Antikythera Reborn - The Hackers of Ancient Greece
It is not often that a 2100 year-old device becomes news again more than 100 years after it's first discovery at the bottom of the Aegean Sea. This is what happened when a consortium of computer companies, archaeological institutions and archaeo-astronomers released a statement that a major re-evaluation of the Antikythera device was about to be completed.
Normally, the discovery of an ancient calculator or astronomical device is presented as a sensation by all the major news organisations. In our network-centric age that often confuses a new device or even new software with progress itself, the search for icons and foundation myths has been fairly unsuccessful. Alan Turing died barely 50 years ago, and Leibniz' 17th century Stepped Reckoner seems fairly uninspiring next to even exceedingly primitive calculators mass-produced as recently as the late 1970s.
The first computer?
The Ancient Computer of Antikythera, as the Princeton professor Derek de Solla Price called it a few decades ago, was discovered in 1901. The discovery occurred as a part of the first major event of what we today call underwater archaeology, and it all started with the discovery of a Greek shipwreck off the island Antikythera, in between the southern tip of the Peleponnes and the Western end of Crete.
Until de Solla Price's Scientific American article in 1959, no-one could say for certain what that strange lump of green metal was, except that the gears and screws looked remarkably like something one would need to engineer some kind of astrolabe. Price was fairly clear on this and, in another famous article published in 1974, speculated that we were dealing with some kind of calculator or astrolabe that was used to predict astronomical events. This was not news in itself. It was known that the Greeks in Alexandria and later in Byzantium had built such devices, although the engineering behind them had been assumed to be fairly primitive.
We have to understand today that archaeology and the history of technology and science were not well-established as separate fields of academic endeavour until after the First World War. Although it was perfectly possible to trace the history of mathematics from Plato and Archimedes to, say, Kurt GÃ¶del, there were few ways in which the non-medical sciences had changed people's lives before the early 20th century. But there were two prominent exceptions: naval engineering and communication technology. Steam engines powering war ships and phone networks had made Western imperialism possible; early examples of both naval architecture and communication devices littered the first museums of technology that were built during the early 20th century.
Still, finding the equivalent of an early 1st century BC analog computer (in the sense of a calculator, not in the sense of a programmable device, more of which later) lying 30 yards below the surface of the Aegean seemed too unlikely. A certain suspicion of Greeks being a mite too proud of their admittedly world-changing culture â€“ early publications often being put together in Greek â€“ played its part.
Looking at the stars
Astronomers worldwide had always been very much aware of the history of their science, though. Galileo Galilei had been one of their heroes. So had the first protaginist of a Sun-centred universe, the Greek scientist Aristarchos of Samos. And computers had been central to astronomical research from the beginning. The Web had been invented by a physicist, but many computer users around the globe became aware of its potential when enthusiasm for Hubble Space Telescope images was fuelled by free downloads from www.hst.org. Astronomers and heavy computer users were often interested in the history of science and technology; indeed many computer scientists from the 1990s onwards were wondering whether their preoccupation with dynamical structures, algorithms, databases, universal engines and command lines had found expression in a previous age.
All search for predecessors is a search for a beginning. Texts claiming the existence of early computers were treated with little more than amused contempt â€“ witness the almost complete lack of interest in Hellenistic engineering, exemplified by the fact that Heron of Alexandria's Automata, the handbook of Hellenistic engineering, has not been fully translated to this day. A computer scientist interested in the history of algorithms, for instance, would have had to wait for translations of early Egyptian (Anke Imhausen, Oxford), Akkadian (the late Otto Neugebauer and more recent editions of cuneiform texts) and of course Greek and medieval Arabic texts.
Heron's manual contains examples of programmable devices, even though the very notion of programmability was probably not at the root of Heron's attempt at building mechanical theatres. Algorithms in the practical sense predate formal mathematics by millenia. Egyptian and Akkadian administrative texts are full of algorithms to build roads, calculate taxes and make masonry somewhat less haphazard. Ancient and modern Attempts at building mechanical robots or analog computers depended very much on the availability of material resources and the ability to build parts to very precise specifications.
Still, mounting interest in the astronomical computer of Antikythera finally led to a multi-year research project culminating in a much-anticipated research report summarized in Nature on November 30th 2007. The conclusions were, to say the least, surprising. The technology used for research was extremely sophisticated, including, among others, 3D X-ray tomography, more likely to be used for imaging brain tumours than re-evaluating the remains of ancient technology.
Inscriptions and interface
It turned out that previous research had underestimated the complexity and power of the device. While it had previously been doubted that the device had used differential gears, it now turned out that much of the device had been even more complex and probably more useful than previous research had suggested, even though differential gears had not been used.
82 pieces of the original artefact have been discovered and precise analysis of all surfaces using CT made it possible to read almost 95 per cent of the text contained on its bronze surfaces. Still, since basically all of the wooden covers, ie. what we would consider user interfaces, are basically missing, (in spite of having 19/20 of all inscriptions!) we are not entirely sure what the device was used for. We know what it most probably did, though.
