I recognise this as well. I write for a living. So I'll do lots of searches to cross check stuff. But if you search to quickly, or to 'weirdly', or whatever you'll have to pick out bridges or zebra's or whatever is the current fashion in Captcha.
the best one is "select the photograph containing a crosswalk". How am I supposed to know what a crosswalk looks like in each & every culture on earth?
I mean, as a human, you are expected to use context clues.
You don’t need to know the markings used for crosswalks in every place around the world to know what a crosswalk looks like based on its purpose. There’s only so many ways to create a pedestrian crossing across a street, after all.
If anything, that seems like an extremely appropriate choice for something attempting to restrict access for bots that wouldn’t necessarily be able to act on the same context clues and intuition.
This doesn't really cross cultural boundaries. For example, the skull and crossbones means nothing to Iraqis despite universally being seen as as sign of danger and caution in US
You're asked to point out the designated crossing area for pedestrians across a street. Sure, some places use crosswalk stripes perpendicular to the street, others use squiggles, others use lines on the sides, and some don't use any markings at all, but it should be plainly obvious to anyone, anywhere in the world where the designated area is based on there being some marking, or control devices, or literally people walking in the photo.
This isn't rocket science. Using contextual clues to figure something out is literally one of the most basic human abilities.
I share your frustration but I’ve come to learn that a lot of people don’t process things contextually and have an extremely difficult time with problems or reading that require picking up context clues.
I assume you don't have to answer correctly on the crosswalk question, you just have to answer the way most humans answer the question when asked... but I have nothing to back that up.
I'm not sure. It used to be that you could just select whatever as long as you do it with a mouse (so have human-like cursor movement). But latetly reCaptcha and hCaptcha have both been yelling loudly every time I didn't select one square that had a car or staircase or whatever it makes your look for even if that object is relatively small and easy to miss.
I think this is because the primary purpose of the excersise is AI training though.
