Here are three examples of poor software design.
First comes the Gimp. Select an area using the scissor select tool. Hit copy. It copies the entire image.
They do a lot work on the fancy fancy detail, but totally screw up the glaringly obvious: when one selects something and hits copy, one wants to copy the selected something.
Mind you, I found a moronic mess in my recently purchased copy of Photoshop. Go to Help in the Photoshop menu. Ask for help on Photoshop. Get help on Safari.
Ok, I’ve got it now. The Gimp is meant to be a freeware replacement for Photoshop. They’ve copied everything, including the idiocies. Open Office did the same with Word.
Fortunately, I could do my work in Aperture.
But Apple’s XCode cancels out their gold star. Do a search and replace in all open documents. It seems to work. Change the search term. Change the replace term. Search and replace again. Rather than doing what you’ve said, it changes the previous result.
E.g. Replace “(i)” with “01”: it works. Then replace “(ii)” with “02”, it replaces “01” with “02”.
In all three cases, the software has obviously being designed this way. Such errors would not have got part testing unless they were deliberate. I imagine arguments. Quite clearly, someone made a wrong decision. Perhaps they were trying to do something clever and screwed up the simple. Perhaps they presumed anyone who disagreed with them was stupid. Perhaps they couldn’t see the forest. Whatever, the error also lies with the organisation that permitted these wrong decisions to remain.