I know of one perticular manufacturer that boasted about how robust their ECU is and how they have never had a failure. I then had a chance to work on one of their engines, another story in itself, and found no spark. With no interface device, yes, thats right, you needed to spend anther £110 on software and an interface device to connect a PC, additional to an already expensive £550, suposedly robust ECU, that requires a non statdard dedicated connector that you cannot buy seperately so have to buy the complete loom. Only to have both ignition drivers fail. This was one of the manufacturers engines and it was their instalation, so it shouldn't have failed, but it did.
Its a bit like my work. I work for a manufacturer, and see loads of faults and issues with our machines, and if I were to just take that into consideration, I would have to think that the gear we manufacture was shite, when infact, the problems we see is only a tiny percentage of the machines we have in the field that we never ever see because they run perfectly without a problem for years.
One other thing to consider is that OEM ECU failures are not uncommon, and why there are loads Network 500 agents around the UK at least. The actual number of failures to cars on the road is tiny, but thats because of the multi million pound research programs OEMs have.
Ibuilt my own ECU and that is most likely why I had so many problems I did. I smoked the P259 chip trying to resolve the FET blowing by seperating out the grounds. like rob says, you do it once, and you dont do it again.