
this is relevant is as well:
<a href="https://play.google.com/store/apps/details?id=com.hitwe.android&hl=en" rel="nofollow">https://play.google.com/store/apps/details?id=com.hitwe.andr...</a>
tldr:
'HitWe' it's a tinder-like dating app with 10M+ installs on google play (haven't checked itunes).
i came across it by browsing dating apps for fun.
most of the bad reviews (1-2 stars) are from people who claim it's 95% fake profiles etc. many claim that women there just ask for their email etc. some even claim to have paid some of the women which have then disappeared (but that's being too naive, IMO. never pay someone you have never met.. ).
there are many many good reviews, but most are only with 5 stars and without any text (which, to me, is another red flag).<p>anyway - the most interesting part here is that they actually managed to fool google, up until this very moment!
google recommended me to view their app.. spam does work !<p>ps - googling (ironic, i know) 'hitwe app scam' showed me this on the first result:
<a href="http://www.datingbusters.com/hitwe-com-exposed-for-fake-profiles-throughout-their-site/" rel="nofollow">http://www.datingbusters.com/hitwe-com-exposed-for-fake-prof...</a><p>i am interested in a response from one of google play's spam engineers/managers ..<p>edit: spelling<p>edit2: it took me 0.5 secs to start sensing that it's a fake-boosted app. a human reviewer at google could have just scanned the top 100x dating apps in a single day and map out the fake apps. what do you think?
Spam was a serious and growing problem before the major providers got serious about filtering. For now I just report clickbait feeds as spam and hope for the best.
in '08 or '09 I met a guy who was creating 100s or more accounts and paying an overseas outsourcing company to curate longterm timelines properly (so not just spam)<p>he genuinely didn't know why he was doing it at the time (he'd been heavily involved in gaming Google rankings previously for credit card companies) and it was at significant cost - but he was completely sure at some point it would be useful
Related research paper, "The 'Star Wars' botnet with >350k Twitter bots":<p><a href="https://news.ycombinator.com/item?id=13445289" rel="nofollow">https://news.ycombinator.com/item?id=13445289</a><p>And another paper covering the topic:
<a href="https://news.ycombinator.com/item?id=13445295" rel="nofollow">https://news.ycombinator.com/item?id=13445295</a>
I get follows from obvious bots (bio: "click here for 5000 follows") very regularly on Twitter, Instagram, and Facebook. Instagram almost always removes the account, Facebook has always responded the account "doesn't violate the rules", and Twitter I'm not sure about, but I always block and report that garbage.
I thought everybody knew that a big chunk of Twitter works this way. I got to see it in action just a couple weeks ago with a pro-Trump network of fake Twitter accounts. I had the misfortune of having one of those @TrumpRulesMAGA4Eva style Twitter accounts reply to something I'd written with a standard pro-Trump meme. A couple people liked and favorited within an hour or so and then nothing. Then late on a Saturday night several days later, a whole network of Twitter accounts favorited and retweeted it. Most of them had similar style names but some had generic ones or random character strings. About half had an egg profile picture. Many of the others had tried to look legitimate with real names and pictures of people but tineye reverse searches showed 10,000+ hits on the images. Looking at the histories of the accounts, they'd all been used similarly for months.
SoundCloud is full of fake accounts. I get a couple of new followers every now and then and it feels like 95% of them have mentions of "buy followers". I always report spam bots whenever I spot them on SoundCloud, Tinder, Facebook and Reddit, so though I only have very few followers on SC I think all of them are real people at least. It is very important to always report spam bots and other types of disruptive fake accounts and I wish that everyone did. If everyone did then it would be less profitable to run such accounts and then there would be fewer of them and we would be bothered much less frequently.
In a time where bots and AI are all the rage, it begs the question: Are some bots on twitter more useful and insightful than actual people?<p>I remember a twitter bot (name escames me now) which would crawl pastebin and tweet updates when passwords / DBs were leaked. It had lots of followers (security researchers who found it very useful).<p>Most humans on twitter are boring and waste people's time with youtube style comments. Most bots are spammy and waste time too. Why not allow both and let people decide who they follow/ban?
I found it interesting that a Member of parliament on twitter had hundreds of obvious bot followers. I presumed he was buying 'popularity'. Maybe more sinister. Lots of them very recent
The paper: <a href="https://arxiv.org/pdf/1701.02405v1.pdf" rel="nofollow">https://arxiv.org/pdf/1701.02405v1.pdf</a>
Standard stuff, you want media attention for your blog, twitter whatever? You'll need to have a bot army to upvote your own stuff, relink and like your content.<p>Having a botnet is now an essential part of building your social media following.
On the same topic. I think he might have discovered the same botnet. <a href="http://sadbottrue.com/article/51/" rel="nofollow">http://sadbottrue.com/article/51/</a>
hi,i'm Brian, i had my friend help me hack my ex's email, facebook, whatsapp,and his phone cause i suspected he was cheating. all he asked for was a his phone number. he's email is (hotcyberlord425@gmail.com)..IF u need help tell him Brian referred you to him and he'll help. Am sure his going to help you do it, good luck
Meanwhile Twitter banned yet another political youtuber @SargonOfAkkad, doesn't seem like he did anything against Twitters ToS. Fed up with Twatter to be honest.