It doesn't take much of a leak. Radiation likes to radiate.

I used to keep my work phone in a Faraday bag sometimes. (I had my reasons[1]). It usually worked. Occasionally, it didn't work and the phone would demonstrate this by doing phone-stuff like ringing even while it was snug inside of that conductive bag.

So sometimes, the radiation was radiating well-enough despite my efforts.

Not so long ago, I was chatting with someone here on HN about blocking RF at GHz frequencies using aluminum foil. I was sure that it would be trivial, and they were sure that it would be difficult. So I tested that.

I started pinging my phone on its LAN IP, and wrapped it in foil. I found that I could increase latency some and also institute some packet loss.

But I couldn't stop it altogether -- not with a sheet of aluminum foil, anyway. No matter how carefully I made the creases, pings simply kept happening. (Having satisfactorily demonstrated to myself that were right and that it would be difficult, I stopped testing at that point.)

---

So here in reality, suppose the [car's] cellular connection finds that it has a connection occasionally. What's to stop it from buffering data and sending it in batches during times when it works? A few dozen lines of code that's geared to that purpose, perhaps? Or maybe a few hundred lines, instead?

Not that the difficulty matters much; the software is all closed up and inscrutable.

If the value of batching data to deal with intermittent connections is greater than the cost of producing the code to do this, then it can be assumed that such code has been or will be written.

---

[1]: An abusive manager I had liked to turn on the tracking system that the phone had. I didn't mind being tracked while I was on the clock, but I placed a higher value on my privacy than on her ability to be a snoopy bitch when I was not on the clock. My Faraday bag solution was adequate for that phone, at that time, with that particular tracking system, and for my particular desires, and I had access to the system with which to validate the adequacy of this success, but it was by no means perfect.

That's just it - move in just the right spot where reflections combine in the right way, and it might be enough to get a ping. So the tracking would still be there just less reliable, with an unknown level of degradation. In the end you still wouldn't have any guarantees.

Modern cars sometimes have telematic units running off dedicated lithium ion batteries, so killing main battery might not do anything

Unfortunately for many modern cars that may make it run less efficiently and clean and have a rough start every time you do it for 30 minutes or more because many sensors are trained on-the-fly from a running vehicle and then the correct calibrator sensor values are then stored in volatile memory which is lost upon power loss.

I use to disconnect batteries all the time when fixing vehicles, but the last decade ive been avoiding it unless I have to because of how poorly new cars run afterwards. And people get really angry when you fix something on their vehicle and then go to drive it later and it hard starts and feels and performs worse than ever. Telling them to "just drive for 30 minutes and then restart your car again and hopefully it goes away" doesn't make people happy or confident in your fix, nor does it make diagnosing issues after replacing a suspected faulty module or sensor easier when it sounds and performs like trash for a long while afterwards.

Usually, the whole antenna is behind the rear-view mirror between the glass and mirror. Often glued together