Hacker News Reader

Comment by rdl | Hacker News Reader

Sounds like not "the death of public key encryption" but the golden age of building technical controls into hardware/software which cannot be subverted by the operator, even in the face of a state agent with a gun.

Assuming the right tech is developed and deployed, this is going to be far better for everyone in a few years. Yes, it will be shitty for a year or two, but by 2020, if we actually have real technical security, it will improve security and trust for end users. Rather than "trust us", it will be "trust us, because...".

View on HN

Those things you speak of will be outlawed and you won't be allowed to use them.

Harsh. dvmmh made a joke a few days ago and got downvoted to oblivion. But I think this comment is a good one: we do not live in a libertarian paradise. The government is comfortable using its powers, including physical violence under the law, to keep people from doing things it does not want them to do.

Sadly, that's exactly how things work. We tend to think we can find technical solutions to create legal loopholes, but they usually end up as contempt of court.

>Assuming the right tech is developed and deployed, this is going to be far better for everyone in a few years.

A, the old "let's use technology to solve a political issue" idea.

Sadly, it does not work. For one, the government has all the technology available for it too, including dedicated, full-time paid researchers.

Second, they can outlaw any of those things at whim.

I think it's important to do both. We can't solve political issues using only technology, but it can also be helpful to build our infrastructure in a way that would make it cost-prohibitive to outlaw these technologies.

For example, we could deprecate and eventually disallow TLS ciphersuites that don't provide forward secrecy, just like we did with single-DES, the NULL cipher, and "anonymous" (unauthenticated) encryption modes.

These criticisms of technological solutions always miss the obvious: both approaches complement each other.

If I send my aunt a letter, I am assured that unless there is a warrant, my letter to her will not be read by my government. She doesn't live in my country though, and her country gives no such assurances. More-so, even if it did, I would have no particular reason to trust her government. On that note, what if I don't trust my government?

Instead of sending her a regular letter, thanks to technology I have the option of mailing her a message encrypted with her PGP public key. Now I have to trust that my government will not beat me with a wrench or compromise my computer to gain access to the pre-encrypted letter (if I posses it), I have to trust her government to not beat her with a wrench or break into her computer, and I have to trust our governments to not outlaw PGP.

So which is preferable, finding political solutions to the "my aunt and I are being beaten with wrenches" problem, or finding political solutions to the "my letters are secretly being read without our knowledge" problem? I assert the former is preferable. Beating people with wrenches is a far more extreme action, it is easier to trust governments not to do something that is more extreme.

All of the problems with technical solutions are ultimately political problems. You can either abandon all technical solutions and only go for political solutions, or you can find political solutions to the shortcomings to technical solutions. Technical and political solutions complement each other, providing assurances that the other cannot. Both need to be pursued.

Don't want the government to read your mail? Encrypt it. Don't want your government to ban that? Lobby your position.

>These criticisms of technological solutions always miss the obvious: both approaches complement each other.

In a small way, yes, but in the long run, politics trumps technology.

You could achieve privacy from the government if you solved the political issues.

But the reverse is not true: if you had the perfect technological solution, they could make using the solution illegal -- including doing away with plausible deniability, even if the technological scheme provides it.

Or, if the government goes totally south, they could fuck you in other ways, making the technological solution totally irrelevant. E.g they could put you in a concentration camp. Where's your tech now?

>Beating people with wrenches is a far more extreme action, it is easier to trust governments not to do something that is more extreme.

Only if you leave in a western democracy (and are not a dissident, activist, or of any concern to the law).

In any other country , trusting the government not to do the extreme is not really that obvious. Heck, even in Western Europe, there have been 3 active millitary dictatorships during the last 40 years (and this is not even counting Eastern European countries, stasi and such).

I agree being in a truly lawless or evil state, even with high quality privacy technology, means you can't live a normal life. I'm pretty sure well before the "get in the trucks to go to the concentration camps" stage, it would be time to demonstrate that cryptographic munitions are not the only kind of munitions. There have been some promising efforts by EFF, etc. to bring the rabid pro-gun crowd into this in various court filings; people who will change states, vote single issue, etc. over being allowed to own a .50bmg rifle (when they could otherwise have a .510dtc anyway) are exactly the kind of people who will go to the mattresses over crypto. I think they also brought in pro-abortion people who vote similarly single-issue.

The US is far from that stage right now. NSA is developing absurdly overpowered capabilities, but is actually using them in a relatively constrained way, at least so far -- they are not being used for even high level criminal cases or really anything except actual terrorism. The problem is exactly the same technology could be used with a change of administration for political repression, domestic criminal enforcement, or whatever.

Since terrorism is maybe 1k people with any connection to the US per year, it's pretty easy to justify deploying privacy enhancing technologies to the remaining 300 million people, and there wouldn't be a politically feasible way to stop it. The battleground would be any extension of CALEA to pure-Internet companies, or worse, to end user technologies. Unless that happens, there's not much to fear in deploying technology overall, although in certain verticals (finance, with various FinCEN/AML requirements), additional regulations apply.

