Wouldn't this be an end goal for malicious bots? I also think this article puts too much emphasis on people "being lied to" by all social media bots with statements like: "One question begs for closer analysis; why would a small percentage, though small, find anything positive about the bots, about being lied to, that the information is not free of salary packet or automated word strings?", when that's absolutely not true for all bots.
For example, a National Weather Service severe weather social media bot just posting severe weather warnings probably isn't lying to you.
An NWS Twitter bot (although I would actually expect it to be a human posting) isn't telling the truth because of the content of its tweets but because it accurately represents itself.
If a Twitter bot named FSB Social Media Account #5527 posted commentary about American politics, there would be very few problems regardless of content.
Does it matter if a robot can build a car? People can build cars.
Of course it matters. A bot can scale to appear as millions of people. So whoever wields the most compelling bots gets a massive force-multiplier for their opinions and objectives. Or more likely, grants that power to the highest bidder.
Not really -- for instance, at HN we pay more attention to comments that explain stuff well, refer to sources, and so on.
The bots might be a blessing, in some ways, if they result in more people looking for good explanations rather than for assertions or herd direction; assessing ideas rather than their sources.
It's the difference between an angry mob and an angry mob of puppets controlled by one guy with an agenda.
Normally, without the puppet/bot amplification, the latter guy could be easily ignored. Now his bots can drown out real voices, and that's a problem.
See: the thousands of letters to the FCC regarding net neutrality that were sent by "bots" impersonating real people. Letters written by actual voters were dwarfed by the fake letters.
There's a more fundamental problem you're hitting on. Somehow we have shifted from responding to rationality to responding to noise. Granted it's always been the squeaky wheel that gets the oil, but now you have the cow squeaking for oil. That was one awkward metaphor... What I mean is that in times past getting involved in issues took a decent amount of effort and time. You had to really care about an issue to become active with it. Now the internet provides a 0 effort means for people to chip in with their 2 cents. And so there's a lot of noise that says absolutely nothing, even without bots. And the worst part is that companies and individuals are increasingly treating the noise as the signal, and that's counter productive.
What is the value of automation in general? Changing opex (training & paying a single human to post propaganda) with capex (buying another server to host a thousand bots posting propaganda, plus some fixed opex to pay a programmer to orchestrate it all). Servers are cheap, and some people with megalomaniacal dreams of influencing society through information have deep pockets.
This quantitative change breeds qualitative change. Any time computers are used against us, be it in information warfare or physical, the reduced cost and decreased risk mean the activation energy is lower and the scale so much greater. The barrier to entry becomes lower, too.
It’s weird that people have an issue with bots when the bigger issue is astroturfing. But no one wants to get rid of astroturfing bacause that’s every brand’s bread and butter.
Tangential, but curious as how one would classify the various national astroturfing propaganda offices (think China, N Korea, India, Russia, etc.) Since they are human operated, they're not bots but actually fit the astroturfing definition but I often see them confused or conflated with "bots".
The article talks about people's perception of whether or not they can distinguish between bots and humans on social media. A more interesting question is whether this perception is correct. Are bots actually getting better at fooling humans or is there simply more cynicism about social media bots? We can't tell from this article.
I'm having trouble understanding the terms being used by the article. First off - bots. Generally I understood them as automatic agents doing something a person otherwise would do themselves. This doesn't seem to be the definition in use here:
> why would a small percentage, though small, find anything positive about the bots, about being lied to
Under the "bots = automation" definition, this is similar to saying "why would anyone find anything positive about computer programs".
Instead, this article is saying:
bots = lies
Does the article actually talk about astroturfing trolls leveraging automation? The automation part doesn't seem to be relevant, as paid trolls are the problem.
In the end, I understand that the article has a problem with lying on the internet, but what does "bot" mean to the society at large these days?
Bots that attempt to make themselves indistinguishable from humans - the bots under consideration in this article - are attempting to deceive people. If something is signed 'Amazon Ad Bot' or 'Moscow Political Ad Agency Bot, funded by the Bob Charles for Senate campaign', that's one thing. If it's signed 'Amy Jones", that's a lie.
> The automation part doesn't seem to be relevant, as paid trolls are the problem.
Without the automation, the paid trolls' reach would be far less. It's like saying, 'the nuclear weapons part doesn't seem to be relevant; it's the hostile dictator that's the problem'.
So far, the most common issue raised in this HN discussion is the poor bots being demonized. Only on HN.
This is similar to the futile struggle to get people to stop calling "crackers" "hackers". "Bot" has now come to encompass "bot", "troll", "sock puppet", and "shill".
Or for that matter, the struggle to get people to stop calling bullies, harassers, bots and sockpuppets 'trolls'. Troll used to mean 'someone who who said/did things for the sole purpose of annoying people'. Not 'annoying person/bot online I don't like'
But as with hacker, these terms have basically lost all meaning now.
