I would imagine that any team with bigtech/bigfactcheck/soros type funding would be able to build some impressive systems to manage forums and social media platforms.
Most of social media has been fake for a very long time. I’ve worked with some of these companies and they all know it too but like to have their inflated numbers so they do their best to only target things causing them problems or that are very very obviously not real, both in terms of bots and paid content.
It’s almost always been this way from the moment Facebook had ads and any social media site had an audience. You could spend $1mil on TV and movie ads to promote something, or you could pay $20-50k a person a year to create fabricated organic content in all the right places and get everybody to talk about things and like it. It’s been an obvious play from the beginning. And it does work very well. I’ve worked on things like that for things unrelated to politics and the returns were great. Back then you could even get by with a team of two people making multiple accounts and make it seem like you had thousands of people talking about and posting something. The sites love it too because they also can now inflate their numbers. You can do other minor things that add up on top of that too, like tweak the settings to show active users as counted over a longer period, like say anyone that logged on that day vs that hour, to dramatically increase perceived engagement.
It’s all very easy to game and when people realized that they let it happen because it was mutually beneficial.
There are two main options for getting the data, scraping or using their API. (the other options are being reddit and having the data locally, or reddit pushing the data to you)
The scrapers will have a delay because they are forced to poll at regular intervals to obtain the data. Your delay won’t be short because you don’t want to be banned for spamming their server. You could operate from unique IP addresses using a VPN, but if you are running a bot farm, it doesn’t matter.
If you are using their API, then you will be rate limited for the same reason. They don’t want to waste bandwidth or have their servers overloaded.
So no matter what, you will have some kind of delay. A minute or two, or at least thirty seconds.
If you are getting replies instantly, that’s because the bot doing the replying doesn’t need to ask for the data over the internet. They are watching for or being notified of it immediately. So either reddit is running the bots, or they are forwarding the messages to the bots as soon as replies are posted. There has to be some coordination.
The problem here is that you're just a theorist. I've been writing bots, scrapers and pollers for over twenty years.
Uh no, I’ve written bots as well…
There isn't a magic filter and you'll find in reality limits are quite forgiving. You'll also find that a few heuristic targets and you don't have to poll a great deal.
You do if you want instant responses, because you need to know about them before you can reply, correct? Given that you need to first check for those replies, you have to be polling pretty frequently if you want your responses to be instant. Far too frequently to be workable if you are doing other things with your bot too.
Everything you're talking about is blind theory without any real metrics. What's the actual request rate per minute? You're pulling numbers out of your butt that on the face of it don't hack. You don't get rate limited hitting F5 twice in 30 seconds.
Well if you are going to whine about it, reddit’s API limits you to 60 requests per minute. If you are doing actual bot like activity, that’s not a lot.
The assumption that a bot would necessarily poll at a fixed interval is also primitive. You can make one that speeds up or slows down based on rate of change.
The reason you would poll at a fixed interval is to reduce requests… I mean I’d assume your bot is not just checking replies. It would be crawling, voting, checking replies, replying, posting, checking messages, among other things. Each of those is an API request.
Comment responses are likely to be inevitably avoided and easily delayed to appear more organic. It's downvotes that you'll notice are more likely to be instant. You could see this pattern on reddit.
Dude, my whole reply is in regard to someone saying that bots are responding instantly. So yeah, obviously you could delay responses to appear more organic, but that’s not what we’re talking about…
Not faster, instant. That is the word he used. Maybe it is an exaggeration on his part, but just for the sake of example, let’s say he is correct.
Don’t you think if you are relying on an API, you would need to do a hell of a lot of polling to have instant responses?
You can only do 60 requests per minute, one per second, and some of those requests need to be spent on performing other activities.
So at a minimum, you would need to interleave requests to check the comment replies to the bot at regular, short intervals, and you need to ensure that no matter what activities your bot does, it does not exceed the limit.
It’s not like the replies come at regular intervals either. Say you get five replies to a comment, well if you want to respond instantly, that means you would need some buffer so that you are always free to send a response without hitting the API limit. So you can’t even fit one action per second. You need to be far below that in case you want to do something more than just check the messages you are receiving.
I’m saying it doesn’t even make sense, from a bot writer’s pov, to have instant responses. As you mentioned, you would be far more realistic if you didn’t do that at all.
So then why would instant responses even be a consideration? Well they wouldn’t unless you were not subject to the limitations of Reddit’s API.
AWS + immediate link clicks + not engaging in discourse/simplistic distractions = this is a bot/AI. The more time you spend on a site, the more opportunities they can collect ad revenues and data and people spend more energy on arguments than coming to a common understanding. The AI is using you as a learning opportunity in interaction.
They are so good hardly anyone can tell the difference. It’s creepy. I caught it once only because it repeated itself once - and I have experience and training. It was shocking.
It might be used as a data point. If these are in fact bots ran by A.I, then they learn from experience. Opening a link, scanning the info, and responding with a comment gives legitimacy. For most users, they will believe that this bot is actually a person. It passes the Turing test.
But the flaw in the design is the fact that the bots click on in without understanding it's a IP tracker.
This guy Kitboga on youtube built an bot to screw with scammers and it didn't seem like it was a difficult project for him to accomplish.
https://youtu.be/Hx5R0Vnp6w4
I would imagine that any team with bigtech/bigfactcheck/soros type funding would be able to build some impressive systems to manage forums and social media platforms.
