Finding where the Captcha images were taken is the most rainbolt thing ever. Most people get annoyed doing Captcha whereas Rainbolt see’s an opportunity 🤣
The most annoying part is that how you click it doesn't even really matter because at that point it knows that you're not a bot so then you're just solving it now you have to prove something to yourself.
nah if this guy was a kid in my neighborhood i would hang out with him everyday, and i would feel like i have the most 200 IQ super genius friend in the whole world, we would have great times
First time I've ever beaten your time. Saw the thumbnail and immediately knew the location as I live in Copenhagen and have seen that exact spot a thousand times before lol
It’s the fact that he can also find interiors (the few times he exposed gatekeepers and the sandwiches they bought at certain shops) that makes him absolutely terrifying..
Super cool seeing you figure it out so quickly! I live in Copenhagen, so I recognised that it was the 6A bus (The A lines are the ones with the red strip on the left side of the bus), but you definitely found it much quicker than I would've done, even with me knowing the city and bus path beforehand.
To be fair, a robot could probably find it far more quickly. Google's reverse image search can pick your image out of the zillions of images on the internet. If given access to street view, I'm sure it could find the location in seconds.
Also: if you aren't aware, pretty much all captchas are used to train machine learning algorithms. Remember when you used to have to identify a blurry picture of some misprinted word in a book? That was training character recognition for Google Books. Now, it's training object recognition for Waymo's self-driving cars. It still works as a Turing test because it's a task that machines will be able to do soon, but can't do *yet*.
If you don't know which squares to click it doesn't matter actually if you click those small slivers or not... That clicking itself is used to train AI image recognition and it is not the actual test... The actual test for humans is how you move your mouse as you do it whether it is robotic or natural movement of the cursor...
that was true once upon a time, but no longer is the case. They use a bunch of heuristics including mouse movement, but image recognition is definitely part of the test now.
@@TrimutiusToo I know I'm late, but it's not exactly true. Obviously the test will not fail because you forget a pixel nor because you clic on an almost emplty case, but it can make the test longer in some case. I try a lot of things and it seem if the object is very big on screen, it's better to clic as few cases as possible, because selecting half the cases will make it suspect. If thz object is linked to others objects (for exemple someone is on a bike) it's better to select only the asked object (do not select the head of the person for exemple). When objects are only partially visible it's generally better to select them, but not always... You also need to be consistent (if you start selecting the few pixels cases on the first image, do the same on the next). Again, unless you randomly clic it will eventually pass, but if you do not respect thoses rules you could need to do four tests instead of one. It's also important to note that it heavely depend on your pricacy settings: if you are on incognito mode with an ad-blocker it will generally ask way more questions that if you have open cookie parameters without ad-block, because it can easily track you.
To complicate matters, I'm fairly sure there actually used to be a shelter at that bus stop. I lived about a kilometer from there back between late 2011 and early 2014.
I can imagine a world in which I send a selfie, and Rainbolt is like. "Definitely American trees, somewhere north could be [insert exact location] oh is that a cherry blossom? Oh that's just [insert exact location]."
Always wandered if I'd ever see something I'd recognise on geoguessr. Todays that day i guess, i cycle past here frequently... and i didnt even recognize it lol
Fun to come in and immediately see my own local bus there! Also you were a bit lucky, the area you highlighted for the search cut out the majority of central Copenhagen
It's important to note that whenever the blue button reads "skip" before you select anything, it's not actually a CAPTCHA. It doesn't verify your answer against any expected answer. Whenever the button starts with "skip", you are doing Google's work for free. In actuality, by solving these you tell Google that you're a good slave that does what he's told blindly, so they pile MORE pointless square selections on you. I find that the only sane way of solving reCAPTCHA is doing the audio CAPTCHA. You hear an awful voice recording, and you need to transcribe a single word out of it. That's it. Not the whole thing. Write one word, and you're through.
As a Dane, immidiately recognized this this being Copenhagen, just from the thumbnail, but it was kinda fun seeing him know it was Copenhagen and saying the buildings look right while looking at buildings there are thousands, if not tens of thousands, of in Denmark.
I found the house of 3 youtubers based on some footage they showed. One was a quick video full of jumpcuts to a Home Depot (or something similar), I was able to find the store in the county he lives and traced back the streets and finally found his house. The other one was just driving around his house a lot testing the breaks and sensors of his new car, that was pretty easy, the third one was a bit more tricky but showed enough footage of him driving around his car that I could see signs and building around his place and it was much easier to find his place. I didn't tell anyone about these, I don't want to publish private information on them but it was a fun challenge.
I googled some of the words that were on his screen when he used the tool. Found out it should be "Bellingcat OpenStreetMap search". I did not use it so far, but screenshots of that tool look exactly like it.
0:33 Captchas have these dubious blocks on purpose, because the humane uncertainty is exactly what it's training for. In these situations, the captcha will pass regardless if you select the dubious block or not, because there's sufficient number of responses for both cases. What the captcha wants to see is how people react to that picture. And knowing that, for example, 68% of people selected the background buses (or the corner of the crosswalk) versus 31% that didn't is exactly what it wants to know. It will only fail if you select a block that no one selects (thresholds are dynamic).
Yo rainbolt, i have a challange for you, this company Horde has put a million NOK somewhere in the woods, with a camera streaming the money. I bet you can't find the location of the box!!
VIDEO IDEA: Geogeussr IRL, have a group of friends escort you and take you to a random place in the world and you have to guess where your at, 3 rounds.
I really wonder if RAINBOLT is going to use his skills for crime investigations at some point, i really think he could make millions with that just by locating crime scenes
0:45 because the car already knows where it is, what it doesn't know is what it is looking at. Also you should definitely get in contact with the Jet Lag team for their next game of hide and seek.
0:37 Não esperava ver Santa Catarina em um vídeo desse. Mas especificamente da região de São José! (Identifiquei pelo fato da pintura do ônibus proveniente da Estrela que atua nessa região)
Ngl there should be a series of rainbolt getting kidnapped and flown into random countries where he gets released in the most random backroad and he has to guess where tf he is