It is hard to state directly, and there is a lot of ambiguity.
The difference I'm trying to establish is between directing specific actions (post this at 3:45 Tuesday) and directing behavior (every hour upvote all posts positively portraying hot air balloons)
Digging too far into the difference between automated and autonomous obscures the larger point I was trying to make.
Fake reviews might be the best example. The lie is that there is a person who bought the product and has an opinion about it. Or a fake account on Twitter might not just be automated, but pretends to be a human who has an opinion, in an attempt to shift what people believe are common opinions, or just to raise follower counts.
Presumably if an automated account isn't trying to look human then it's unlikely to be mistaken for one, although if it's just there to bump up a number, it might not have to try very hard.
In my opinion this is due to people increasingly reappropriating the term 'bot' much in the way 'troll' has been. Troll originally had a clear meaning. It was people posting things exclusively to obtain incensed reaction. Now somehow 'troll' has increasingly been applied to anybody stating something somebody else strongly disagrees with.
And now the same thing is slowly happening to bot. It's not only people 'trolling', but it's being done by 'bots'. I believe this is largely a product of the increasing antagonism and radicalism in different 'sects' of our country today. By referring to people you disagree with as 'bots' and 'trolls' it works exactly like wartime propaganda - it dehumanizes and delegitimizes the 'enemy'.
Competent trolls don't evince normal human reactions; if their manipulation is effective, trollees will have their emotions toyed with yet won't be able to break through the troll's facade and get them to respond in kind. Perhaps the experience of being trolled successfully is not unlike interacting with something inhuman... or robotic.
That reminds me of the writer who managed to get ahold of a major troll of hers and talk to him on the phone. And... he apologized. An interesting listen.
I’ve noticed the conservatives? Alt-right? People on T_D on Reddit? Have started calling liberals “NPCs” (non player characters in video gaming terms). On both sides it’s a worrying trend of dehumanizing the other. A trend of cutting economic ties (boycotts, getting people fired) is also worrying, because you don’t go to war with people you trade with. Cutting economic ties is a sign of impending conflict.
Countries do go to war with countries they trade with. People were saying record high levels of trade would prevent war ... in Europe, right before world war 1
To be fair they were totally right that going to war would be a terrible idea but they did it anyway - given that World War 1 pretty much killed off the ruling aristocracy, lead to countless suffering, and stagnation. If they were smart it would have prevented the war but as Bismark predicted it all blew up over a fool thing in the Balkans.
It is like MAD - the assumptions are based upon rational actors which is why they are terrified of proliferation beyond just others getting in their clubhouse.
Unfortunately, the term "NPC" seems to be just a childish insult more often than not. I guess the word "shill" is typically an insult and rarely ever used to actually attack an actual bot or paid-marketer (even though that's the implication).
Troll is the most accurate word I think. I think people are generally happy when a paid employee interacts with the audience (technically a "shill" but... in the case of MMORPGs and such, people like feeling that the game-masters are listening to them).
I'm personally going to stick with the word "troll". I don't really care if they're paid or not, or foreign or not. If they're hampering the discussion, they're a troll. If they're helping the discussion, then they are welcome. So "Troll" is the best distinction, and the main one that matters.
NPC is the right-wing version of 'bot'. These days the term 'Bots' can refer to all sides of social media shilling, but seems to be most commonly used in relation to (perceived?) Russian interference and pepe-/the_donald-type memes.
(I'm not a fan, because I feel 'bot' should refer to, well, actual bots, and the better term would be 'shill' or 'sockpuppet', but whatever)
NPC, on the other hand, seems to be used mostly by right-wing actors in reference to left-wing activity/interference.
Personally I'm inclined to think it's largely projection on the side of the NPC-memers/(alt-)right, but I'm sure there's still a bunch of truth to it. Everyone with half a brain uses 'bots' these days.
How about a browser extension or app that highlights suspected bots as you're using Twitter or other social media? How difficult is this problem to solve?
It would be really easy for the developers to highlight their political adversaries as 'bots', and people would be none the wiser since they forfeited their ability to perform critical thinking to have a 'bot' do it for them.
I'd suspect bots will be able to become much better at avoiding suspicion much faster this way. Any popular extension is basically a free A/B testing environment for bot producers.
Twitter's API limits would stifle anything that could be useful. Software that scrapes Twitter and becomes popular would be shutdown for violating ToS.
Also, exposing bots on Twitter is how you get the eye of Sauron to notice you. Twitter portrays bots as real users to investors and their advertising clients.
I have a plan - instead of combatting the bots make their influence smaller by limiting the access to them. Hike up the mobile internet prices would be a good start.
