I think a non distinct null parameter would be fine. My general argumentation is not about a closed argumentation system(standard) like I think you propose but by problems who are embedded into their context. I think good tools can be adapted to their context. It is not always easy and sometimes not even worth it. But you can argument about it. So my argumentation is that it is not important if it is sensible to the standard(curch ;-) but to the context it is used. So in my context a customization would be very useful. You can argument that my context is very special. Actually I cannot see so many cases there you want a unique index which is not constrained on null. And for that you can always use a partial index which makes it much more clearer. ;-) My point was not that the STRICT mode would change the behavior but would add some customization for null values. Anyway it was only an idea and I think I need to convince my company to pay for it which is not so probable. ;-)