Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

The coming automation of propaganda

warontherocks.com

Post reply on HN
Print view View original

71–80 of 93 posts

Re: The coming automation of propaganda

#71

Post by pjc50 »

throwaway94857 wrote: post #41
im3w1l wrote: post #9

Imagine you go to an online forum and that there is a 99.99% chance that a given message was written by a bot. Imagine further they are sufficiently well written you can't tell from the post itself whether it was made by a bot or not How would people respond to that? Well the obvious consequence is that you have to stop paying attention to random posts. You have to depend on something outside of the forum itself to f…

I believe the age of the anonymous internet is over. It's time for real world identity linked accounts for participating in online conversations. This is necessary for 2 reasons. First, it gives us confidence the people we are speaking to are (probably) not bots. Second, it reduces the need for online censorship beyond what is required by law. So much of the hateful garbage posted online is only posted in the first p…

Some of the worst garbage is posted under people's real names and bylines in newspapers. Or on the Presidential Twitter feed.

Mandated real names are just pre-doxing everyone so they can be censored by real world retaliation. If you're already powerful, this has no effect. If you're vulnerable, it's silencing.

Re: The coming automation of propaganda

#72

Post by lasagnaphil »

scarejunba wrote: post #36

This is the best thing to happen to human information processing since the printing press. The democratization of propaganda will mean that no longer will people be slave to the voices that are loudest. Now, all ideas will fight equally. No longer will Judith Miller be powerful and lead us into war because we will not trust her. Because when everyone is a liar, we will not choose one liar to trust.

The problem is, who controls the medium? The Internet is not simply a free space that everyone can fight for their voice anymore, because the social networks, online spaces, and the people hosting the websites are slowly being more monitored and censored. For China, the censorship is performed by the government, and for the US, this is performed by the corporations and investors.

Re: The coming automation of propaganda

#73

Post by trqx »

zxcmx wrote: post #11
joe_the_user wrote: post #5

The idea is that something like the (formerly) OpenAI text generator could be used to enhance "boiler room" propaganda operations. The idea is one person could decide on a message and hundred of articles pushing the message could be generated. Maybe that would work. Maybe you'd see so much of this people would start to get more selective. Maybe it wouldn't matter because people believe what suits them rather than jus…

IMO the end state is way worse. Imagine campaigns that: Have a general idea of your life stage, interests, career, personality etc, based on advertising profile and commercial surveillance. Can tailor each message to the recipient. Can a/b test the crap out of every message for target groups, optimising for best response. Can deliver fully customised messages in a natural voice (aka google duplex). Conversational ai…

That sounds a lot like how current web ads works to me, only with some further optimizations.

Am I wrong ? I hope so.

Re: The coming automation of propaganda

#74

Post by jdietrich »

whatshisface wrote: post #27
im3w1l wrote: post #9

Imagine you go to an online forum and that there is a 99.99% chance that a given message was written by a bot. Imagine further they are sufficiently well written you can't tell from the post itself whether it was made by a bot or not How would people respond to that? Well the obvious consequence is that you have to stop paying attention to random posts. You have to depend on something outside of the forum itself to f…

Here's a hot take: you shouldn't be trusting intenet comments written by humans anyway. If a bot comment contains something you should actually listen to (sources, evidence) then you should listen to it, and if a human comment is just a Reddit-style popular opinion dogpile it is just as bad as a bot comment.

That's a very 20th century view of propaganda. Modern propaganda doesn't need to be factually incorrect to be effective, it just needs to direct the attention of the audience in a manner that suits the propagandist. You can push pretty much any agenda in a very convincing manner simply by choosing to emphasise or de-emphasise certain facts; we are in many ways more powerfully influenced by what we consider to be important than what we believe to be true.

If you only read reports of vaccine-related harms, it's easy to come to the conclusion that vaccines are a terrible mistake or even a conspiracy, because human brains aren't very good at prioritising the singular fact that vaccines have saved millions of lives over a constant stream of emotive stories. If you only read about the inconsistencies and errors in contemporary reports of the Holocaust, it's easy to come to the conclusion that the whole thing is exaggerated or an outright hoax, because it's just deeply unpleasant to think about six million corpses.

I don't need to lie to you or dictate what you believe, I just need to influence what you pay attention to; if I control your attention, I can guide you towards a conclusion and make you believe that you came to it of your own initiative.

Re: The coming automation of propaganda

#75

Post by throwaway13337 »

This could actually encourage filtering for better quality content.

I could totally see bots writing most of the YouTube, reddit, and Instagram me-too-style content.

What about your average HN post?

If people want to not look like spam, they're going to have to start being more insightful. This is good.

