The other side of this is apparently also true:
LLMs Reproduce Human Purchase Intent via Semantic Similarity Elicitation of Likert Ratings - https://arxiv.org/abs/2510.08338
Bots are getting good at mimicking engagement
251–260 of 312 posts
Re: Bots are getting good at mimicking engagement
#252Earlier quoted context omitted.
YouTube didn’t stop counting views with ad blockers. It was one blocker extension that added the view counter api to the block list.
That's splitting hairs, but in a way that's important to the conversation. If YouTube served gigabytes of the video file for 40 munutes and human watched it for that time, but they didn't send a request to `youtube.com/api/stats/atr/` and periodically to `/stats/qoe?`, did the video actually get viewed? I think a reasonable person would say that the person viewed that video. Only a programmer would suggest that wasn'…
Youtube might not really care about accurate view counting in that way. In fact, Youtube likely does not care at all about traffic that won't view an ad, and they have demonstrably been hostile to that traffic for a while. Someday youtube hopes that nobody with an adblocker ever views any video, and has no intention to track their view attempts.
If that causes a problem with Youtubers making a paycheck from external sponsors, Youtube really really does not care, because sponsorships are money that youtube doesn't get!
Youtube is downright hostile to creators who don't make the "right" content, which means new videos at a perfect schedule that all have the exact same "taste" such that viewers can mindlessly consume them in between the actual ad breaks youtube cares about. The more and faster people accept this, the more likely we get improvements.
It took over a year for youtube to agree to let people swear in videos without punishment, including videos that are set as "These are not intended for children", after they unilaterally introduced this swear downranking several years ago.
Youtube cares about Mr Beast and that's about it. If you do not run your channel like Mr Beast, youtube hopes you die painfully and think about your mistake. Youtube actively drives creators making actual art, science, knowledge, and other types of videos to burnout, because Youtube considers creators to be a renewable resource worth exploiting, because there will always be 15 year olds who want to become influencers.
It is not "deeply invasive tracking" or "programmer thinking", it's entirely business. Google's business is ad views, not video views. They want to measure what they care about
Re: Bots are getting good at mimicking engagement
#253Earlier quoted context omitted.
Google ads are often like paying someone to stand in front of your business and hand out flyers to everyone going inside. Sure, a lot of your visitors were handed a flyer by google before they showed up in your store, but were any of them new customers?
Google ads is bad but wait till you reach the ads on all the streaming platforms. Watching the same ad roll for the 20th time in the same evening is like the door to door salesman that rung your doorbell for the 20th time and kept loudly screaming about their product while you’re trying to enjoy a peaceful meal. If anything, I’m going to actively start hating your brand or product.
It's entirely intentional. Companies would rather you "hate" their brand than forget their brand.
The reality is that a shockingly high percentage of people just don't get upset about that. Pay attention to anyone using a web browser without an ad blocker. They are so deep in the learned helplessness hole they don't even complain as their web page takes ages to load and reflow while they want to read, and they just silently click skip on all three youtube ads.
The ads work great on them. Think of how many people bought HeadOn.
Re: Bots are getting good at mimicking engagement
#254Since we're talking about bot detection, any library or service recommendations?
I know of fingerprint.js which is semi open source but I have never tried them out. Any feedback?
Re: Bots are getting good at mimicking engagement
#255I’m puzzled by this: I thought it was well-understood, at least in the industry, that traffic numbers were at least mostly nonsense, and that ad click metrics especially were suuuper shady, typically more than half fraud; yet OP, in the business of “accurate ad spend analytics”, only just discovered this!? It just doesn’t ring true. That aspect of the story isn’t novel at all, and someone in that line of work should…
It's more a matter of "how much" personally, I knew it was not reliable but I thought it would be around 5-10% and recently more around 20%. I was far from thinking close to 75%!!
