How to achieve 1 meter accuracy with Android Was: Re: Dropgang vulnerabilities

Punk punks at
Thu Feb 28 20:51:34 PST 2019

On Fri, 1 Mar 2019 01:17:11 +0000 (UTC)
jim bell <jdb10987 at> wrote:

> Depends very much on the city.  In the big (relatively) city closest to me, Portland Oregon, there are many hundreds of streets that 'leave the city'.  Sure, they could all be camera'd, but what good would that do?  The pictures could be stored, and would be, but how could it be known if any specific frame represents a "useful" image?  
> Eventually, some kind of AI could be developed,

	There's no such thing as 'artificial intelligence'. And the time isn't "eventually" but "right now".

	There is already ordinary software that does motion detection and image recognition which is all you need for surveillance. Or if you want to put it differently, all the hype about 'AI' is just more promotion of automated surveillance. 

> but I doubt whether this would find most activities which would be useful to identify. 

	I don't see many reasons for doubting that automated surveillance can work, or is already working. 

> Placing or retrieving a dead-drop would be one of the most undetectable events that could occur:   Drive to a block, get out, walk around in a large grassy area, bend over, pick up something, walk away, drive away.  How much video of this would surveillance have to catch to determine that the person surveilled was doing something suspicious...

	If a few people make short trips to some rural area, that's suspicious enough. If they get caught on camera wandering around that's prolly even more suspicious.

	But there's more. Locations are being used as a sort of private key. Problem is, an attacker may be able to drastically reduce the 'key space'. Sellers in a given city are likely to use a few selected places. So all the pigs have to do is make a few purchases to learn what areas the sellers use. 

	article states 

	"Potential dead drop locations can be identified and surveilled by law enforcement" 
	and then makes the bullshit claim

	"the number of potential dead drop locations is very high which requires a lot of resources to surveil" 

	but no, the number of locations isn't necessarily 'very high', and once general locations are known, surveilling them doesn't require a lot of resources. 

> particularly if they didn't already know something was going on.  
> This reminds me that 23 years ago, I first learned about "3M Louvered Film",     a plastic sheet product that prevents viewing at angles greater than a pre-defined amount.  It's now generically available.   
> This stuff could be placed over a car's license plate, so that a camera well above (or to the side, or both) couldn't read the plate.  To be sure, that's not necessarily an unmixed blessing.  While it effectively makes a car look like it doesn't have a (readable) license plate, that in itself might be considered suspicious.

	Of course, the moment the system detects a car whose plate it can't scan, cops get called. So that doesn't sound like a good move at all. 

	You could try using a fake plate, if you know the plate number of a car that looks like yours, but that's again rather risky.

	Bottom line seems to be, you can only use the car to do ordinary stuff like going to mcdouglas to buy hamburgers. Then walk to a nearby 'dead drop'.

> A few weeks ago, I realized that cemeteries would be excellent locations for placing dead-drops.  (no pun intended, but I'll take what I can get !).  There are few people who visit cemeteries, 

	so the 'anonimity set' is small, which is not good. Also, as mentioned above, cops make a purchase, learn that cemeteries are used - then surveil them. 


>but the idea of a person visiting a grave is very plausible.  And, some amount of 'searching' is to be expected, so it doesn't look suspicious.  Further, there are plenty of gravestones which can be used as markers for the placement and retrieval of those dead-drops, even more precisely than WAAS GPS or L1+L5 GPS.
	well at least there's that...

> The war continues.  
>                          Jim Bell
> ×
> ×

More information about the cypherpunks mailing list