ballistic trajectory versus active evasive behavior.
A target with a 300mm frontal diameter, capable of evading at 10g, is expected to require 6 rounds to achieve a P_hit of 0.5 at 200m range, based on the parameters of your typical Phalanx CIWS. Said Phalanx weapon system might fire a burst of 1s duration in such an engagement, which is 75 rounds.
For a missile engagement, that's a problematic number, but for a drone that's probably only moving a few dozen m/s and carrying a small payload, it's an entirely satisfactory engagement.
Phalanx is an early 1980s weapon system. The Army is going to be fielding directed-energy SHORAD systems next year, equipped with 50kW lasers.
A few dozen m/s is probably a good estimate for quadcopter style drones, but low hundreds m/s should be easy for fixed wing drones, so at 200 meter engagement range (that paper makes it pretty clear that maneuvering targets are hard beyond that) there is only time to hit a few drones before they arrive.
So maybe drone swarms will work even against current CIWS.
Sure, but current drone swarms wouldn't.
A drone optimally randomly dodging at treetop height is a much more difficult target.