I thought that the article did a good job of explaining the problem of changes in the world making things change appearance or shape, but not some other causes

I thought that the article did a good job of explaining the problem of changes in the world making things change appearance or shape, but not some other causes of icons not making sense. To me, icons indicating actions (save, print, etc) work well when they depict (wholly or partly) a physical object that is obviously used to enact that action. I assume that this is because our visual cortex is tuned to recognise physical objects that do actions, not visual representations of actions themselves. We see a depiction of a hammer and know that its for driving nails, but its hard to depict driving nails without depicting a hammer. You could presumably do it (i'm not a graphic designer) but you're making people work harder to understand your depiction. The problem is that sometimes changes in the world result in the recognisable physical object no-longer being recognisable and/or physical. This can happen we make an action abstract ("save to floppy" becomes "save to cloud storage"), virtualise an action ("pick up phone and talk" becomes "make skype call using laptop"), or merge a once-external object into a multi-function object ("save to floppy" becomes "save to internal disk"). (What others have I missed?) I don't know what to do about this. You could use a logo or graphical device/convention to indicate actions (e.g. skype logo indicates make a voice call) but you're also increasing the cognitive load on the user of the interface and increasing visual clutter.