Hacker News new | past | comments | ask | show | jobs | submit login
Stop Calling It “Military-Grade Encryption” (signata.net)
90 points by timothy-quinn 47 days ago | hide | past | web | favorite | 75 comments

Yeah, this part rings totally true:

> I actually do a double-take when a company says "Military-Grade", as I start to question how well they know cryptography, and if they may have picked an inferior algorithm thinking it's safe.

"Military grade" might sound good to marketing people, but to everyone that's actually involved in digital security, it sounds like it's made by people who have no idea what they're doing.

Another phrase like that is "patented technology" or "patent-pending". Maybe it sounds good to some people, but to to engineers it's a big red flag.

I'm fine with the phrase "military grade" if the combination of software and hardware is FIPS 140-2 Level 4 certified.

When it is FIPS 140-2 level 4 certified, they will let you know early enough so when it's missing you can easily tell.

Yeah, anything I've ever encountered that had certifications didn't say "military grade" or whatever, instead they always listed the certifications relatively prominently.

What part of the device is certified? The logic unit, the whole asic, the board, the appliance? Software too? Or what??

My worst red flag is "proprietary". It gives me an instant head ache because it makes me think of bad quality, terrible documentation and just over all pain.

In the context of encryption, "proprietary" means "home grown". Or rather "home groan".

Propretary is a huge red flag. Never roll your own crypto!

I've noticed this even with fast food commercials which say things like "our proprietary blend of spices" in a tone implying you should be excited about it.

I'm fine with proprietary spice blends. At worst, it doesn't taste the way I want. But with encryption, "proprietary" means "untested", and that's a really bad idea.

Military grade makes me think of symmetric key encryption which unlikely to be used in most of these situations.

In regards to "Patent Pending" or "Patented Technology" - these are phrases that are legally defined.

Neither are snake oil (unless used in a fraudulent manner..), and can generally be interpreted as the product in question is objectively innovative.

When it comes to cryptography, "innovative" generally means "broken in new and interesting ways".

If you want security, use something which is old and well tested.

It is a sign they are more interested in lawyering than engineering.

The salesmen at the company I work with at the moment (industrial/commercial IoT) have adopted "banking-grade" for laymen, and have learned to recite the algorithms and/or protocols for anyone who expresses further interest.

Stereotype incoming: This might be a cultural thing too. Military grade sounds like something that would play well to the American ear, whereas these guys are British.

When I hear some Americans about their banks, I'm not sure "banking grade" will leave a positive impression there.

Ya, if you don't have the world's largest and most well (over)funded military 10 times over, "military grade" doesn't sound quite as serious.

"Military-Grade" encryption. In my experience, this involves writing passwords on daily orders and on the big white board in the ops room, giving them out over the phone to anyone with a suitably clipped voice, and laminated guides to zipping files with a password - from which everyone then copies the example password they can see in the laminated picture - physically tied to "the encrypting PC" so that nobody can wander off with the guide.

On the other hand, we _could_ try to get to a world where "military grade" encryption means it's so fire and forget that actual military people won't fuck it up. Which would be very good for everybody else.

Nobody calls a truck engine "military grade" because it's hard to maintain and only works on artisan made nitromethane, a "military grade" truck engine has to be maintained by an ADHD kid who got mediocre grades in shop class and run on whatever counts as "fuel" in the country the truck is passing through.

Passwords for example, are a bad idea, so we want to design a system that doesn't have those. The password step being mindlessly replicated is a bad idea in large part because passwords are themselves a bad idea, if the instructions involved a system that randomly generated session keys the same mindless replication would ensure all recruits used randomly generated session keys - which would be safe.

It's this comment that should be laminated...


To me, military-grade means that you run it over a physically separated network and post armed guards who will shoot all intruders.

I will remember that for the next time someone calls something "military grade".

"So you've got armed guards ready to shoot anyone who messed with it?"

And the device itself can be run over by a car, dropped in salt water, and still work!

As seen in the Crypto Snake Oil FAQ from the 90s: http://www.interhack.net/people/cmcurtin/snake-oil-faq.html#... If the first 20-30 years of "please don't" didn't work, I am reluctant to think this post will help...

Military Grade iirc referred to the encryption schemes defined in the Rainbow Series in the 80’s and 90’s today it is a buzz word but I’m not sure it was always that undefined.

Well, until the mid-90s or so, encryption was considered military equipment, so it was kind of true back then.

Is there a version of this one but for recent years? Thanks!

I never really took it as a factual statement on how amazing the used encryption is. I always understood it as just another marketing term for the sales process. It probably sounds very good for a non technical decision maker and everyone who’s technical enough to understand encryption will look at the more detailed information anyway.

