Chapter 17 It Would Be Called Milcode
The house might have been divided, but both sides seemed to be working towards what I hoped was still a common cause, and an ending that would be happy for all of us. While the southern tribe worked furiously and late into almost every night to come up with some kind of working proof of concept, the prototype of a dream, my attention was drawn to money and plastic. When I wasn’t chasing funds and fame, I was focused on finding a housing, a case, and a home for our new phone.
One of the few things we all seemed to agree on was that if we were to make our mark quickly on this thrilling and mysterious and secretive new world, we would have to be the best in every way. We could give no quarter. And that included looks.
Because development had accelerated much faster than we expected, one of the biggest pressures was turning out to be less about the phone and more about what it would wear. We wouldn’t really know the final dimensions needed for the housing until the insides were completed and ready for production. But if we only started the design, the tooling, and the production of the plastic phone case at that stage, we would be pushing the launch data out by at least six months. And possibly lose the Irish army contract and our very first order and ally.
But there were no companies making telephones in Ireland and few anywhere making phones whose housings would be big enough to fit what we eventually expected the space our insides would need. My biggest hope, and perhaps my last, was the voluminous Hong Kong Trade Directory, which listed hundreds of manufacturers of all kinds of electronics including telephones. But no matter how many and how much I begged, all required a minimum order of a thousand pieces as a first date and perhaps months of waiting time before we’d even see them.
I was stuck, I had no answers. But I knew that I only had to wait a while and they would manifest. As if aware of yet another gathering storm in need of diversion, the magic once again steered me towards safety. Even if it nearly ran me off the road.
The few enough days that I would visit my parents’ home in the suburb of Leopardstown, a former leper colony, I would take a shortcut through the Sandyford Industrial Park, my former playground. When I was a child it was like so much around us then, a collection of rolling fields and fraying old horse farms as far as the eye could see. A gentle and endless green playground where we would wander and explore or just trespass to take the shortcut home from school and past the reservoir when the days were warm. Now it was just a sea full of breeze block businesses and warehouses, including the one owned by our first investor and my older brother’s employer.
But what caught my attention on that particular day was a giant poster of a drop-dead gorgeous telephone in the front window of the office of the Irish division of a British telecommunications company called Plessey Communications. I whipped the lumbering Volvo around so fast I nearly ran over the kerb and into a bus shelter, and I pulled up to a stop and just gawked.
Shea be damned, I muttered. I’d found her. Even though it was just a photo of her, I could immediately sense that not only was she probably the one, but we might never break up with her. Assuming she was interested in us. But even that didn’t matter. We’d make her.
Now I just needed to find out how we could meet. Plessey was a part of the giant GEC Marconi Company of Great Britain and had just launched that very attractive-looking and expensive ISDN desktop telephone system depicted on that poster facing that street and my Volvo.
Unlike the traditionally shiny plastic round edged toy-like phones that seemed so popular in spite of their awful looks, the Plessey ISDN phone was sharp-cornered, matt-finished and wedge-shaped, and seemed like it would be the perfect fit for a very contemporary and highly-priced secure phone.
The color was a very gentle gray green mix, not unlike the Captain’s uniform, and I knew instantly it would speak well to most of our conservative and gray green customers. White would have seemed inappropriate and amateurish, and red perhaps a little too cliched. And because it was an ISDN phone it had all the same data ports that we could use to encrypt faxes and data too. All the right parts in all the right places and such a beauty indeed.
The closer I examined the phone, the more it became obvious that whoever designed her must have been thinking about us too. In addition to the main 12-button keypad, the phone was tooled to accommodate an additional thirty keys and giving us all the functionality we would need now and probably well into the future.
It also came with a backlit 24-character alphanumeric two-lined LCD display that would also be perfect for displaying the numerical handshake that would confirm the true identity of each party on a call on our future phone. Truly perfect in every way. Truly another sign. Now we only had to determine if she was willing and available.
The flags of luck were blowing strong and stiff in the wind and I’m sure it annoyed Shea to no end because I was delivering almost perfectly on some of our greatest challenges, and throwing a little extra in as humiliation. It turned out that Plessey also had a sister, nearly a dozen in fact, but my attention was immediately drawn to Plessey Crypto, another manufacturer of secure telephones and based in Wavertree in Liverpool.
Plessey Crypto had hundreds of military and government customers from around the world thanks to the reach of the British Empire, but had so far failed to develop any kind of phone that would rival the STU 3 from the Americans. Just 18 months before we met, Plessey had merged with the giant GEC company to create Britain’s largest telecommunications company.
