yes.

The easiest one is this image is tagged as bad, so the checksum for the file (basically a unique code for that file and can be computed by any one with a copy) will be instantly noted as against the rules.

Changing anything about the picture will reset the checksum, such as drawing on it, resizing it, etc.

Or they can use image recognition for more processing power, looking at random parts of the image trying to get a match against a bank of known images to decide if it's banned or not.

That's expensive. It's best reserved for images that have been preselected for filtering, such as by posts from someone who only posts spicy memes after 10pm, and doesn't tag their wife in memes but tags their wife in pictures of their kids.

Or they can rely on the spread of an image. If they crawl thedonald.win and see an image's checksum, and that same checksum pops up in 500 feeds within an hour, it's probably spicy and needs to be examined for "election interference potential."

That's why only your rarest or homemade memes that you do not share anywhere but Facebook get upvoted by your friends, and your shared mainstream memes go ignored. They're too lazy to have a human examine a unique meme for the damage it could be doing until it's reported, and it's not worth the processing power for deep-learning AI to check out the one-offs.

yes.

The easiest one is this image is tagged as bad, so the checksum for the file (basically a unique code for that file and can be computed by any one with a copy) will be instantly noted as against the rules.

Changing anything about the picture will reset the checksum, such as drawing on it, resizing it, etc.

Or they can use image recognition for more processing power, looking at random parts of the image trying to get a match against a bank of known images to decide if it's banned or not.

That's expensive. It's best reserved for images that have been preselected for filtering, such as by posts from someone who only posts spicy memes after 10pm, and doesn't tag their wife in memes but tags their wife in pictures of their kids.

Or they can rely on the spread of an image. If they crawl thedonald.win and see an image's checksum, and that same checksum pops up in 500 feeds within an hour, it's probably spicy and needs to be examined for "election interference potential."

That's why only your rarest or homemade memes that you do not share anywhere but Facebook get upvoted by your friends, and your shared mainstream memes go ignored.

yes.