Military = Safety. There is a reason they didn't choose "Terrorist-Grade encryption"

Although that would probably need to be higher grade to work. Technically it would be the better name.

The best advertisement for me was that terrorists use Telegram and Russia wants to ban it.

According to several world powers, the encryption terrorists use is superior and warrants further pro-surveillance legislation.

What is funny is that some powers that be really want backdoors in current crypto, because of course terrorists are stupid and will use backdoored crypto just like the rest of us.

Seems like they still do consider strong crypto to be just like munitions, because they seem to think if you just make it illegal then the gargantuan black market supply will just dry up and ordinary citizens won't be disadvantaged.

Plus the stakes are high in the military. If encryption fails, missions may be compromised and people die.

The stakes can indeed be high. However risk exposure (impact multiplied by probability) is often less than people think. An operational order betraying a strategy, for example "take Wimbledon Common at 4am tomorrow," must remain secret only until the order is carried out. So (gross simplification) any encryption that takes longer than from now to 4am tomorrow to break will suffice.

The harder problem is guaranteed delivery, which is impossible using only technology. You might email that order to the General in charge, and if he's out on a toke and doesn't read it within 5 minutes forward the order to his 2IC. And if the 2IC is on the john your only recourse is to forward the message to the nearest permanently staffed guaranteed action point (GAP). The GAP will send out one or more despatch riders to hunt the recipients down in meat space and deliver the message.

In any other scenario the same security best practices apply as in the commercial world, albeit with much more money.

Edited for spelling.

I tried googling "permanently staffed guaranteed action point (GAP)." With no results.

Is this the militarys mitigation of the two generals problem?


It's more like reliable messaging (say TCP over IP) as opposed to best-effort messaging (where no guarantee of delivery exists). Best-effort messaging better describes a solution to the two generals problem.

Yeah... This is a purely marketing reason. You're not going to convince people to change that with technical arguments and everyone who cares also knows that the term means zilch.

While we're at it, can products stop using the phrase "aircraft grade aluminium" to sell trivial items such as forks, wallets and, I kid you not, Post-It note holders.

Yes, there are differences in quality possible with aluminium, as there are with most materials, but to make "aircraft grade" a main selling point for your sunglasses case is snake oil marketing nonsense.

Never know when you’ll need to whittle yourself a new aircraft component with you’re genuine hand forged pocket whittler after you’ve crashed your handmade single seater in the Tasmanian highlands.

Unfortunately there are specific alloys of aluminum that are used for aircraft, and some that are totally unsuitable for aircraft.

So there actually exists "aircraft grade" aluminum. It's not a quality thing so much as a question of physical properties.

This is just selling snake oil. I am not sure but this term is coined by people who are in marketing, who does not really understand the real strength and weakness of a encryption. Like a snake oil salesman they want to use a jargon to create an image in people's mind that they are extremely secure.

But in reality as the article points out AES, RSA and others are all used in military and all are military grade, but all have one or the other drawback if not used carefully. This kind of advertising is generally misleading.

Hopefully if someone challenge such claims in court will be nice. But I believe it will be hard to prove if the statement itself is untrue, its just that it is created with a spirit to deceive or conflate the meaning.

By calling it Military Grade, it's easier to Ban it...

Actually, that's what already happened. And now they're trying to do it again.


Posted on July 24, 2019 by Bruce Schneier on Security> Yesterday, Attorney General William Barr gave a major speech on encryption policy -- what is commonly known as "going dark." Speaking at Fordham University in New York, he admitted that adding backdoors decreases security but that it is worth it.


In 1998, the EFF and John Gilmore published the book about "Deep Crack" called "Cracking DES: Secrets of Encryption Research, Wiretap Politics, and Chip Design". But at the time, it would have been illegal to publish the code on a web site, or include a CDROM with the book publishing the "Deep Crack" DES cracker source code and VHDL in digital form.



>"We would like to publish this book in the same form, but we can't yet, until our court case succeeds in having this research censorship law overturned. Publishing a paper book's exact same information electronically is seriously illegal in the United States, if it contains cryptographic software. Even communicating it privately to a friend or colleague, who happens to not live in the United States, is considered by the government to be illegal in electronic form."

So to get around the export control laws that prohibited international distribution of DES source code on digital media like CDROMS, but not in written books (thanks to the First Amendment and the Paper Publishing Exception), they developed a system for printing the code and data on paper with checksums, with scripts for scanning, calibrating, validating and correcting the text.


The exposition about US export control policies and the solution for working around them that they developed for the book was quite interesting -- I love John Gilmore's attitude, which still rings true today: "All too often, convincing Congress to violate the Constitution is like convincing a cat to follow a squeaking can opener, but that doesn't excuse the agencies for doing it."



