I remember the first interception tests of missiles had terrible success rate, but it got better. That's why you test them. Now Iron Dome has around 90% success

I remember the first interception tests of missiles had terrible success rate, but it got better. That's why you test them. Now Iron Dome has around 90% success rate at interceptions in real conditions. And it is hard to get the precise info but it looks like a lot of the failures happened at a time were a malfunction put the system offline. It is worth doing R&D on these if an attack is likely. If it is not, getting rid of MAD without having a replacement mechanism is scary, but in this crazy world and the current proliferation, MAD is becoming an increasingly dangerous bet.