43 comments
[ 0.22 ms ] story [ 82.4 ms ] threadFor example, a National Weather Service severe weather social media bot just posting severe weather warnings probably isn't lying to you.
If a Twitter bot named FSB Social Media Account #5527 posted commentary about American politics, there would be very few problems regardless of content.
Of course it matters. A bot can scale to appear as millions of people. So whoever wields the most compelling bots gets a massive force-multiplier for their opinions and objectives. Or more likely, grants that power to the highest bidder.
Meanwhile, labor in Asia is quite cheap ...
Also, they might use bots for the "easy" people, and real human workers for more alert people.
The bots might be a blessing, in some ways, if they result in more people looking for good explanations rather than for assertions or herd direction; assessing ideas rather than their sources.
Normally, without the puppet/bot amplification, the latter guy could be easily ignored. Now his bots can drown out real voices, and that's a problem.
See: the thousands of letters to the FCC regarding net neutrality that were sent by "bots" impersonating real people. Letters written by actual voters were dwarfed by the fake letters.
This quantitative change breeds qualitative change. Any time computers are used against us, be it in information warfare or physical, the reduced cost and decreased risk mean the activation energy is lower and the scale so much greater. The barrier to entry becomes lower, too.
> why would a small percentage, though small, find anything positive about the bots, about being lied to
Under the "bots = automation" definition, this is similar to saying "why would anyone find anything positive about computer programs".
Instead, this article is saying:
bots = lies
Does the article actually talk about astroturfing trolls leveraging automation? The automation part doesn't seem to be relevant, as paid trolls are the problem.
In the end, I understand that the article has a problem with lying on the internet, but what does "bot" mean to the society at large these days?
Bots that attempt to make themselves indistinguishable from humans - the bots under consideration in this article - are attempting to deceive people. If something is signed 'Amazon Ad Bot' or 'Moscow Political Ad Agency Bot, funded by the Bob Charles for Senate campaign', that's one thing. If it's signed 'Amy Jones", that's a lie.
> The automation part doesn't seem to be relevant, as paid trolls are the problem.
Without the automation, the paid trolls' reach would be far less. It's like saying, 'the nuclear weapons part doesn't seem to be relevant; it's the hostile dictator that's the problem'.
So far, the most common issue raised in this HN discussion is the poor bots being demonized. Only on HN.
But as with hacker, these terms have basically lost all meaning now.
Hell, someone the other day passed me their phone and asked me to take a “selfie” of them...
"Do this on my behalf" automation is different than "behave with these parameters" autonomy.
There is a grey area, but the differences are who is making decisions, who is being represented, and what is the motivation.
Is it though? The "on my behalf" implies that there are parameters specified.
The difference I'm trying to establish is between directing specific actions (post this at 3:45 Tuesday) and directing behavior (every hour upvote all posts positively portraying hot air balloons)
Digging too far into the difference between automated and autonomous obscures the larger point I was trying to make.
Presumably if an automated account isn't trying to look human then it's unlikely to be mistaken for one, although if it's just there to bump up a number, it might not have to try very hard.
And now the same thing is slowly happening to bot. It's not only people 'trolling', but it's being done by 'bots'. I believe this is largely a product of the increasing antagonism and radicalism in different 'sects' of our country today. By referring to people you disagree with as 'bots' and 'trolls' it works exactly like wartime propaganda - it dehumanizes and delegitimizes the 'enemy'.
https://www.thisamericanlife.org/545/if-you-dont-have-anythi...
It is like MAD - the assumptions are based upon rational actors which is why they are terrified of proliferation beyond just others getting in their clubhouse.
Unfortunately, the term "NPC" seems to be just a childish insult more often than not. I guess the word "shill" is typically an insult and rarely ever used to actually attack an actual bot or paid-marketer (even though that's the implication).
Troll is the most accurate word I think. I think people are generally happy when a paid employee interacts with the audience (technically a "shill" but... in the case of MMORPGs and such, people like feeling that the game-masters are listening to them).
I'm personally going to stick with the word "troll". I don't really care if they're paid or not, or foreign or not. If they're hampering the discussion, they're a troll. If they're helping the discussion, then they are welcome. So "Troll" is the best distinction, and the main one that matters.
(I'm not a fan, because I feel 'bot' should refer to, well, actual bots, and the better term would be 'shill' or 'sockpuppet', but whatever)
NPC, on the other hand, seems to be used mostly by right-wing actors in reference to left-wing activity/interference.
Personally I'm inclined to think it's largely projection on the side of the NPC-memers/(alt-)right, but I'm sure there's still a bunch of truth to it. Everyone with half a brain uses 'bots' these days.
2. What's an acceptable false negative rate before the browser extension is politically problematic
Also, exposing bots on Twitter is how you get the eye of Sauron to notice you. Twitter portrays bots as real users to investors and their advertising clients.