Schneier on Security

Clive Robinson • March 13, 2020 7:53 PM

@ Jonathan Wilson,

Here is another article on EARN-IT that says its not as bad for encryption as everyone says it is

I wish people would stop touting the “lawfareblog” because it’s becoming like “brightbart for the establishment crowd”.

The last dozen or more of their articles I have read show significant bias pushing very close to falsehoods in one way or another with “opinions couched as truths” and all sorts of other little nasties to worm into your brain.

In short the old advice of,

If you are going to read the newspaper, first learn to read the newspaper

Is very much true of the lawfareblog, they have become increasingly appologists for the more draconian parts of the establishment. People realy should realise that the writers “realy try hard to put squirrels in peoples heads”. They do this by making things that are logical, sensible and straight forward look like they are anything but, and therefore not as it appears to be, but something that is realy the equivalent not of the “bear trap” it is, but some “milk and honey in the promised land” benign fairy tail…

So it’s not at all surprising to see that the article is from one of the worst offenders when it comes to actually pushing not toeing the “anti-encryption” line who has been subject to quite a bit of rightfull villification on this blog befor. That is it is non other than ex-NSA general counsle and advocate of weasle wording for lying to congress Mr Stewart Baker…

The only place he should realy be given house room is in the “snake pit”.

Clive Robinson • March 14, 2020 8:21 AM

@ Yet Another Sockless Pupppet,

Or better yet, as @Clive often admonishes, to “energy gap” the encryption, ie, pencil it yourself using an OTP in such a way that you leak no light, sound, etc signals.

I have said the OTP using pencil and paper and a little discipline is a secure “Hand Cipher” which is also about the simplest hand cipher to learn how to use. Also that it’s not dependent on any technology that can be backdoored, or easily evesdropped on remotely and that it will also work with any “human interface” at any communications end point, which some technological solutions may well not (or on which further down).

Importantly I’ve also made the point it is a “simple example” to get people to understand the idea about where the “communications end point” is and where the “security end point” is, and how they should be kept seperated via a suitable “energy gap”.

Because I’ve also noted as you have,

That’s a bit onerous if you are trying to communicate with more than one interlocutor.

Specifically in the Keying Material (KeyMat) Key Managment (KeyMan), where Key Generation (KeyGen) and Key Distribution (KeyDist) the required secure storage and all the process auditing is a mountain of work at the best of times. That is the OTP in it’s true form “moves the problem” from one place in the “system” to another.

However I have talked about the use of various Crypto Secure Digital Random Bit Generators (CS-DRBG). Back in the 1990’s I used a variation of RC4 and BBS in a “card shuffling” and “waggon wheel” configuration. Whilst it’s still secure for what it was designed to do, I would not use it for “Generating Pads” with these days. Times have moved on and so have algorithms, so using two or more of the “AES finalists” in a variation of “counter mode” and a “mixing function” along with BBS would be more appropriate these days. Which obviously needs some “computing grunt” which would be in a seperate “device” or “secure token”.

If you dig up my comments about “authenticating transactions not communications channels” going back to the last century you will find I’ve talked on a number of occasions about “secure tokens” and some of the issues involved with trying to make them usable in human terms, whilst trying to keep them secure, part of which is always “putting the human in the chain” as a kind of “choke point”.

The reason for this is, whilst it’s easy enough to design a keyboard and display system on a “secure token” that you can use as easily as a smallish mobile phone, the real usabiliry problem is getting the various plain/cipher texts in and out of the device into the communications channel end point device without compromising the “secure token” by extending the communications channel into it via a side channel of some form…

To see why this can happen you have to dig down to the lowest of the electrical physical channels and thats a “simple circuit” to see how it and all systems built on it can be compromised[0].

It’s been variously pointed out over the decades that RS232 serial through to full on USB these days can “just be plugged in” however few when making these suggestions realise just what nasties could also “go across the wire” and in how many ways, thus how difficult it is to prevent (hence mitigation by “energy gap”).

For instance take what most call RS232 as a short hand[1], it has the basic Shannon Channels of,

1, Local Tx to Remote Rx
2, Local Rx from Remote Tx

That “work against a common ground” thus you get a “two wire” as a minimum for unidirectional data to “three wire” for bidirectional data, and five wire when you add “hardware flow control” (CTS and RTS circuits).

But then there are a whole load of other hardware level Shannon Channels for “signaling” including in some cases “break” on the basic Tx-Rx channels. The standard calls the twenty Shannon Channels it has “circuits” to emphasize that they are physical wires and electrical voltages not any waveforms that convey other signalling or information. Few realise the standard alows for a secondary set of bidirectional communications circuits, data clocks, circuit direction changing as well as “loopback” testing.

The point is that these twenty basic circuits can all carry information thus can form “side channels” that work “out of band” of the assumed data channel (Tx-Rx circuits). Likewise other protocols used as standard like ASCII define “in band” control sequences that can also be used as Shannon Channels to carry information overtly or covertly.

