On the Lies about Self-Encrypting Drives (SEDs)
Robert Thibadeau, Ph.D.
2018-11-6
A serious paper by Carlo Meijer and Bernard van Gastel of Radboud University has declared a “self-encryption deception of solid state drives” which is, if the truth be known, a deception itself. I, as the inventor and originator of SED technology both in its SATA Security and Trusted Computing Group (Opal) forms (among others), have resisted responding to the lies that have appeared about the non-proprietary, industry standardized, forms of SEDs.
But not this time! Because the work out of Radboud U. is good work and deserves to be read for its truths. However, these authors also engaged in lies that need to be noticed by other competent people. The history of SEDs and good truthful criticism is also published in my book “How to Get Your Privacy Back” exclusively on Amazon and summarized in the attached short, highly relevant note “How to Get Your Lies Back.”
We see three levels of lies. One is perhaps excusable lies, one inexcusable lies to promote otherwise good research, and finally inexcusable lies by people wanting to make a name for themselves. The second two are described in the book’s chapter “The Power of a Lie: Confuting and Counteraccusation” which explains why lies are so powerful in the specific case of Donald Trump’s lies. It applies here as well.
To keep this note short, I will simply list the truths and the lies.
In Meijer & Gastel:
Truths: They have found weaknesses in the security hardware against attacks that would have been infeasible if the engineers at Micron had been properly supervised and informed. Samsung engineers did a better job and in fact had versions that were not inadequately engineered. Both Micron and Samsung have said they fixed their problems, as the authors note.
Lies: There are lies of omission and commission.
Omission: The standards workers have on thousands of occasions said publicly and in print that non-standard proprietary solutions, like software Bitlocker, should not be trusted any further than you can trust any software running in a modern OS such as Windows, Apple, Linux. SEDs were designed not to be controlled by OS software but pre-boot software. This giant lie of omission made by the researchers led to lies of commission of the counteraccusation and confuting class.
Commission:
“Data protected by bitlocker is also compromised” is a lie that says that bitlocker protects data. Bitlocker is in fact OS software managing encryption.
“Evil maid attack” is a weakness is not somehow now out of scope. It always was in Opal and SATA Security and in all software encryption as well.
‘Storing keys in software registers is why software is more secure than hardware” is a lie because ASIC controllers are comprised of CPUs running protected firmware with even better register protection. Indeed in most all of my implementations, the DEKs could not be directly read, modified, or written by the firmware; period.
‘Suspend mode will force key leakage’ does not mention that hibernate mode is routinely required by competent IT implementations of SEDs. Published standard documents state this best practice for over a decade.
‘Hot plugging attack is possible’ is a lie because, like evil maid key loggers, it is explicitly out of scope and this was publicly disclosed by the industry standards groups in 2002 or earlier. Software has these same problems.
“DEK left unprotected” is claimed for hardware encryption when, in fact, the authors cite appropriate controls employed by Samsung that effectively protects the DEK.
“Lack of entropy in randomly generated DEKs” is a lie. First, all DEKs, and KEKs should be randomly generated as specified in the TCG specs. There is a whole chapter in my book that not only says this but identifies a bunch of other weaknesses, more likely in software, due to non-random DEKs and KEKs. This includes backdoors that are impossible to detect in software or hardware implementations.
‘Bitlocker relies fully on hardware encryption’ is a horrible lie. Bitlocker violates the hardware recommendations by not using pre-boot isolation and the recommendation never to manipulate MEKs and KEKs in OS software.
There are more lies like these but the worst is that the authors assert that software encryption running in an OS is safer than SED encryption engineered under informed competent guidance. This is a plain lie excused as an opinion.
They correctly note, to nobody’s surprise, that encrypting data multiple times, with more than one encryption system, is safer than using one system alone. In further subtle lies of omission and comission they mention NSA,as untrusted, when it is well-known that NSA has long recommended a minimum of two independent encryption layers in their public written recommendations. Their reasons are not restricted to hardware encryption.
As I have said for over for over twenty years, SED solutions, properly implemented, are safer, cheaper, faster, less power hungry, less complicated, more assured interoperable, and more valuable than OS based software encryption such as bitlocker (see www.drivetrust.com, and my book).
But the real inexcusable lies came in the reviews, most notably the social media that followed the otherwise excellent Radboud article and the fair reporting in Techcrunch. As one mentioned, John Hopkins, not historically notable for scientific rigor in their reports, just made up much more serious absolutist lies that covered all SEDs as useless. To quote Techcrunch,” Matthew Green, a cryptography professor at Johns Hopkins, described the BitLocker flaw in a tweet as like jumping out of a plane with an umbrella instead of a parachute.”
It is not. Some parachutes made by some makers fail more often than others, some are engineered improperly and always fail, and none are infallible in a gale. This does not mean you should not use parachutes for safety.