The ability to compete with the Americans in the growing field of secure communications was in part a matter of national pride for them too, and so they agreed to sell us their beautiful ISDN phones, even just a few at a time. And even if our stated goal was to tear out their very expensive insides and just keep that beautiful but completely plastic facade. As long as we agreed in exchange to consider the better looking crypto sister as a potential partner and future home for our new secure phone that would now look exactly like theirs.
Back in the crypt, Shea had split the development of the phone into small teams, of just one or two on each, all overlapping and with Shea as the common connection, the foreman, the salt master.
We had decided, or at least I had decided and shared the edict, that in order for such a small startup on such a small island to be taken seriously in this dangerous new world, our product had to be superior in every way and down to every detail, from face to soul. We couldn’t give our competitors any reason to suggest that we just got lucky. Even if that were true.
Brendan had logically suggested that the three core elements to focus on, our path to glory, would be the speech coder, to analyze and synthesize the human speech and prepare it for encryption; the encryption engine, to encrypt the user’s speech and manage the keys that would make each call completely secure; and the modem, essential for sending that now compressed and encrypted data stream in a usable way over even poor quality analog phone lines anywhere in the world.
There were other less challenging but equally important technical tasks to complete. We would obviously need to build our own original telephone from the foundation up, component by component, and make sure it had all the normal functioning of a telephone so that it could make and receive calls and work as a perfectly innocent telephone when not in secure mode.
It would all need to fit on a single and compact printed circuit board. Shea concluded that because different processes within the phone would need different power and fluctuating electrical inputs, we would even have to build our own power supply to accommodate the differences.
The most important team and focus was the development of the speech coder. In order for the phone to not only work and work well, but to be a first of its kind and better than anything else ever built, the quality of the heavily encrypted speech coming out the receiving end had to sound no different to an ordinary local call to your parents on a weekend and on any telephone. A massive challenge given the constraints of speech synthesis and digital encryption in an analog phone world, and no wonder the Americans and everyone else seemed to be struggling.
In order to achieve that ambitious goal, we would first have to digitize the speech, the user’s voice, turn it into data, so that it could be put through the highly complex process of analysis and synthesis. Then compressed so that this massive amount of data could fit down mother’s phone line, and then encrypted to such an advanced degree that even the NSA would be powerless in any efforts to decrypt and eavesdrop.
At the other receiving end, the reverse process would have to take place and without any noticeable delays or reduction in the quality of the voice. The phone conversation, now a digital datastream, would have to be decrypted using a complex process involving the factoring of massive prime numbers.
The data would once again have to be analyzed to extract the key elements of the voice, the pattern that would make it unique and identifiable to just that user, then expanded and returned to its apparently original form.
Except that what the user at the other end would hear was not the real voice of the other party but the voice coder’s best attempt at reproducing an indistinguishable digital likeness. CELP would take just the unique characteristics of each voice, create a codebook, a guide to rebuilding that voice, then send just the characteristics and the code book down the line where the receiving phone could rebuild the voice based on the code.
And in the middle of all that, a complex key exchange system and an unbreakable user verification protocol that would render any attacks or deception completely pointless.
Brendan was so excited, as excited as he could get, the day he brought us the news that he’d heard CELP was about to be recognized as the new standard in speech coding by the US Department of Defense.
It was a clear vindication for both he and us although for us it was very double-edged. It just added even more pressure and the recognition that all our lagging competitors would have to speed up their transition to CELP and maybe even beat us to the line.
Brendan was also excited that based on separate discussions with some of his colleagues, a chip no bigger than a couple of stamps would probably be enough to bring the new CELP to life, and not the car-sized supercomputer referred to in the original proposal.
The version he proposed would eliminate most if not all of the weaknesses of the current generation of voice coders, and be easily able to accommodate different voices (it was difficult for women to use secure phones because the voice coders couldn’t handle the pitch), different accents, speed and melody of speech, sore throats, congested sinuses, background noise, and a terrible phone connection.
The processing requirements seemed massive when the process was explained to me, little as I understood any of it. Brendan mentioned casually one day that in order to achieve all that was being proposed, the CELP coder or circuit in each phone would be required to conduct anywhere between 2 and 12 million instructions or computations every second.
And that was just for the speech analysis and synthesis. It didn’t take into account all the additional computational power that would be required to perform the complex math at the heart of the encryption and exchange of keys, not to mention all the other basic functions expected in an executive telephone.
In yet another sign and just right on time, Motorola announced the availability of their 56001 floating point digital signal processor. It boasted almost the perfect speed and processing power, around 10 million instructions per second, that Brendan had estimated we would need.
