Thank you for posting a well-balanced summary of the situation.
I think the biggest difficulty with balancing privacy concerns against other factors -- or even encouraging debate about and awareness of the issues among non-technical friends and family -- is that an item of data is itself neutral. It is how that data is used or combined with other data that may or may not be in any given party's interests.
Sometimes the exact same technology or data could be used for very good purposes, useful and generally harmless purposes, or hostile purposes, depending on the context. For example, consider automated number plate tracking of motor vehicles. If your child has been kidnapped and a witness caught the plate of the kidnapper's vehicle, you're going to appreciate the police being able to find and intercept that vehicle as quickly as possible. If you're an urban planner responsible for keeping transport infrastructure as efficient as possible in the face of a rising population, an aggregated data set showing how real travellers want to move around your city could be very useful, helping you make decisions that improve the system for everyone. If you're that same urban planner but on the side you're working with a load of jewel thieves and abusing your access to historical movement records to figure out when specific wealthy residents are usually away from their homes so the thieves can break in and then abusing your access to real time tracking to confirm that the residents really are out or warn if they come home early, that's not such a happy ending for the data subject. Analogous issues arise with many kinds of personal data, including more sensitive areas like financial or health data.
In the modern world, with vast databases and powerful data analysis tools and effectively instant communications and effectively unlimited storage, sometimes seemingly innocuous data can also give away a lot more about you than you might want or need. Things like what you bought at a store, or who you were tagged with in photos on social media, or a recording of you walking across a street on CCTV, can be used to determine many apparently unrelated things about you with relatively high (but, significantly, not complete) reliability, again sometimes quite sensitive ones that you might very much prefer to keep to yourself. Consider the store loyalty programme that determines a girl is pregnant from her purchasing patterns long before her partner or parents know. What about the person outed as gay because they were tagged with a whole group of openly gay people on holiday? Oh, by the way, unlike their friends in the photos, the first person lives in a country where homosexuality is still frowned upon and being open about it has real consequences. Also, gait analysis said you looked nervous as you walked into the airport, but it couldn't tell whether you just read a news story about a plane going down or you have inside knowledge of a terrorist threat, so unfortunately you won't be flying today. Just wait until automated text analysis software reaches the point that it can effectively de-anonymise posts like this one, and suddenly a lot of people who thought posting under a pseudonym was going to hide their criticism/whistle-blowing/advocacy of some controversial subject realise they weren't as safe as they thought and those comments are now permanently recorded on some public web site.
We therefore need to move beyond black and white assessments like "data collection is dangerous" or "social media sharing is fun". What matters is not just what data is collected, but also who has access to that data, what they are allowed to use it for (including for how long it's stored, what else it can be combined with, and the like), and crucially, how these things can effectively be controlled or regulated to ensure that everyone is playing by the rules when data is data and once someone has it for one purpose it can readily be used for another.