Here in Germany, authorities treat IP addresses as very weak forms of circumstantial evidence due to the issues you mentioned (CGNAT, compromised devices). Nobody is going to be convicted purely based on an IP address.
(Disclaimer: I'm not a lawyer, exceptions apply)
Making a statement just to avoid an edge case is what we hackers like to call "boilerplate". We don't find it useful, and generally prefer to avoid it, both for the writer and the reader.
"This person seems really knowledgeable on law" which leads to
"Must be some sort of a lawyer"
They then go on to assume they have gained valid legal knowledge.
I have heard that even starting torrent program will get you fine to your inbox.
How do they trace it?
However, there's bunch of companies monitoring popular torrent trackers and since it's all peer-to-peer, they can easily figure out your IP address as soon as you start downloading a torrent they're monitoring (they usually work for a particular distribution company and have a list of files they're interested in). Popcorn Time and similar apps are a common and obvious target.
Once they know your IP address, ISPs are required to disclose your personal data (§ 101 UrhG) and they'll send you a nice letter requesting a payment (Abmahnung). If you do not pay, they might sue you (but probably won't).
If they do sue you, it's on them to prove that it was indeed you who used the IP address at the time, which is - in most cases - an impossible feat.
So, yes, you're highly likely to get in trouble if you torrent in Germany, but you'd probably win of you fight it.
You ban it, and they'll switch to something else, and you'll have done nothing but make things slightly less convenient but with relatively large cost to enforce. Why not just do the job properly and ban general purpose computers, limited to government approved and locked down hardware, os and apps?
One IP per person really doesn't hold.
Edit: this is incorrect apparently.
I don't have AT&T so I didn't pay much attention to the details. One possible approach: https://github.com/jaysoffian/eap_proxy
It has the same affect on you as your neighbor does.
Bandwidth would be unaffected. Or more exactly, on WiFi (but not Ethernet): as affected as if your close neighbor used that amount (since WiFi itself has limited bandwidth). The connection from the router to the cable modem has lots of headroom, so that would be unaffected, as would wired connections.
IANAL, but wouldn't a presumption of innocence require the prosecution prove that their evidence is not mistaken?
In theory perhaps, but in practical terms your mileage may vary widely. I'm not sure about the rest of the USA, but in the Chicago area Comcast appears to have pretty much abandoned rotating dynamically-allocated IPs, and I suspect many other carriers have something similar. On the other hand, experience last year with attempting to whitelist a remote user living in Germany showed that at least one ISP there appears to rotate home IPs every few days.
A quip. But also, I think, how things like this work.
You can ask the ISP. They are supposed to store metadata.
What you say is true, but prosecutors/investigators have little incentive to jump through another hoop. If they are willing to prosecute someone with just circumstantial evidence of an IP address, they aren't really looking to find the perpetrator, only a fall guy.
Similarly, if I torrent pirated media in a Starbucks, should the store be charged?
In my state, that was specifically to cover "well, if you deny driving the vehicle..." situations, where people would say "Oh, it wasn't me, I lent my car to someone/my kids drove/whatever", and law enforcement replying "Who?" "Oh, I don't know. Sorry."
Now it's to increase the burden of proof. It's reasonable to assume that if you own a vehicle, you were driving it. It's furthermore reasonable to assume that you should be able to identify who was driving the vehicle, from the face, even if you do have multiple people driving.
If you then refuse to identify the person, or "I don't know", then it's tantamount (in the law's eyes) to saying you weren't taking due regard to the care of operation of your vehicle (after all, if you don't even know their name, how do you know they're licensed?).
Computer crimes are the same way. It may be possible to prove what equipment it's implicated in a crime, but that's rarely sufficient to make any individual responsible without other evidence. People give WiFi passwords to their guests, equipment is hacked, etc. Establishing that an owner of a router or computer is responsible for anything it is used for unless proven otherwise opens a massive attack surface for blackmail, extortion, etc.
> Cars are stolen with significant frequency, for example.
Presumably you'd report your car stolen, no?
> People also drive them without permission (e.g. a spouse gives the keys to a visiting in-law).
Then you'll be able to identify the person via a picture of their face, most likely, no?
> Establishing that an owner of a router or computer is responsible for anything it is used for unless proven otherwise
I'm not trying to say any different, in fact I'm saying that it is a "reasonable" assumption, and as with any reasonable assumption there are notable exceptions, by which showing the driver's face is a reasonable method by which saying "No, Your Honor, the person driving my vehicle is not me". In some states you may not even be obliged to identify the driver, only show that it "is not you".
As outlined in the article:
* most residential IP addresses are dynamic
* more and more carriers are using cgNAT as they exhaust IPv4 addresses
I'd like to add a third case that I feel is often overlooked:
* most consumer or ISP provided routers never have their firmware updated (unless the ISP pushes the update themselves) and are probably vulnerable to mirai, KRACK, and many others.
Taking a these into account, even if LEO is able to get the correct information for an IP address, there's a non-zero chance that the people had/have a compromised device.
No average consumer is going to be able to prove to LEO or a judge that their device was compromised, and I somehow doubt doing forensic analysis of the person's network is top of LEO's list of evidence to gather. Additionally, many compromises can be ephemeral so by the time the deed is done and the police show up, the router has been rebooted and evidence of the compromise is gone. There is zero chance that LEO is dumping RAM of the router before seizing it as evidence.
Combine this with the fact that people may have an encrypted device and may legitimately forget the password before being requested to decrypt the device. I'm sure this can happen, as being arrested and going to trial can be quite stressful.
The US government is already indefinitely detaining people for not decrypting their devices. [1] I'm not saying the suspect in this case is innocent or guilty, but consider what precedent is being set there.
I don't see heading anywhere good...
[1] https://nakedsecurity.sophos.com/2016/04/28/suspect-who-wont...