Give me a long enough lever and a fulcrum on which to place it, and I shall move the world.
- Archimedes
The Apollo program’s Lunar Module was the world’s first digital vehicle. When Neil Armstrong “took control” to land on the Moon, his control inputs went into the incredibly sophisticated Apollo Guidance Computer, which interpreted those inputs, and then controlled parameters such as yaw, pitch, and roll, which allowed Neil (and those who followed him) to safely descend to the Lunar surface.
The AGC was a marvel of engineering, a miniature digital computer that was a key enabler for the entire rest of the program. Despite the Saturn V rocket, the infrastructure, the Apollo spacecraft — without the AGC, the landings would never have happened.
One — of the several — reasons why the Soviet Union would never land humans on the moon was because of the lack of sophistication of their system controls (and the subsequent total lack of software¹). The Soviet equivalent of the AGC was the Globus ИНК (INK), which was a terrifically complex² analogue (mechanical) computer with gears, differentials, and cams, all intermeshing to provide a navigational output which a cosmonaut could use to figure out where they were, and where they would be likely to land. The Globus was fundamentally inflexible, in a way that the AGC was not. The Globus could only predict a spacecraft’s position based on predetermined inputs (unlike the AGC, the Globus couldn’t take inputs from sensors to determine a true position); and the Globus could only perform the function which it was pre-set to perform³, and this meant that many of the feats of the Apollo program which permitted its success would have been impossible to accomplish⁴.
The Apollo Guidance Computer was introduced in 1966, at the core of the Lunar Module design — a design paradigm that was digital-first, and unashamedly so, in the sense that the design would not have been possible without the AGC. And while more or less every road-going vehicle today features embedded computers, all too often these have been introduced as a means of replacing a mechanical component or system with something of equivalent functionality for reasons of either cost reduction or sometimes improvements in efficiency. As a result, the true advantages of a digital-first design cannot be taken advantage of. These vehicles are stuck using the equivalent of a Globus, when what is needed is an Apollo Guidance Computer.
For instance, mechanical carburettors were replaced with mechanical fuel injection, and then with electronic fuel injection, to improve reliability, reduce cost, and enhance efficiency — all of which was realised — but was still fundamentally built on a tech heap that stretched back to the dawn of internal combustion, and replicating an existing function, but better.
Silicon, and computing power, can be “layered on” to improve what is already there, and modern cars and trucks have got computing power that would make the Apollo program blush. But as we grapple with the requirement to shift to more efficient, electrically-driven road vehicles, the optimisation of this energy shift requires a clean sheet design with completely integrated, flexible, and connected core computing and software which allows for leaps beyond what was possible with a mechanical system — a totally different paradigm requiring different design principles.
A key principle here is that silicon is cheap — and with the steady advancement of Moore’s Law⁵, it is now almost laughable not to include embedded silicon wherever possible⁶. The cost per transistor — even with recent supply chain issues — is almost so cheap as to be inconsequential.
But silicon in itself is only really the fulcrum point of the lever that is the software control.
And if all you’re doing is replacing mechanical components with electrical and electronic ones, then you’ll never be able to provide the meaningful movement of a long lever that is going to be necessary to grapple with the transition to a low-carbon economy.
And this speaks to the underlying paradigm shift required to accomplish this. There are a lot of industries still grappling with the transition to digital, and it requires a fundamental shift to a software-first approach and a software-first competency. If the mindset is still where electronics and software is secondary to the mechanical paradigm, where digital is regarded as a “bolt-on” or replacement, then this will never take advantage of the opportunities afforded by electric.
Ultimately, electrification isn’t merely about swapping a combustion engine out with an electric motor — this is the relatively easy part that both existing players and startups are doing. This approach will never be able to enable the benefits of electrification, except at a very superficial level. Switching to electric requires a wholesale change in everything, from drivetrain architectures to data networks to reshaping — in our case — what a truck really is. This is a fundamental change not only in technology, but also in core competencies, culture, and design principles.
To illustrate with another example from the aerospace world. The F-16 was the world’s first fly by wire, digital proportional control aircraft. It differed not just in how the inputs to the control stick caused the flight control surfaces to move — but in the fundamental design of the aircraft as being one that was inherently unstable. Like the Apollo LM, the computer flies the plane — the pilot inputs what he wants to happen, and the computers figure out how to make that happen. Today, it would be unthinkable to design any modern aircraft any other way. This is a design paradigm that, once the change took hold, ended up propagating and driving all other paradigms to extinction.
In the land of commercial vehicles, the old way of doing things is still the only way to do things. But when you try to make a dinosaur digital — at the end of the day, it is still a dinosaur.
¹ A software engineer for the AGC, Don Eyles, gives a fascinating insight into programming the AGC in his book Sunburst and Luminary, named for two of the programs that the flexible, powerful, and efficient AGC ran. ^
² The Globus is taken apart and investigated here. Look at all those gearwheels! ^
³ For instance, the Globus could only operate for a fixed orbital inclination and for circular orbits, rendering it useless for rendezvous and docking. It was only replaced in 2002! ^
⁴ Apollo 14 had a critical fault on the way to the moon. Demonstrating the power and flexibility of the AGC, the flight software was reprogrammed by the astronauts when they were already orbiting the Moon! ^
⁵ The continuing dominance of Moore’s Law has dramatically reduced the cost of transistors, increased the transistor count, and, through shrinking die size, massively reduced the energy cost per calculation. This is the main reason we have highly sophisticated and connected pocket computers now, less so because of improvements in battery technology. ^
⁶ To return to the comparison of the AGC and the Globus, the AGC deployed 17,000 transistors. The Globus had one. ^