It was also the perfect size to fit one or even more inside our compact phone, and at a price that would still keep our final production costs, even in small quantities, below a thousand pounds. We would still need at least a second identical processor dedicated to just the massive mathematical challenge of encryption and key management, but it was all still very feasible.
We also located a local distributor for Motorola semiconductors who was beyond excited that we would be the first customer so far to show any interest in their prized and brand new DSPs. They wouldn’t be cheap, he warned us, especially in the small quantities we wanted. The sales rep cautioned us that we might have to get approval for use first, because ofexport restrictions.
“They’re still mainly used for radar and weapons guidance systems,” he said. “That’s why they’re a bit sensitive.” If for any reason we were denied, he suggested, maybe Brendan could order them for himself as a researcher at the university. So at least we had a backup plan. We kept pushing forward.
Three weeks later the first handful of chips arrived and the excitement could be felt from front to back. But it was tempered with anxiety. If the DSPs didn’t work, or if we simply didn’t know how to get them to work, we would once again face doom. Doom was not an option even if it seemed our constant companion..
After another couple of weeks of testing just one of the chips, and verifying his calculations with some of his professors and colleagues at Trinity, Ciaran announced that in his modest estimate we would also be able to achieve the levels of security we first spoke of in the Buttery at Trinity.
Ciaran had opted to embed himself north of the border, amongst myself and the women. He said it was quieter, and the light was better, and it meant he had his own office even if at the expense of Gemma. After weeks of huddling with Brendan and Daragh and sometimes Shea, Ciaran said he was now very comfortable with his original suggestion, of creating a semi-proprietary encryption algorithm that would be loosely based on something called triple DES. And adding some adjustments to the design standard to overcome some lingering concerns that the NSA might have altered the original design to create a backdoor.
If 56 bit DES were broken, said Ciaran, or would be in the next few years, we could just go with a slightly longer key length to make it secure enough for the next few decades. Or we could go all in, way over the top, a 516 bit key length with a public key exchange like RSA or Diffie Hellman.
DES was the most commonly used and until recently, the most trusted encryption algorithm in the world, one heartedly supported by the NSA. Which of course made everyone suspicious about its real strength. Ciaran suggested that if we went as far as 512 bits, not only would we be making a very strong encryption algorithm, but a very strong statement. Maybe too strong. That the world was now ready for unrestricted encryption for all and that we, Ireland, would be the vanguard.
“How strong could we make it?” I asked, and Ciaran’s eyes lit up. “512 bits means we’re unbreakable forever,” he said. “But if we tried for 1024, we’d be unbreakable forever and ever.” His eyes were sparkling now “Now that would be a statement.” But after consulting with Daragh and Brendan the three concluded that in order to hit the highest target possible we’d probably need to add another DSP. Which would not only add to the cost but would need a completely different housing in order to fit all the new components.
So we decided to launch with a semi-proprietary symmetric algorithm with a key length of 512 bits, but state very clearly in all our documentation that a 1024 bit version would be available soon.
Every single bit would be used in every exchange, and every time a new call was made a new key would be generated. So not only would it take all the most powerful computers in the world working together billions of years to come close to cracking a single phone call, they’d have to start all over again for the next call. And the next. And the next.
Given that the visible universe as far as we could tell was at the time a little over 13 billion years, Ciaran was confident that to break a single key, using every single tool available on the planet, would take at least a few hundred trillion years. Give or take. Either way, we would be unbreakable.
While the south struggled with shaping the belly of our new beast, Gemma and I focused on an equally important task. What would we name it? For months it was always and only referred to as “the phone,” and while sticking with that name might be an audacious marketing strategy, it might be a little too soon for such a small startup to be so cocky.
One of top contenders for a new name for our newborn was the SVT 100, or Secure Voice Terminal. It sounded significant and befitting, reminding everyone that the phone that would no longer be referred to as just “the phone” was indeed not just a phone anyway but something far more capable.
The name also mimicked that of our greatest rival the STU 3, but at the same time alluding to something far more advanced. In the end we settled on MilCode, short for military coding, a name we felt would be significant enough for our military customers yet short and catchy enough for commercial markets. But we still designated the first version as the SVT 100. Just in case. And just because.
At the suggestion of the army, we decided that we would also try to bring MilCode up to the NATO Tempest standard. Tempest focused on another growing threat, a sideways attack, of monitoring the electrical signals, the pulse, the heartbeat, the breathing of the phone, through things as fundamental as sounds and vibrations generated simply by using the phone.
