It's more plausible that someone who was on the same WiFi (or at the same location) was searching for Roadhouse.

Obviously Google knows who your coworkers are. Only one of them searching for Roadhouse (combined with some other matching critera) might be the reason you see it.

Sounds plausible. No matter what they say, Google "stickies" some searches to IPs, even when you're not logged in and clear all cookies. This is one of the reasons I started using a VPN.

How does a VPN help? Unless you're switching IPs constantly, Google will still associate the information with you.

Your actual identity is irrelevant, it's the "virtual" identity they've built of you that's important.

If you use a VPN client on your computer/phone your searches won’t be associated with other searches coming from your office.

I think most VPNs use NAT, so many users share the same IP address. Otherwise the VPN would be mostly pointless.

Are you suggesting your phone recorded you and your coworkers and then suggested ads based on that?

Maybe I'm not paranoid enough, but I find that.. improbable? Not out of the question, I guess, but still pretty improbable.

Maybe you could run some experiments and try and disprove the null hypothesis?

It's been extensively tested and debunked by many, many people and yet it's a myth that continues to perpetuate HN. The idea that these big companies or Big Brother are streaming your phone microphone data 24/7 is a lot more romantic than "the public wifi network I'm connected to is riddled with analytics."

The idea that your phone's microphone isn't constantly parsing speech into text and then occasionally uploading certain keywords to whichever major advertising platform (FB, Google, Amazon, etc) has an active app on the phone doesn't seem so far fetched. We know Android and Apple phones are generally always listening for their keyword phrase (Siri, Hey Google, etc) for voice-based AI assistants and that their speech-to-text support is very robust. Uploading a text log is very data efficient and becomes even more efficient if it merely sends a positive for certain keywords it's told to listen for based on the individual's demographic profile and advertiser's interest in said demographics. I've had enough "coincidences" happen to believe this can be explained otherwise.

We know for a fact that some phones are secretly listening and uploading stuff.[1-4] It's done by spying apps. How do we know? Because there are tons of researchers reverse engineering stuff trying to find this stuff. These researchers are extremely hungry for a big find, for some big publicity. They're searching very hard to find a big player doing it, but none of the big players are doing it, so they only make small headlines with the small players they find.

[1] https://www.zdnet.com/article/hundreds-of-apps-are-using-ult...

[2] http://christian.wressnegger.info/content/projects/sidechann...

[3] https://www.nytimes.com/2017/12/28/business/media/alphonso-a...

[4] https://web.archive.org/web/20190309034942/https://recon.med...

If Google were that smart, it would be pretty impressive. But I doubt we are.

Some coworkers and I were talking about the Kentucky Derby over lunch the other day, a random topic that nobody in the group really was previously interested in. Nobody searched it or typed it into anything. Later that day my coworkers were seeing ads for the Kentucky Derby. Anecdotal, yes, but very creepy.

Are you sure it’s not recency bias? Like when you learn a new word and hear it everywhere?

I use youtube for one esports channel, ever - I want the live streams when the events are on live. They can't figure this out. It never ever pops up when its live, I have to manually search for it. I wish I knew why exactly that was - either they know what I want to watch, but that I don't watch other things so they recommend other things in the hopes that I become a more regular user, or they flat out have no idea how to profile people except expressed interests from outside sources (which I think is the most likely option). It seems incredibly uncanny one moment and then dumb as a rock the next.

>(O)r they flat out have no idea how to profile people except expressed interests from outside sources

I think this is the closest to the truth if you meant that the way I'm interpreting it. YouTube seems to base its search algorithm around certain, broadly defined parameters that are chosen purely based on advertiser demand. If I watch one "meme compilation" sent to me by a friend, my entire recommended section will be meme compilations for weeks afterward. These videos grab eyeballs, particularly eyeballs in my demographic category as a general rule. However, if I search for car repair videos surrounding a specific model, I will never see anything relevant in my recommendations except for general "auto enthusiast" videos that target a very broad audience (broad in terms of the demographics of this hobby segment). It seems like YouTube is very focused on serving videos that please their advertisers and that their major advertisers are only interested in very generic target categorization.

I'd say this is best explained by the fact that producing a compelling video advertisement is several orders of magnitude more expensive/difficult/labor intensive than simply choosing a set of keywords for a "click here to buy" ad to show up in google search results. This means microtargeting is effectively impossible as it just doesn't make sense to spend $15k+ filming a professional quality video advertisement (plus cost per impression/click expenses) just to sell a hyper-niche product like a manufacturer-specific diagnostic tool. That said, I am a huge fan of two specific channels which seem to draw a large enough and unique enough audience to get hyper-niche ads for things like capital pipeline maintenance equipment so maybe YouTube just doesn't have the volume of audience analytics to make such ads work on anything but the most obvious signals in the data.

None of that sounds like coincidence to me. Google detected activity on some topics around you (I'm guessing your coworkers searched Roadhouse, and I know Plex has Analytics), and decided you might be interested too.

Somehow I feel better for my bad English pronunciation, because even when I purposely try, I am unable to make Siri, Alexa, Google Assistant, or Cortana listen to what I am actually asking them about. This morning, for example, I tried to search “How to make pickles?” and ended up with “owls, too, may tickle” :-|

I have good pronunciation but a very deep voice and I am completely unable to make Google Assistant understand what I am saying. I think it just doesn't work for anyone who's outside of some idea of an "average" voice.

"Owls, Too, May Tickle" sounds like it could be the title of a childrens' book.

I'm guessing this is more because of google analytics being absolutely everywhere. If G knows your social graph and that your peers looked up movie X on website Y - that's enough to tweak your feed. Pure speculation.

That's the definition of big brother, I wish there would be at least a button to understand why we see certain things...