Several reports have suggested a temporary loss of power to three engines, which just strikes me as an odd number. I could see one, two (same wing), or even all four being subject to software or otherwise issue, it is just that three, if confirmed, strikes me a bit more odd or unprecedented.
The problem is we know nothing of the actual software fault. It could be that the gitch rolled back power on the engines one at a time and then when the third engine rolled back, the software switched to an "alternate law" (or rough equivalent) that prevented the fourth engine from rolling back and then the other three engines starting rolling back up but not in time to avert the crash. Of course this is pure speculation on my part and is provided strictly as an example of how three engines MIGHT roll back and not the fourth.
Separately on Boeing installations, the engine software queries a pin to determine where it is installed on the aircraft (#1, #2, #3, or #4 position). If this pin was not correctly installed during assembly (the cause was described as a final assembly quality fault) perhaps all four engines thought they were in the #4 position and three engines rolled back when the data they were getting stopped making sense. Again pure speculation on my part. But the bottom line is that when software is involved there could be any number of explanations of how the engine software interacted with each other and with aircraft software to cause a three engine power roll back.