The US Department of Commerce has officially stated that publishing a World Wide Web page containing links to foreign locations which contain cryptographic software "is not an export that is subject to the Export Administration Regulations (EAR)."* This makes sense to us--a quick reductio ad absurdum shows that to make a ban on links effective, they would also have to ban the mere mention of foreign Universal Resource Locators. URLs are simple strings of characters, like http://www.eff.org; it's unlikely that any American court would uphold a ban on the mere naming of a location where some piece of information can be found.

Therefore, the Electronic Frontier Foundation is free to publish links to where electronic copies of this book might exist in free countries. If we ever find out about such an overseas electronic version, we will publish such a link to it from the page at http://www.eff.org/pub/Privacy/Crypto_misc/DESCracker/ .

* In the letter at http://samsara.law.cwru.edu/comp_law/jvd/pdj-bxa-gjs070397.h... , which is part of Professor Peter Junger's First Amendment lawsuit over the crypto export control regulations.

"Military-Grade Deception" is a legitimate term.

> I actually do a double-take when a company says "Military-Grade", as I start to question how well they know cryptography, and if they may have picked an inferior algorithm thinking it's safe.

While I despise the term as well, this seems like an exaggeration. It's simply marketing, nothing else. Developers implementing security aren't also making sales decks.

The thing is, real cryptographers and computer security experts never use language like this. You'll never here Daniel Bernstein (or whoever) call ciphers "military grade". Moxie Marlinspike never calls Signal "military grade", even though it's probably superior to most actual military systems. They know better.

What do you believe the thing is?

It’s just marketing wank. The majority of marketing is like that, to be honest.

Everyone is trying to over sell everything in every which way.

It’s wholly unremarkable that advertising drivel is completely out of key.

Like I said, these real cryptographers aren't the ones going out and selling the software. Marketers use whatever terms their audience likes to hear.

If it's a three person SaaS start-up or whatever, there's no guarantee there's a dedicated marketing team. And even if there is, the engineers/managers don't know better than to tell their marketers "hey don't use that kind of language". And the marketers selling a security product don't know better?

Sure, it's entirely possible that the product itself is absolutely solid. It would be a bit silly to rule a product out on this basis alone. But when evaluating a product, it's fair to "do a double-take" when seeing language like this.

Obviously it is all depends on the classification level of the platform in subject.

Military projects I've worked on, none of the known algorithms were in use. However, those are not available for commercial use whatsoever.

The author is right about one thing though, since if it is available in the commercial world, it is better off be called as "Industry Standard Encryption".

> Military projects I've worked on, none of the known algorithms were in use.

What do they use? I expected that us military would be 99% FIPS. (Other military could be GOST, I guess)

In a conversation I would interpret a mention of military grade encryption as meant to defend against nation state actor adversaries who have theoretically unlimited resources to mount an attack.

In that sense, I would expect it to be much more stronger than financial grade or consumer grade encryption system where adversaries are less stronger.

Certain attacks like supply chain attacks (factory, transport etc of hardware components) or special access attacks (Certificate authorities, BGP, DNS, ISPs etc) or social attacks (patsy or spy with MICE/RASCLS) that nation state actors can pull off which others cannot (without prohibitively significant effort or negative consequences).

So, usually if someone says military grade, I would look at it as being resistant to even these threats.

Of course, it is always about the system holistically and not just the AES, RSA, SHA-2 etc algorithms.

I've seen a lot of people refer to their practices as "Industry Standard" when they know what they're doing, and a lot of marketing teams use "Military-Grade".

I don't think this particular hill is one to die on - the public have a perception that the military are amazing at using the latest and greatest technology, so the marketing teams will always continue to use it. Fighting it will be about as effective as fighting clickbait.

Instead, let it guide your own choices. Any sign like this that shows the marketing team wrote the Security Policy page is one where you should absolutely start questioning if they know what they're doing.

To be fair to the marketing zombies, I kind of see why it would be easier to sell "Military-grade" rather than "Standard".

To our definition of "Standard" this means up to scratch and implemented correctly to specification - to the normal folk this is probably interpreted as "regular" or "basic-tier".

I'm a very important and aspirational small shipping/accountancy/dog-walking firm. I don't want that potato-tier regular encryption - I deserve super-duper fabtastic encryption like the banks have.

Maybe we should call it Fabtastic Encryption 12.0 (people like numbers after their software).

Double ROT-13 is for the plebs who can't afford better encryption. We do QUADRUPLE ROT-13! Take that, crackers!

Those are puny numbers. Don't talk to me until you're using ROT-26, or better yet, ROT-52.

