Could the incoherent sum of baselines be used as anti-coincidence filter?

[caseyjlaw]

Seems that very strong RFI is often spatially incoherent. These events can trigger a huge mess of candidates, but are easily rejected by eye. Perhaps an anti-coincidence filter could help remove them sooner.
Idea would be to sum over baselines and take the amplitude to generate an incoherent, wide-field data stream. Then dedisperse and search for transients. Any detections are likely (but not guaranteed to be!) dominated by RFI.