Was this designed by Goodwill Industries engineers? Sounds like a dumb ass screw up.
My guess is that this “slipped by QA” because it is intentional behavior and therefore not regarded as a defect. To test whether a WiFi network is truly active or not, the connecting device is going to have to transmit a packet.
What is the threshold for a “keep alive” polling interval? People in dense public WiFi areas walk through a kaleidoscope of overlapping signals. Is everyone’s WiFi device supposed to send a packet storm of TCP ack/responses in an attempt to connect to a newer stronger signal every 100ms?
I have a notebook with a netstumbler module running sometimes when I’m on the road which collects found networks into a database without discrimination. I see everyone’s devices broadcasting a network, from phones with enabled hotspots to distant laser printers in homes and businesses, RF guns at Petco, and even vehicles themselves are mobile hotspots straight off the showroom floor, like every Chevy Silverado even if the old coot driving it doesn’t even know he’s doing it. A typical cop car might be broadcasting four networks alone including the personal phone on a policeman’s duty belt.
You can’t just have a flurry of zillions of competing connectivity requests from overly talkative devices saying “Yoo-Hoo?!” ten times a second.