Probably it is because I'm not a native speaker but I'm not seeing any difference between Military-Grade or Industry Standard Encryption. Can anyone please clarify what the benefit is? Are these well defined terms and does Industry Standard Encryption imply that it's continually updated?

> MD5 is thoroughly useless as a hashing algorithm[...] but it was used by the military and banks in the past, so it's technically "Military-Grade"

If it was used by the Industry in the past wouldn't the same hold true for Industry Standard Encryption?

Edit: quote formatting

They’re all just marketing terms and have no technical meaning. The author just wants the marketing terminology to better reflect the technical reality of all applications using the same technologies at any given time.

I suppose you can try to be clear that you maintain a standard as it evolves, that doesn't really work with 'grade' (but admittedly it doesn't stop you saying 'military standard'...)

Most of the software either claim it military grade encryption or bank level encryption, which is literally AES256.

Out of topic, my phone case - Spigen provide military grade protection too. :)

Am I the only one that gets a NET::ERR_CERT_AUTHORITY_INVALID when I visit the page in Chromium? Quite ironic, considering the title of this item.

You can use add-ons like Certainly Something (I don't know the name of an equivalent for Chromium but I bet there is one) to show you exactly what's in these certificates.

By far the most likely cause of an error like this (in which the browser clearly connected but didn't like the offered certificates) is that an intercepting proxy aka a MITM or middlebox is between your browser and the remote site and it fucked up.

Things that are intercepting proxies (some of which you might have classified wrongly as something else)

* Most 3rd party AV "solutions" or "endpoint protection" on your machine itself

* WAFs

* Any kind of "Next generation firewall"

* Government or ISP "filters"

All these products are pretty bad, and most are worse than useless. Recommendations to get one or more of them for "security" are probably this era's "Rotate passwords every 30 days" in terms of the actual security behaviour that results as distinct from what the policy proponent imagines will happen.

The EU has strict food labelling laws[1]. They stop companies making stupid claims like "100% asbestos free!", or claiming to be a health food when a major indicator is way out of line with guidelines.

I think marketing in general could learn something from that. Screen out ridiculous claims that have no basis on the viability on the product, and force people to focus on product and deviances from industry "standard".

For example, your transport layer might be using the same ciphers and key length as a military installation, but if Maureen in accounts can log in from home with the username and password she's used on every site since 2002 and access 400,000 customer details, and download them to an unencrypted file on a personal computer... You're not meeting GDPR obligations, let alone military or banking standards.

... And by tangential extension, I think non-developers might be surprised just how many companies still have absolutely zero effective access control to data. No storage encryption. No plans to warehouse or delete old data. Just records in a database (or shared spreadsheet) where a username and password, and sometimes just network access, will give you PII for every customer in the last 20 years.

[1]: https://ec.europa.eu/food/safety/labelling_nutrition/claims/...

"Military Grade" fails to clarify which military or from when, the enigma code was military grade but it wouldnt stand up to much now

The Caesar cipher was the best encryption available, once upon a time.

Isn't it just an expression by now though? Like how "weapons-grade" is used for many things that aren't nuclear material.

Anyone else getting a "Untrusted SSL Server Certificate" on the page?

If most other people were seeing it, someone would have diagnosed it by now.

Untrusted why? Dig into the browser's cert details to find out. In Chrome devtools (ctrl+shift+i), select the security tab.

The other person who mentioned an SSL error said chrome said untrusted authority. It's a cloudflare-issued cert, dated Tuesday 00:00, so there shouldn't be any issues unless you've disabled the Baltimore Cybertrust root which backs cloudflare's intermediate cert. Or unless there was a transient error with the cert cloudflare was serving. Like maybe they were serving a bad cert, but only at one of their POPs? While theoretically possible, it doesn't make sense that they'd have any untrusted certs around anywhere.

    depth=2 C = IE, O = Baltimore, OU = CyberTrust, CN = Baltimore CyberTrust Root verify return:1
    depth=1 C = US, ST = CA, L = San Francisco, O = "CloudFlare, Inc.", CN = CloudFlare Inc ECC CA-2 verify return:1

OP here. What browser/OS do you use? Our blog is run on Ghost so we don't actually control the TLS cert used, but we can try to find the cause of why it's not trusted and fix it.

Hi! John from Ghost here, had a quick look and can't see any issues on this side with the cert. It's a relatively new certificate, though, so there could be an old cache on an edge node somewhere which just hasn't fully updated yet.

I suspect this is a one-off, but please do reach out to us if you still have issues, I've already passed this along to the team to fast-track support if you do send us an email.

First guns then nmap.

Ok, I don't often comment on web design, but why is there the effect of shining an extremely dim torch at a dark page? I'm finding it very hard to read.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact