This is a false dichotomy. If you care about apprehending criminals actually responsible for crimes you should care about this too. A reoccurring subtext is that law enforcement cares more about quantity than quality, but data quality is hugely important. These dragnets are the equivalent of a House TV Series body scan, and the arguments that characters in the show made apply here.
One of the consequences of a dragnet and too much low quality data is the risk of circumstantial evidence leading to the conviction of the wrong individual. The minds of investigators and district attorneys are susceptible to all the biases that plague any other field including confirmation bias. Project Innocence has revealed heaps of wrongful convictions based on low quality data, even when police corruption wasn't at fault. Dennis Fritz and Ron Williamson for example were convicted for murder based solely on bad circumstantial evidence while the real perpetrator remained unpunished for over a decade.
It's not just privacy advocates who should care about stopping these kinds of surveillance techniques. That is unless the dragnet apologists don't care about who is being punished for crimes just as long someone is punished whether or not they actually committed a crime.
What are the odds the "two people" are "both agents"? That's like the lamest off the record attribution ever. Also amazing that some random agent using Google broke the case. Apparently anyone could have figured out who DPR was.