What exactly is your understanding of networked fusion?
Unless there is a central processing unit that collects, fuses and subsequently re-transmits the fused sensor picture every platform will do the sensor fusion on its own anyway utilizing onboard and offboard sensor data. From my understanding of your post, you consider networked fusion as the exhange of raw data only, rather than completely pre-proessed tracks. As stated you need some pre-processing anyway and tracking is nothing but the correlation of successive detections. The primary question would be to what extend it makes sense to share raw data in the first place? Exchanging raw data makes, from my point of view only sense when your platforms sensors can't establish a track own their own for whatever reason or when your ESM fails to identify a specific emitter because it's unknown to your emitter library. I can't immediately think of any other good reasons for the exchange of raw data or detections.
As far as the fusion is concerned the source with the lowest data quality/accuracy is typically the bottleneck for successful fusion, though you may well fuse good detection/track data and exclude those that don't pass the quality threshold for fusion. The same concept could be applied when fusing offboard data with onboard data. A minimum quality threshold could be agreed as a standard by all parties and ownship transmissions are degraded to match a defined threshold if necessary. The otherway round your onboard processing may use different quality threshold if necessary rejecting those data that whiöe above the agreed quality standard, might still be considered as inappropriate for your own standards. So you cam effectively choose between fewer high quality data or more low quality data with all entailed consequences.