Thanks @phnx!
I must admit I’m having a hard time thinking clearly about this. I’ve been over that Nature paper and I think it’s scary but I’m struggling to get a clear picture of what this means in practice.
Let me make up an example.
I start carrying a phone I bought with cash with a SIM card I bought with cash. I blithely go about my daily business carrying it everywhere, generating a nice trail of data points (from cell tower data) which doesn’t have my identity attached.
The paper mostly seems to talk about how dissimilar my trail will be from all the other trails everyone else is generating - four randomly chosen points will probably be enough to pick my trail out. It won’t at this point be known to be my trail, but those four points match my trail and my trail only.
So let’s say I use a credit card with my real name on it at four shops over the space of (say) a month. If we assume those four points are “as good as” random (which I think they probably are in this context), since the physical locations of the shops are known and my phone was turned on and its location being tracked when I visited them, someone who has a) the credit card data b) the shop locations c) my “anonymous” mobile phone trail can combined them and attach my real identity to the mobile phone trail, revealing my entire location history.
If I have that right, the key takeaway here is probably (as you said, I think - I don’t want to put words in your mouth) that it is surprisingly easy to do this kind of correlation. If you’d asked me before I read the paper, I’d have said “sure, you can do this - but you probably need quite a lot of data points”.
Where I am struggling is relating this to my threat model. I am not hiding from the government. (I might like to, just on principle, but it’s well past the point of diminishing returns for me.) I just want to be “left alone” and not tracked, profiled and monitored 24/7.
It doesn’t feel implausible that a random commercial surveillance firm buys both that “anonymous” location track and my credit card history and could tie them together. It’s just really hard for me to judge whether they are going to. Is it worth it for them, when 99% of people are just generating a constant stream of data directly linked to their identity? The commercial surveillance people aren’t specifically interested in me, and maybe I’m just slightly too much work for them to bother with.
I’d still like to thwart them if I can, and maybe this is sufficiently little effort for them that they are doing it anyway and it’s even more important I thwart them. But at this “they can, but will they?” level, it starts to feel more like guesswork.
It has occurred to me that except for a mostly theoretical need to be contacted urgently by family, I really wouldn’t suffer much from having airplane mode on almost 24/7 and relying almost exclusively on home and public wifi. If I really wanted to access the net and didn’t have wifi, it wouldn’t matter that much to reveal my location (from a KYC-ed SIM, let’s say) once in a while by turning airplane mode off for a few minutes - it’s the constant tracking that I really resent.
And maybe it would be good for my mental health etc to realise that I do not need to be contactable instantly 24/7 even by family - most of the time it wouldn’t be a big deal if I was out of touch for a couple of hours. And as I think THO’s video suggested, the reality is that most places I actually go would have free wifi anyway, so it wouldn’t take long before I built up a collection of wifi passwords so my phone was automatically connecting to wifi whenever I got to my destination, so I’d only be out of touch between leaving home and getting there.
Further thoughts/comments welcome, of course. I feel I’m starting to come to some sort of action plan which will give me the modest privacy I want, although I’m also feeling a bit annoyed that I have to work so hard to get it.