The excellent researchers at Radboud U. made a substantial contribution to our proper oversight of commercial product safety claims, but they should not be forced into excusable lies by people who want only to hear the worst in overreaching lies. What this led to were the inexcusable lies, like Trump’s.
There are good lies, excusable lies, and inexcusable lies. Here is…
How to Get Your Lies Back
Robert Thibadeau, Ph.D.
October 6, 2018
800 Words
In my book, “How to Get Your Privacy Back,” I argue privacy and good lies of commission and omission rest on strengthening the three historical legs of the privacy stool: technology, law, and social contracts.
Philosophers in the 1600s in the Age of Enlightenment or Reason proposed “social contracts” to refer to informal contracts held between people and entities entrusted with valuable secrets. Social contracts had been disrupted by the printing press as people figured out kings lied a lot. Kings lost their privacy. Social contracts needed to be based in truth: in Science, Rules of Evidence in Law, and distributed in print. It worked.
Privacy lost and regained has happened not once but many times before. Lies, the opposite of truth, can destroy our privacy, but good lies protect it. We regain our privacy and lies through new social contracts that steer new technology and law, not the other way around.
Indeed, privacy lost and regained goes back to the evolution of natural language itself, to the beginning of our species. Natural language is the unique innate capability of humans. It evolved to enable Homo Sapiens to communicate valuable secrets to each other. Animals with a neocortex think like us, but they just can’t communicate their valuable secrets as truth or lies among themselves like we can.
People in one tribe could tell their children the truth about the tiger that they had never seen, and could lie about it to people in other tribes. The supreme evolutionary survival power of natural language in the controlled dissemination of truth and lies is plain.
The same natural language that takes our privacy can be modified to repair it even on the global scale brought about by the Internet. We can get our privacy and good lies back. However, we must use our language, in its naturally evolved way, in dialog.
Natural language dialog about acceptable truth and lies is how to get our privacy back. We already know that science and the rules of evidence are pretty good at helping that along, but new social contracts in “informal agreements” or what some lawyers call (overly broadly) “norms” are needed. We need active dialog between individuals and the trustees of our privacy. Existing fiat truth and lies should change on the Internet.
The fiat AGREE Button should be killed and replaced with negotiated privacy. This is a radical departure from the direction our parochial laws have taken us. It requires better technology than a single “If-THEN” one-way demand in a web site: Did he AGREE to an incomprehensible contract or NOT? But we already see from Alexa chat, AI natural language technology could provide all people and their trustees with automated natural language agents to negotiate acceptable privacy and lies.
Negotiated privacy should be claimed with our right to privacy. Negotiated lies should be claimed with our right to lie, too. New social contracts can arise out of natural language dialog, as they have in the past.
My book makes predictions on how technology, laws, and social contracts will change In the Age of the Internet. The major prediction is a new system of global law, based on the existing edifice of global telephony, that puts malicious lies in Internet jail, not people.
Putting people in jail is a sovereign right of countries, but putting lies in jail, should be globally managed by its own system. So, for example, if a lie is adjudicated to have killed people, that legal system can put the lie in jail, and perhaps sent off for academic study. Such a legal system can determine if a person or entity should also be publicized as the originator or distributor of the malicious lie, whether it be a lie of commission or omission.
This Internet Legal System is new. Privacy laws should enable negotiated privacy and negotiated lies of commission and omission, fully automated, not privacy and lies by fiat. Otherwise, history shows, we can be locked in the dark ages for a thousand years, dickering endlessly about the details of AGREE buttons put to us by fiat.
The conversation should not be among lawyers, companies, or governments, it must include everyone. The core to getting our privacy back, and getting our lies back, are the social contracts we all must informally develop in our natural language. Law and technology must actively promote the development of new social contracts by promoting negotiated privacy and negotiated lies, just as social contracts did in creating new technology and law for the kings and their vassals in the Age of Enlightenment. We must use our natural language, not some artificial version of it, for strengthening our social contracts and turning our social contracts into the simplified artificial languages of Law and Technology when possible and appropriate.
Four Principles of Human Privacy
- Human Privacy is controlled by Technology, Law, and Social Contracts, but Human Social Contracts come first and rule.
- Unlike Poison, too little Technology or Law and uninformed Social Contracts, will kill you, but enough will save your life.
- Like Poison, the bad guys are always using more than enough Technology and Law, and confuse your Social Contracts, to kill you.
- Getting your Privacy back is the same problem as Getting your Lies back. Good lies protect your privacy, bad ones will kill you.
Robert Thibadeau, Ph.D., is an Experimental Psychologist in Memory, AI, and Machine Learning, one of the founding Directors of the Robotics Institute at Carnegie Mellon University, and Inventor of a number of Security and Privacy technologies including “self-encrypting storage devices” now used by over a billion people a day for privacy preservation. His privacy policy book is available exclusively on Amazon and Audible; just search his name, “How to Get Your Privacy Back,” or www.privust.com.