It’s comparable to the difference between still photography and video. At the time, almost any time you wanted to take a picture of anything specific inside a cell, you had to “fix” the cell at that position in time, and then get your labeling agents into the now dead cell in order to get an image of that one point in time for that one cell.

By adding GFP to the end of any protein you were interested, you suddenly had the ability to track almost any protein, in real time! The same cell, only with infinite timepoints now. Tweak GFP so it emits red light, and attach it to another protein of interest and NOW you can see if protein A interacts with protein B both spatially AND temporally.

And that’s just the start.

Of course, you can already use bacteria/yeast/immortalized cells to make metric boatloads of GFP… Though I can see the appeal of the purification-less “drip it and forget it!” method!