"In iOS8, Wi-Fi scanning behavior has changed to use random, locally administrated MAC addresses Probe requests (management frame sub-type 0x4) Probe responses (management frame sub-type 0x5)" (e.g. http://arstechnica.com/apple/2014/06/ios8-to-stymie-trackers-and-marketers-with-mac-address-randomization/ ) This sounds like an interesting privacy enhancement and worth adding. (The currently cited case of shops monitoring all visitors is not as applicable to laptops/desktops as to phones, but the same kind of monitoring can be done for global tracking, e.g. from workplace to mall or airport to airport, or for targeting a specific individual.) (Filing against NetworkManager as the top-level component because I don’t know the underlying architecture well enough, feel free to reassign.)
Yeah, I've seen the posts on the wpa_supplicant lists about this, and it's much more a driver/supplicant thing than NM at this point. Upstream wpa_supplicant has indicated that they would like to default to this behavior once it has kernel support. So I'll move the bug over to the kernel for that. http://lists.shmoo.com/pipermail/hostap/2014-June/030405.html
Are you using this as a tracking bug? This really needs to happen upstream.
At first glance, this seems interesting...I'll Cc Johannes to see what he thinks -- he probably already has an answer in mind... :-)
As Dan said, there was a discussion on the hostap list, which was probably not the appropriate venue but IMHO neither is RH's bugzilla, so I'll not track it here. In general, I see no big issues with this, but programming the hardware for it might be tricky and will likely need driver support.