Intelligence agencies should be capable of taking what you saw on that video to the next level. If anything they have the cash to do so.
Very likely this an existing capability of the Shadownet that Millie Weaver was covering this year.
Most of social media has been fake for a very long time. I’ve worked with some of these companies and they all know it too but like to have their inflated numbers so they do their best to only target things causing them problems or that are very very obviously not real, both in terms of bots and paid content.
It’s almost always been this way from the moment Facebook had ads and any social media site had an audience. You could spend $1mil on TV and movie ads to promote something, or you could pay $20-50k a person a year to create fabricated organic content in all the right places and get everybody to talk about things and like it. It’s been an obvious play from the beginning. And it does work very well. I’ve worked on things like that for things unrelated to politics and the returns were great. Back then you could even get by with a team of two people making multiple accounts and make it seem like you had thousands of people talking about and posting something. The sites love it too because they also can now inflate their numbers. You can do other minor things that add up on top of that too, like tweak the settings to show active users as counted over a longer period, like say anyone that logged on that day vs that hour, to dramatically increase perceived engagement.
It’s all very easy to game and when people realized that they let it happen because it was mutually beneficial.
Most of the trends are heavily botted, unless they support Trump or Trump like people.
Twitter benefits from their trends looking like more people are behind them as well.
When bitcoin went sentient.
Have you heard the story of princess Scota and the linguistic ties to Phoenicia?
https://youtu.be/4DS-ngVWge4
Not sure if this is exactly the one, but this guy has a lot of dank content like this
If the bots respond instantly, then reddit is running the bots.
There are two main options for getting the data, scraping or using their API. (the other options are being reddit and having the data locally, or reddit pushing the data to you)
The scrapers will have a delay because they are forced to poll at regular intervals to obtain the data. Your delay won’t be short because you don’t want to be banned for spamming their server. You could operate from unique IP addresses using a VPN, but if you are running a bot farm, it doesn’t matter.
If you are using their API, then you will be rate limited for the same reason. They don’t want to waste bandwidth or have their servers overloaded.
So no matter what, you will have some kind of delay. A minute or two, or at least thirty seconds.
If you are getting replies instantly, that’s because the bot doing the replying doesn’t need to ask for the data over the internet. They are watching for or being notified of it immediately. So either reddit is running the bots, or they are forwarding the messages to the bots as soon as replies are posted. There has to be some coordination.
Uh no, I’ve written bots as well…
You do if you want instant responses, because you need to know about them before you can reply, correct? Given that you need to first check for those replies, you have to be polling pretty frequently if you want your responses to be instant. Far too frequently to be workable if you are doing other things with your bot too.
Well if you are going to whine about it, reddit’s API limits you to 60 requests per minute. If you are doing actual bot like activity, that’s not a lot.
The reason you would poll at a fixed interval is to reduce requests… I mean I’d assume your bot is not just checking replies. It would be crawling, voting, checking replies, replying, posting, checking messages, among other things. Each of those is an API request.
Dude, my whole reply is in regard to someone saying that bots are responding instantly. So yeah, obviously you could delay responses to appear more organic, but that’s not what we’re talking about…
Not faster, instant. That is the word he used. Maybe it is an exaggeration on his part, but just for the sake of example, let’s say he is correct.
Don’t you think if you are relying on an API, you would need to do a hell of a lot of polling to have instant responses?
You can only do 60 requests per minute, one per second, and some of those requests need to be spent on performing other activities.
So at a minimum, you would need to interleave requests to check the comment replies to the bot at regular, short intervals, and you need to ensure that no matter what activities your bot does, it does not exceed the limit.
It’s not like the replies come at regular intervals either. Say you get five replies to a comment, well if you want to respond instantly, that means you would need some buffer so that you are always free to send a response without hitting the API limit. So you can’t even fit one action per second. You need to be far below that in case you want to do something more than just check the messages you are receiving.
I’m saying it doesn’t even make sense, from a bot writer’s pov, to have instant responses. As you mentioned, you would be far more realistic if you didn’t do that at all.
So then why would instant responses even be a consideration? Well they wouldn’t unless you were not subject to the limitations of Reddit’s API.
Just about everywhere is totally infested. It's the greatest problem facing public discourse.
AWS + immediate link clicks + not engaging in discourse/simplistic distractions = this is a bot/AI. The more time you spend on a site, the more opportunities they can collect ad revenues and data and people spend more energy on arguments than coming to a common understanding. The AI is using you as a learning opportunity in interaction.
People should never have started to talking to automated machines.
They are so good hardly anyone can tell the difference. It’s creepy. I caught it once only because it repeated itself once - and I have experience and training. It was shocking.
GPT-3 AI can more or less pass the Turing test nowdays. If it or neural nets like it aren't already being used for bot spam, they certainly will be.
I believe the site is crawled / posts are sorted or searched for by keyword. Then some guy in China (or wherever) making 1/10th of a penny responds.
Or...Reddit has a Ministry of Truth which functions on the same level.
Why would shills click links though? seems like a quick way to get your shills farms get infested with spyware and shit.
It might be used as a data point. If these are in fact bots ran by A.I, then they learn from experience. Opening a link, scanning the info, and responding with a comment gives legitimacy. For most users, they will believe that this bot is actually a person. It passes the Turing test. But the flaw in the design is the fact that the bots click on in without understanding it's a IP tracker.