Agreement here.
We cannot just give up the power of augmented reality, and computer-augmentation in general. What I believe we should push for is avoiding the “cloud”, namely avoiding streaming data for offsite processing by third parties. Local processing should be preferred, even if it means external batteries. This does not address local recording issues, though, but that cuts both ways.
Luckily, live streaming can be detectable by analysis of the RF emissions of the device; constant stream of packets (with lower-tech (diode detector) manifesting just as carrier wave bursts) with certain size/timing characteristics will tell that with fair reliability.
I wonder if it would be possible to armor mobile devices with hardware/software detecting anomalies in its behavior, so an external compromise (e.g. the NSA software toys they have) would show up. Every such device would act as a “honeypot” - compromisable, but the compromise leaves traces and logs that can be analyzed and the device then can be “recruited as a double agent”, for feeding the adversary with false intel. Even the existence of such tech could deter the adversary from using remote exploits, from fear they’d get detected. (Does not help against conventional analysis of communication patterns, using email/phone data, which reveal the group topology and activity fairly well. This is however a threat not directly relevant to wearable electronics.)
The higher danger of surveillance by The Authorities lies in HUMINT; namely, in recruitment of agents from in-group or in infiltration of the group by police, and the resulting abuses of trust-relations. The affairs of British police vs environmental activists are well-documented, including an agent fathering a child with an activist.
Here is where technology acts not only as a threat but also as an equalizer. As it happened in Britain, occasionally the attempts to recruit an insider are recorded by said insider and then published. Photographs from the group settings can, when confronted with other groups from other time, yield intel about suspicious individuals (good infiltrating agents are a rare resource and have to be recycled; this may force them to be single-use at best). Then there are cases where cellphone videos from bystanders served as evidence that a cop lied (and said cop, instead of successfully pressing charges, got charged himself - see e.g. the older case of cop vs cyclist from some bike ride from I think NYC).
And then there are times, usually when things are actually happening, when live streaming (even, and perhaps especially, to third parties) can be a great benefit; a cop or other authority figure can ask you to erase pictures (which can be later recovered from the SD card using forensics tools), or seize the recording equipment together with data, but their power about data already saved on a third party server, possibly in a different jurisdiction, is, to say it euphemistically, greatly limited. Such recordings cannot mysteriously vanish or the cameras becoming “out of order”, as it happens when police-controlled hardware records authority misconducts. The keyword here is “sousveillance”.
The tech cuts both ways.
