Agreed, a system with false positives is worse than no system at all imho because it trains you to disregard warning signs.
I'm a bit leery about glass break sensors, to say the least.
Does Point try to learn a baseline for temperature/acoustics/etc., and trigger when a deviation occurs? Or do you just say "when glass breaks + these other thresholds are tripped, ..."
Because the latter would probably solve the problem fairly well, but the former would be outstanding.