Re: Bots are getting good at mimicking engagement
#256So if I am running Facebook Ads and I get a load of fraudulent traffic, who is instructing these bots to make the fake clicks and why? Facebook? They have the motivation. They are getting paid per click. But I don't think they would dare. They would get sued into oblivion if they got caught. A competitor? They could be trying to burn up my marketing budget. But it seems it would require a bit more technical sophistic…
Look into methbot. Typically, it's people who game the algorithm (for ad dollars), and use these bots to click on other things (including ads) to make it more difficult for FB / Google / etc to detect all the traffic going to their pages and content is their own fake clicks, trying to train the algorithm to send other people to their stuff.
I read https://en.wikipedia.org/wiki/Methbot , but I don't understand how they were making money.
Re: Bots are getting good at mimicking engagement
#257Earlier quoted context omitted.
> Does it really matter if it's all fraud Uh, yes. If you get 47 sales on $10k in ad spend (pay per click) and $9900 of that $10k was fraudulent then you got 47 sales on $100 of ad spend. Imagine if you could stop those fraudulent clicks.
The point is, if $10k brings in 47 sales and that’s not enough then you stop buying those ads. It doesn’t really matter why it’s not working. You take your marketing spend elsewhere. You can’t stop fraudulent clicks just like you can’t stop your SuperBowl ad from playing while your viewers are in the bathroom. How much of ESPN’s viewership happens at bars where nobody is watching? At some point it’s not reasonable to…
> You can’t stop fraudulent clicks just like you can’t stop your SuperBowl ad from playing while your viewers are in the bathroom
Yes, you literally can stop fraudulent clicks. I'm not sure where you get the idea that it's not possible?
This is nothing like the Super Bowl analogy.
Re: Bots are getting good at mimicking engagement
#258Earlier quoted context omitted.
> Does it really matter if it's all fraud Uh, yes. If you get 47 sales on $10k in ad spend (pay per click) and $9900 of that $10k was fraudulent then you got 47 sales on $100 of ad spend. Imagine if you could stop those fraudulent clicks.
You can't just "ignore" fraudulent clicks as though those aren't baked into the original price. The only thing you should care about is your own sales vs your own ad spend. Your calculations make as much sense as pouring out 90% of a can of beer, claiming you have just made vodka, and simultaneously trying to pay only for the 10% of beer that is left.
> The only thing you should care about is your own sales vs your own ad spend.
You do understand that when I stop/reduce the fraudulent clicks that leaves extra ad budget available to be spent by clicks from real people who will actually buy things.
I have 0 incentive to not work to stop/reduce the fraudulent clicks. My ad spend stays the same but my sales go up.
Re: Bots are getting good at mimicking engagement
#259Earlier quoted context omitted.
> Does it really matter if it's all fraud Uh, yes. If you get 47 sales on $10k in ad spend (pay per click) and $9900 of that $10k was fraudulent then you got 47 sales on $100 of ad spend. Imagine if you could stop those fraudulent clicks.
And imagine how much more you’d have to pay for each of those clicks if everyone could stop those fraudulent clicks. In equilibrium it shouldn’t change the total ad spend.
The ad spend itself is not the issue.
I am fine with spending $10,000 on ads (or whatever amount).
The issue is that when I know $5,000 of it was spent on clicks that had 0% chance to convert. For every fraudulent click I can block/prevent that is one more click that can be made by a real person who may actually make a purchase.
Re: Bots are getting good at mimicking engagement
#260Earlier quoted context omitted.
Look into methbot. Typically, it's people who game the algorithm (for ad dollars), and use these bots to click on other things (including ads) to make it more difficult for FB / Google / etc to detect all the traffic going to their pages and content is their own fake clicks, trying to train the algorithm to send other people to their stuff.
I read https://en.wikipedia.org/wiki/Methbot , but I don't understand how they were making money.
Methbot was running the sites showing ads, so they got a cut. Doesn't transfer to manipulation on Facebook Ads, where I don't think that's a thing in any scenario.