You really seem to be keen on not getting it.

What you are saying essentially boils down to "Don't waste your time with PGP because they could just stick you in a concentration camp." Being stuck in a concentration camp is a problem that needs a political solution (or in that particular extreme case, a violent solution...), so strive for one. Nobody is suggesting that everyone abstain from finding political solutions to political problems.

All the problems you seem to have with technical solutions are things that can be solved with political solutions. If you are advocating the possibility and pursuit of political solutions, as you seem to be doing, then why do you think the problems with technical solutions insurmountable?

Here is my proposal: Everybody use cryptography everywhere, to the full extent that we can manage. We then find political solutions to the political or physical threats to cryptography. If they try to ban cryptography, we fight back politically. If they start beating people with wrenches, we fight back politically. If they try to throw us in camps, we fight back politically (...and violently...).

There is absolutely no reason not to adopt technological solutions where they exist.

Situations like these probably rarely produce a real "winner". It is going to be an arms race between those favoring personal privacy and those favoring government snooping.

Just keep in mind that government operates basically on an unlimited budget and has access to a wide range of harassment opportunities for non compliance in matters like these. This fact alone will keep them at least at a dead level with potential counter measures.

No, it's a win because the government is not the only, and in most cases not the worst, threat.

Insiders (like Snowden and Manning, ironically) are one of the biggest threats. Being attacked by non-USG outsiders is a major threat ("hackers", state sponsored or not). Your business partners being hacked, or having lax security, is another threat. USG is probably near the bottom of the list of actual threats for most people, in terms of frequency; they just have the most amazing resources.

If the "evil NSA boogeyman" is enough to get people to finally care about security, that's great. It's like people preparing for zombies by buying water, freeze dried food, radios, shotguns, etc., when in fact they're at the same time preparing for the far more likely earthquake or civil insurrection.

> Insiders (like Snowden and Manning, ironically) are one of the biggest threats.

They're a threat to those who like running a Surveillance State, not to the average citizen.

Whistleblowers are a help to the average citizen, yes, but perhaps the parent poster meant "insiders with their own motives", which might not be benevolent.

"They" specifically are... generally, the point is that 'insiders' are probably not motivated the same way but have the same access to data which they could leverage against a target.

Insiders are the biggest threat to any organization. Doesn't matter if it is USG and Manning or that Icelandic kid and wiki leaks or whatever. Independent of whether the org is good or evil.

I always assumed "zombie apocalypse" was code for "civil uprising" anyway.

Even in Oakland, I am way more worried about things like an earthquake, blackout, etc than a civil war, though.

I can think of at least one counterexample, namely the failure of the NSA-promoted key escrow system in 1990s (aka Clipper chip)

And removing publicly available software crypto from the most serious export restrictions, at roughly the same time.

I've recently begun designing such security systems(a few weeks ago). They are quite ugly at the moment and would require entirely new approaches to systems design. I can only hope that more people begin to think like this as it is the best possible evolution in security past things like RSA encryption IMO.

In terms of hardware safeguards, is it really an improvement when we have to rely on proprietary hardware safeguards that may have government "special features" silently built in, vs open source software that can be inspected by the public?

The point is to make the hardware itself feasible to verify. There are ways to do this arbitrarily well. Traditional HSMs are a very bad choice for "interesting" users for exactly this reason -- look what Crypto AG did to the Iranian Government and others, and other vs proofed commercial devices.

I hope it works out similarly to how proliferation of public wi-fi led to increased adoption of https.

Did you know it is illegal to broadcast encrypted signals on HAM radio? Wonder how long it will take for that to come to the internet.

I didn't know that! I used to be a member of FidoNet (pre-internet network that provided e-mail and usenet-like services) back in 1990s and any commercial activity and encryption was similarly prohibited - relay operators were even supposed to read their users' mail and delete any messages that violated the rules.

I don't think there's any chance of this coming to the internet though. There are just too much uses for encryption besides conspiring for terrorist attacks.

Ham radio is a special case, and it is actually the Hams trying to block crypto. There is actually an FCC open rule making issue right now on the issue, and arrl is against it.

The solution is to develop protocols which use crypto but use a public key for now.

I saw that. Apparently issue is with self-monitoring. Also, would only be allowed for emergencies.

Just wondering - will older, but still decent hardware, start to become increasingly valuable since the hardware controls don't exist there? Will the value of my circa 2000 Dell PowerEdge take on new life?

I'm looking at ways to do an open source HSM using either trivially auditable/passive components (batteries, wire, wl gore fabric) and old cots components which you could buy from arbitrary sources, or buy a bunch of and inspect destructively.

Maybe we need a manufacturer to step up and make a line of computers that is open source and contains self auditing features such as checksums for all hardware. There has to be a way to do it. Some of the latest Kickstarter projects that are hardware based would be in a good position to try and pull that off.

Reply on news.ycombinator.com