Re: The coming automation of propaganda

#76

Post by patientplatypus »

This is mildly related, but it's an interesting data point.

I was at a bar tonight and I chatted with a guy who works for a major cable provider. Apparently they generate 10 petabytes of data an hour on what you watch, when you watch it etc. That's a lot when you take into account that its all text columns in a MSQL database somewhere. I think they have enough room to have it stashed in S3 for a week before it goes into cold storage because otherwise it's just too bloody expensive.

What are they storing you ask? There's a whole tree on what you're demographics are, a tree on what your device is, a tree on what you are watching etc. And they're all cross referenced with what ad you watch, for how long, when in the show etc.

All of this whizbang tech to find out how to make sure you spend that extra dollar on whatever crap they want you to. I'm sure that the coming automation is pretty horrifying, but what we have now is pretty terrible as well.

Re: The coming automation of propaganda

#77

Post by techntoke »

throwaway13337 wrote: post #75

This could actually encourage filtering for better quality content. I could totally see bots writing most of the YouTube, reddit, and Instagram me-too-style content. What about your average HN post? If people want to not look like spam, they're going to have to start being more insightful. This is good.

I don't think automation will work or has worked. Look at emails and robocalls. I still get plenty of spam, even if it gets filtered into a spam folder, there is no reason I should have to peruse my spam folder but still some important emails get labeled as spam. I believe the issue is that corporations as a whole simply can't be trusted, and that the only solution is decentralization with open source technology and making it easy for users to create their own filters and control their own data.

Also, the bots I see on reddit, YouTube and Instagram generally aren't pushing liberal agendas but right-wing. The automation I see are networks of influencers working together to push out the same narratives.

Re: The coming automation of propaganda

#78

Post by nradov »

jackfraser wrote: post #50
throwaway94857 wrote: post #41

Earlier quoted context omitted.

I believe the age of the anonymous internet is over. It's time for real world identity linked accounts for participating in online conversations. This is necessary for 2 reasons. First, it gives us confidence the people we are speaking to are (probably) not bots. Second, it reduces the need for online censorship beyond what is required by law. So much of the hateful garbage posted online is only posted in the first p…

> It's time for real world identity linked accounts for participating in online conversations. You might be able to implement this in a small, affluent European country, but nowhere else. Consider the fact that in the USA, it is a standard talking point of the left that the right's continual push to ask for ID to _vote_ has been branded as a racist policy; it seems that we're simply to accept that large swathes of pe…

Divulging classified information is not a crime unless you hold a security clearance.

Re: The coming automation of propaganda

#79

Post by makomk »

jdietrich wrote: post #74
whatshisface wrote: post #27

Earlier quoted context omitted.

Here's a hot take: you shouldn't be trusting intenet comments written by humans anyway. If a bot comment contains something you should actually listen to (sources, evidence) then you should listen to it, and if a human comment is just a Reddit-style popular opinion dogpile it is just as bad as a bot comment.

That's a very 20th century view of propaganda. Modern propaganda doesn't need to be factually incorrect to be effective, it just needs to direct the attention of the audience in a manner that suits the propagandist. You can push pretty much any agenda in a very convincing manner simply by choosing to emphasise or de-emphasise certain facts; we are in many ways more powerfully influenced by what we consider to be impo…

The trouble is, that definition of "modern propaganda" desacribes the entire press. For example, take the FBI's use of the latest mass shootings to push for encryption backdoors - those shootings make up a tiny proportion of deaths, but we prioritize emotive stories that get large amounts of press. This has also lead to some terribly ill-conceived gun control campaigns that downplay the importance of the guns most often used to murder people.

Re: The coming automation of propaganda

#80

Post by IIAOPSW »

colordrops wrote: post #69
IIAOPSW wrote: post #55

Earlier quoted context omitted.

Your counterargument is therefore not with the principle of Epistocracy but rather with the implementation. I've thought about this before and I think I have a way to construct the voting test in a way that is robust to being "hacked". Copy and paste to follow: 1. There is already a de facto voting test for some fraction of Americans. Its called the citizenship test. And I for one find it offensive to basic reasoning…

Way too complicated and certainly would be gamed. And it only filters out for lazy people, not intelligent or informed people. In any case why wouldn't you make the questions about the office that the candidate is running for and what they would be responsible for? That seems like the most relevant knowledge to a vote.

>2. Each candidate on the ballot gets to choose 5 multiple choice questions. The questions must be directly relevant to the powers of the office.

The rest of point 2 is just defining valid questions in a way that is very hard to game and legally unambiguous enough for a judge to enforce fairly.

Post reply on HN