Mac Spoofing preinstalled? - Qubes 4 rc4

58 views
Skip to first unread message

Tristan Fleming

unread,
Mar 3, 2018, 2:21:28 AM3/3/18
to qubes-users
Has mac spoofing been preinstalled in Q4 rc4? I cant find any documentation on it, but the cmd: ip link show wls1

changes every 10 minutes or so. I actually want to stop it and I cant seem to make it static. Lil help!!

Ive tried everything I can think of. Im not entirely understanding the network topology.

I have figured out that Macchanger is installed on the Debian Template. Im otherwise at a loss.

Chris Laprise

unread,
Mar 3, 2018, 6:16:01 AM3/3/18
to Tristan Fleming, qubes-users
Its not a Qubes feature. When wifi is not connected to an AP, its now
the Network Manager and wpa_supplicant default to randomize the mac
address. The mac address shouldn't change while you're connected though.

Here are a couple links:

https://blogs.gnome.org/thaller/2016/08/26/mac-address-spoofing-in-networkmanager-1-4-0/

https://www.qubes-os.org/doc/anonymizing-your-mac-address/

If you haven't tried already, it sounds like you want to change the
wifi.scan-rand-mac-address setting.


--

Chris Laprise, tas...@posteo.net
https://github.com/tasket
https://twitter.com/ttaskett
PGP: BEE2 20C5 356E 764A 73EB 4AB3 1DC4 D106 F07F 1886

Yuraeitha

unread,
Mar 3, 2018, 8:37:34 PM3/3/18
to qubes-users

In addition to what Chris said which I agree with, be careful if you're relying on this type of privacy. If it fails, even for a brief moment, then your real identity has been exposed. And this is very likely to happen.

If your goal is just general increase of privacy, and not absolute privacy with risks carrying in it (i.e. avoiding dictators, or police state, and who knows what other baddies), then you should probably think of something else, than take huge risks or risking your life. If you don't have any big risks on the line though, then yes, maybe it can be worth it even if it sometimes fail. I'm not an expert though, but all you need to know is that it can fail, and the above is the implications if it does indeed fail.

Even if the risk is, say, less than 1% every time its used, or even less than 0,01%, over time as time goes on, that risk will increase slowly, until it will eventually happen to you, the risks will keep going up. You won't even have to be extraordinary unlucky or lucky, it can happen at any time as it moves from almost 0% risk to almost certain 100% risk.

As will anything which has a low risk, if used many times, that risk will grow, until you almost certainly will hit you. So too is the same logic for why immortality is impossible, because even if you don't age anymore and is eternally young and healthy, the odds that something unnatural will kill you, like say getting run over by a car, increases as time goes on. Of course it probably won't be a car, it could be anything that eternal youth is not protecting you from, like getting shot or falling down from a high place, who knows, you get the picture. That's why, as I understand it, macspoofing is discouraged. Of course it could be useful, if you don't mind it failing once in a while, and only want mostly better privacy, and not fully protected privacy, i.e. if you only want to detour dodgy marketing and companies making profits of knowing everything about you, then this might be a valid strategy to reduce it.

So until that feature becomes 100% reliable, all the time, then it will always carry risk. The question is better asked, is it a risk you can survive getting run over by?

Reply all
Reply to author
Forward
0 new messages