The threats were believed so real and potent, and especially from the Soviet Union, an entirely new type of security called EMSEC, or emission security, was born. And from that birth spawned Tempest.
It was believed that captured signals, even the crudest and most rudimentary like the strokes of keys or the flashing of LEDs, could be converted back into useful intelligence. In order to reach the Tempest standard and avail of the opportunity to win some massive NATO contracts, we would have to layer all parts of the phone with filters, shielding, and masking to reduce emanations and emissions to an acceptable minimum.
As a final coup de grace to any potential competitor, and just to add to the humiliation and intimidation, we also decided to add a crypto ignition key. Although it was a minor addition, it was also very significant because it was a little mocking.
In the NSA secure phone, as in most others, the ignition key was central to the operation of the phone because it contained the secret codes that would turn the call from plain to secure.
No key meant no call. A compromised key meant a compromised call, and simply managing the small plastic keys was a considerable security headache that often required both a dedicated high security safe and a lengthy protocol just to access the key, use it for a call, and then return it.
The crypto ignition key contained a small chip which in turn held a unique code paired to a specific phone and central to launching a secure conversation with that phone. Simply being in possession of that key, without permission, compromised the entire security of the phone it belonged to.
We didn’t choose just any key, though, but the very same make and model the NSA used and made in Savage Minnesota. And while they used the key as a way to initiate and secure a call, and therefore a significant cost and disruption, we decided to use the key simply for housekeeping, and a way to access the phone before launching secure mode. Just an irrelevant affect. And a mocking warning that anything they could do, we could do so much better.
Our third and final crown was turning out to be much more significant than we had assumed. If both the speech coding and encryption worked as expected, they would be valueless unless we had the right modem. A slow clumsy modem would almost completely erase the advances we were offering on speed and voice quality. Without the right modem, the perfect modem, our secure phones would never work.
Whatever modem we used would not only have to be fast, the fastest possible, but also small enough to fit inside the phone so we wouldn’t need to add any external devices. And of course affordable enough so that we could maintain our cost advantage.
Fate never abandoned us, not for a moment, and we soon discovered that a growing Irish electronics company called Cornell not only made a tiny and affordable modem chip set that would work perfectly inside our phone, it was also one of the first in the world to meet the new speed standard capable of transmitting up to 14400 bits per second. That was the highest speed and quality available, anywhere, but also capable of cleverly falling back to slower speeds if the line quality were poor.
The faster the speed over an ordinary analog phone line, the better the speech quality at the other end. Most of our competitors, few as they were, were still struggling with a speed of just 4400 bits, a third of the optimum, and hoping to soon make it to 9600, which had become our starting point. And another reminder to all comers that we were better and faster and a force that they would have no choice but to reckon with.
As we all became more confident that what we had dreamed on paper had a better than good chance of coming to life, and very soon, my dream started to wander off its star, just for a moment. Ciaran suggested confidently that as long as everyone else held up their part of the bargain, we would be the proud owners, and purveyors, of the most secure communications system in the world. Barring an error in coding or a backdoor secretly injected by one of us, our phone would be completely unbreakable, to anyone, from anything, and forever.
That would make us so many friends, or at least fans, like the army had suggested in their original sales pitch on that crisp morning. And make us so much money too, millins in fact. But it might also make us a handful of enemies whose rage and reach would outweigh all fans and fortunes.
I calmed myself by convincing myself that we would all be good friends in the end. The Americans, and anyone else who might feel threatened by our work, would probably admire us so much for beating them so easily that their better option would be to partner with us, to learn from us, or better still, acquire us.
Surely they’d see the sense in that. If we could achieve so much, so quickly, with so little, and at our first attempt, how much could we change this entire new world of security and encryption, of secrecy? And not to forget privacy. This was my noble fallback.
We weren’t protecting bad secrets for bad people from better people, not anymore. I was convincing myself that this phone of ours, this glorious encrypting machine, would give every human the opportunity to keep their most private secrets secret. A most basic human right. As long as they could afford the anticipated $3,000 per phone, and they’d need at least two. But it felt like a fair exchange. Human rights might be free, but their deployment and enjoyment might not be.
I also began to believe my own deceit. Or perhaps conceit. Perhaps I wasn’t such a fake, such an imposter after all. Perhaps I truly was a visionary and others could finally see that, sense that, in me. I didn’t have to understand what I was doing, I just needed to keep doing.
I just had to imagine, and then find others who could add the shape and texture to the vision and bring them to my side. Perhaps more than having this magic, I was the magic. Everything I’d touched, even glanced at so far had somehow risen to my summoning. Perhaps that’s how it would always be.