An example of such a covert channel that many ICTsec practitioners are aware of is seen in computer networking with secret “knock codes”[2] to open up ports that would normally be closed. However even several millennia ago Knock Codes developed into “Prisoner Codes” where by two people could pass information backwards and forwards through dungeon walls etc simply by tapping.

So when using an external token it’s undesirable from a security asspect to make a connection you can not see “covert channels” on, that could be leaking secret information “on demand” rather than all the time.

I could go on further but, hopefully people that have read this far, get the idea of just how difficult stopping side channels can be[3].

[0] A “simple circuit” is a wire and some kind of return path (usually but by no means always a “ground” or “common” return). You put a “Transmitter” (Tx) at one end (eg a battery and switch/key) and a “Receiver” (Rx) at the other (eg a buzzer or lamp). Such an electrical circuit, from an “Information theoretic” view point, is a basic “Shannon Channel”. By simple observation it can be seen that such a “simple circuit” has absolutly no security properties what so ever. That is if you want to make a secure circuit there are a minimum number of physical and information theoretical additions you have to make. Such additions are by no means perfect individually, thus you need to have multiple additions. However the problem is between each addition there are numerous security gaps of many forms which can result in any number of “side channels” that can carry information (see “phantom circuit” “party line” phones as one). That is whilst your wire may be just used by you for “Direct Current” (DC) signalling it can carry not just DC but also “Alternating Current” (AC) of various frequencies ‘from DC to near Daylight’ that cover the “Electromagnetic” (EM) spectrum as well as mechanical vibrations at audio and above frequencies. All such “energy” can have “information impressed” or “modulated” on to it in various ways not least of which is capacitive or inductive coupling that gives “cross-talk” on long lengths of cables that run in parallel, which is why “individual circuit pairs” are frequently made as “twisted pairs” to try and canncel cross talk from adjacent circuit pairs out (see a piece of Cat 5 UTP network cable).

[1] What most call RS232 is in fact a collection of entirely seperate standards that are frequently but not exclusively used together. The actual RS-232 standard defines connectors, and Data and signaling “circuits” including their basic electrical standards (grounding, voltages, slew rate and thus cable capacitance limits). But not how the circuits are used in terms of signaling rates/frequencies, or character sets/types/function and any error correction or framing.

[2] Just how old “Knock Codes” are is unknown but there are records of them being used going back several thousand years amongst “secret societies” and those in illicit activities, it’s even mentioned in Vatsyayana Mallanaga’s compendium on “life arts” the “Kama Sutra” from around two thousand years ago. The Knock Code most will have heard is the “Shave and a haircut” used in amongst other things cartoons. Knock Codes act like a password to authenticate people so they could gain admittance or give warning (early “duress code” etc). Computer network knock codes work in a similar way, you send a series of requests to one or more network port numbers, a script running on the remote computer will then enable say the SSH port for say a minute to enable you to log in then use another “secret command” script to keep the port open, otherwise the script times out and the SSH port gets disabled again. The idea was to reduce issues from “script kiddies” first “port scanning” then trying to “brut force” entry thus clogging up logs and resources.

[3] I’ve also mentioned in the past that there are various “end run attacks” an adversary can use even against pencil and paper OTPs and other hand ciphers. The simplest is a covert CCTV camera overlooking your “usuall work space”, so don’t have one and move to different places randomly. Remember a web-cam in the screen edge of a laptop will see quite a bit, thus working in another room would be sensible. But laptops like mobile phones also have microphones and your writing style may leak information, so remember to leave them in another room and shut the doors as well. But also laptops have WiFi etc this can be used as a form of passive radar that can “see” the movment of you pencil even through quite solid looking walls cellings and floors. If you look back on this blog I’ve described a number of times in the past how to design and build an “RF Cage/room” that you can add sound insulation to and rubber shock mountings to turn it into a home SCIF. However having a permanent home SCIF will raise suspicions in anyone who sees it, which is why I’ve also indicated how you can use common household objects that are not in any way suspicious to quickly build a temporary enclosed space that is shielded and uses “active” techniques to make surveillance harder.

Clive Robinson • March 14, 2020 4:48 PM

@ tfb,

You could really only prevent such a thing legally by outlawing general-purpose computing.

Why do it by a legislative ban?

After all as you note with your preceading,

so long as it’s sandboxed properly I think it meets app-store guidelines.

So what need to go with legislation when you just have to lean on those who realy own your phones and smart devices?

That’s the real point, whilst you can download a programing language like Python onto your “Smart Device” try doing something actually usefull with it and you will find that the “sandbox” of permissions makes it little more than a novelty for just about every one. The reason being you might just find a way to “root your device” and that is very much “VERBOTEN” by the OS designers, device manufacturers and service providers, as you might not “Render unto Ceaser…”

There is a prise to pay with those “walled gardens”, which always were in reality the lowest common denominator of a “Canary cage and Opium den”.

Freedom at the end of the day only happens when you yourself, not others take responsability for your safety.