Compress JPG

Not 80% of the detail, and not 80% of the size. It is a number that scales the quantisation tables the encoder uses to decide how much of each block of the image to round off, and the relationship to file size is steep and non-linear. Going from 100 to 90 typically halves the file and changes nothing you can see; going from 50 to 40 saves very little and looks noticeably worse. The useful range is 75 to 90, which is why the slider starts inside it.

Because the second pass compresses the first pass's mistakes. A JPG encoder works in eight-by-eight blocks and rounds off detail within each one; save that result again and it rounds off the rounding, and the block boundaries it introduced last time are now edges it tries to preserve. The damage accumulates in a way that is invisible for one or two passes and obvious by the fifth. Always compress from the original file, never from a copy you already compressed.

No. JPG puts its errors where contrast is highest, and a screenshot is nothing but high contrast: black text on white, window borders, table rules. You get a faint grey halo around every letter and a file that is often larger than the equivalent PNG or WebP would be. Switch the output format to WebP and you will usually get a smaller file that is also pixel-exact. This is the single most common way people make a file worse while trying to make it smaller.

Barely, and it is the wrong reason to do it. An Exif block with a location, a camera model and a timestamp is a few kilobytes against a photograph of several thousand, so stripping it is a rounding error on the size. It is worth doing for what it removes rather than what it saves, and Clean does it without re-encoding the picture at all - which compressing cannot claim, since it redraws every pixel on the way through.