News for hackers. Stuff that matters.
Re:The Follower Factory by kibwen on Saturday January 27, 2018 @01:44PM (#16247416)
> If Twitter applied this analysis, the bot makers would adapt to it.

The problem with the arms-race argument here is that it implies that just because an arms race is inevitable that the only acceptable course of action is total capitulation. Twitter is bigger than any of the botters, and if it wanted it could raise the cost of entry of botting high enough that nobody would be able to afford the bots in any quantity large enough to make a difference. They just simply don't want to.

(As for false positives, I've seen plenty of people whose accounts were suspended for reasons unclear to them; Twitter seems to care very little whether its captive userbase is pissed off or not.)

Re:The Follower Factory by kbenson on Saturday January 27, 2018 @01:56PM (#16247479)
I don't think it implies that unless you've already accepted they aren't doing anything. I think it's entirely possible that a lower class of bots are routinely blocked, but a higher class are always mutating to get past the latest filtering, so has a large fairly stable population.
Re:The Follower Factory by dmoy on Saturday January 27, 2018 @01:57PM (#16247480)
> Twitter is bigger than any of the botters, and if it wanted it could raise the cost of entry of botting high enough that nobody would be able to afford the bots in any quantity large enough to make a difference.

How would they do this without triggering enough false positives to destroy their product?

I mean seriously, if you are capable of doing this that well, you could probably walk into a job paying >>$750k/yr at any of the large tech companies, because they'd be falling all over themselves trying to get at your magic.

Re:The Follower Factory by kibwen on Saturday January 27, 2018 @02:10PM (#16247564)
> How would they do this without triggering enough false positives to destroy their product?

This implies that Twitter has demonstrated that they care about practicing discretion when suspending accounts, and that they care what 99% of their (human) userbase thinks, and that Twitter's product is Twitter accounts. But they haven't; they don't; and it's not. The state of being pissed off at Twitter is the perpetual existence of all Twitter power users. Unless you're a celebrity, Twitter doesn't care what you say about Twitter. And the only thing important to Twitter's business model is keeping celebrities on Twitter. The reason that they don't give a damn about cracking down on bots isn't because they don't have enough $750k engineers, it's because dealing a blow to the egos of their Influencers by banning their botted followers would threaten their bottom line, both in terms of outrage from the 1% of Twitter accounts that matter and in terms of depressing their wildly inflated metrics.

Re:The Follower Factory by anigbrowl on Saturday January 27, 2018 @04:20PM (#16248127)
TWitter trashed my very real account that was tied to a disused email address after I tried to log in from my phone and misremembered my password and refused to lift a finger to help me get it back despite me offering to provide all sorts of other verification information.
Re:The Follower Factory by nl on Saturday January 27, 2018 @11:42PM (#16249855)
That’s nonsense. It’s trivially easy, and as noted elsewhere they can just add a verification check for suspect accounts.

It’s easy to see how much harder Facebook and Google accounts are to create bots for than Twitter. Twitter doesn’t care.

Re:The Follower Factory by ImSkeptical on Saturday January 27, 2018 @02:07PM (#16247542)
Seems like, if you detect a bot, you should impose a captcha on every x actions. Not enough to ruin your customer's experience, but enough to make botting harder.
Re:The Follower Factory by dmoy on Saturday January 27, 2018 @02:14PM (#16247592)
I would be absolutely stunned if Twitter et al. hadn't already tried this...
Re:The Follower Factory by jimnotgym on Saturday January 27, 2018 @03:19PM (#16247897)
I wouldn't be suprised at all. Go find an openly racist account and report them. Two weeks later Twitter will agree with you...but not ban the account
Re:The Follower Factory by dawnerd on Saturday January 27, 2018 @04:31PM (#16248180)
I’ve come to the conclusion after reporting hundreds of obvious spam/racism/hate/etc accounts that twitter just doesn’t care. It increases their metrics and that’s all they want. Makes the share holders happy to see numbers going up and sadly in turn it does increase ad views.
Re:The Follower Factory by ImSkeptical on Saturday January 27, 2018 @04:14PM (#16248101)
I would be curious to know how this failed, if they did try it. Are captcha automatable? You could make it a bit better by challenging every X actions, and have X decline with a failed captcha.
Re:The Follower Factory by dmoy on Saturday January 27, 2018 @06:24PM (#16248700)
Captcha has issues in that:

- captcha that isn't hard for humans is often easy to automate

- captcha that is hard to automate frustrates a lot of humans too

- and to add insult to injury you can "automate" the hard ones too by paying like $1 per 1000 captchas or something (think Amazon Turk like stuff)

Re:The Follower Factory by pm90 on Saturday January 27, 2018 @06:31PM (#16248729)
> They just simply don't want to.

This really does seem to be the case. And I find it incredibly distasteful that they were used by DJT in his campaign and continues to be used. It was rarely a place where politicians posted anything, and now it seems to be the first place where any of them do.

Re:The Follower Factory by slavik81 on Saturday January 27, 2018 @09:21PM (#16249377)
> It was rarely a place where politicians posted anything

Twitter was founded in 2006 and Barack Obama joined in 2007. Today he has 15.4k tweets. Seems like a lot to me.

Re:The Follower Factory by ball_of_lint on Saturday January 27, 2018 @11:28PM (#16249813)
There are plenty of ways besides suspending accounts to make the cost of botting too high to be profitable. A captcha for example is fairly frictionless for a typical user.