Assuming that is the case, one would think it would be in NSA's self interest to be less stringent with their FISA requests so as to achieve a higher rate of rejections, thus creating an image of propriety for the court.
If memory serves, NSA seems to pride themselves on the low rejection rate. Not that there's anything wrong with that—if in fact it's due to stringent preparation—it just looks bad from the outside.
Even if every LEO is trying their best to avoid filing questionable warrants, they will end up being mistaken far more than 0 out of 48,642 times. What's the point of having this oversight apparatus if they provide no apparent oversight whatsoever?
"Reuters writes that the 1,457 requests made last year by the NSA and FBI were all approved"
1,457 < 48,642.
> "Even if every LEO..."
NSA != LEO. Do you believe their requests were not being made by lawyers with the relevant act open on their other monitor?
> "no apparent oversight..."
The court's previous rejection rate was .03%. .03% of 1457 is less than 1. IOW, this is the expected outcome even if no greater care was taken to write bullet-proof requests. Also, this:
"The memo did reportedly show an marked increase in requests that were modified by the court before approval: 80 applications were changed, as opposed to 19 in 2014"
http://phys.org/news/2016-01-evidence-bad.html
Regardless of domain, perfect agreement can indicate a problem.
But how would I refute the "Federal judges warrant rejection rates are very low because the law enforcement doesn't want to present questionable warrants in the first place" argument?
It seems to me that a place to start is to ask where the FISA warrants get reviewed - in regular (non-terrorist) cases there's all kinds of warrant review and challenge and argument before a trial, but I'm not sure that kind of review happens with FISA warrants.