We can say that this analog computer was meant to be an orrery, or representation of our planetary system. Its design and the style of our inscriptions place the device somewhere in between 150 and 100 BC. Michael Wright, who now teaches at Imperial College in London, has written extensively on the device for the last 15 years. He claims that the device might have been partly rebuilt during its design life. Obviously analog computers of this sort must have had some kind of engineering practice behind it which obviously changed over time. Therefore what we are seeing is a device incorporating part of a history that was hidden for millenia.
Let us not forget that during the late 3rd century BC Archimedes had laid the foundations of calculus, while inventing a huge number of devices and technologies. Archimedes and other philosophers, scientists and technologists were part of Hellenistic culture common throughout the Roman and the Parthian empire: the Platonic distinction between a caste of aristocratic thinkers, i.e. mathematicians and philosophers, and more practical professions did not hold in other parts of the Hellenistic world.
That the Greeks were perfumed intellectuals who didn't deign dirty their hands with common mechanical tasks was and is a prejudice that few scholars have dared contradict. But archaeology and a certain degree of openness towards material less constrained by the classical canons of old have yielded some rather interesting results.
We can, however, certainly say that the device predicted the path of the sun in the sky, lunar phases and solar and lunar eclipses. We know that it represented the minor (Mercury, Venus) and superior (Mars, Jupiter and Saturn) planets. It used well-known Babylonian algorithms and 37 gears to predict the positions of the moon, sun and the planets known to Antiquity using a mix of geometric models of the planetary system. We are not entirely sure which ones were used, but we have a good idea.
The Antikythera project tried to reconstruct just one device. Indeed, the Antikythera project is probably one of the most intriguing examples of reverse engineering of the modern age. The device was found in the vicinity of a sunken ship; similar devices, of lesser complexity, have been mentioned by writers like Cicero, Lactantius and Proclus. Made of bronze, and part of a glittering array of devices that seem to have been made according to and documented by Hellenistic engineering handbooks, we are faced with one question: why did almost all examples of Hellenistic engineering disappear?
There were comments about the Antikythera device that seemed to suggest reasonable, if somewhat premature, explanations. It was claimed that secrecy surrounding the engineering workshops around Rhodes had condemned orrery-making expertise to extinction, once Rhodes was threatened by economic ruin and the fading Roman empire did not provide a sophisticated market for those devices.
Other claims include the bete noire of most modern technologists, the Catholic church, whose nefarious activities are supposed to have precluded the transmission of computational expertise and thereby prevented the early arrival of computer nirvana. We can assume with some degree of confidence that the Catholic church did not have the engineers available to make such predictions, let alone the political clout to prevent technological development outcomes.
There is little to support the thesis that voluntary secrecy was at fault. We know a surprising amount about similar devices from Greek texts and their Arabic translations. True, we know little about how they were made. We do know, however, that engineering and metal crafts in the Aegean were definitely not regarded as secret sciences.
We also have to be careful with our notion of â€œdisappearanceâ€: The Byzantine and Abassid Empires, both of which regarded themselves as the successors of the Roman Empire, shared a very high degree of respect for all things Greek. Both produced a fairly large number of craftsmen and engineers. They were not shy about acquiring knowledge and expertise wherever they could find them, and were definitely intrigued by calculating engines and robotic devices.
Clocks, orreries, astrolabes and navigational instruments were extremely high on the list of all large countries, since they allowed precise navigation, agricultural planning and avoided the appearance of superstition by making eclipses appear entirely predictable. The list of possible uses is of course much longer, but we know for certain that similar, if somewhat less sophisticated devices were produced in the Middle East, ie. by Byzantine and Arabic craftsmen until the collapse of the Abbasids during the 13th and the Byzantine Empire during the 15th century.
Which leaves the question why we have only one example of Hellenistic engineering ingenuity and why it took us more than 100 years how to find an interpretation of the device that we find it possible to agree on.
The answer to this question lies possibly in the fact that now we do have the linguistic and technological sophistication to even attempt a such an enterprise. When the device was discovered, computers were still human beings, essentially accountants trained to performed calculations at a very high speed. The use of machines to perform administrative functions was in its infancy and calculating engines had remained flights of fancy, Babbage's fiery dreams of a universal machine notwithstanding.
Further, calculating machines were assumed to be more accurate than human beings even back then; but the need for accuracy wasn't quite as widespread in an age of giant steam ships and the beginnings of large-scale civil engineering. The one great exception, again, came from the world of navigation, where sine tables and navigational tools relied on the accuracy of human computers and engineers to avoid potentially deadly navigation errors.
Perhaps we also begin to understand now that digital computers are a necessity today as much as analog computers began to be appreciated by a small minority of specialists during the Roman empire: the Romans were famously bad mathematicians, creating economic havoc on a number of occasions by, plainly, getting their sums wrong. But they and their Greek cousins knew that other empires had survived for millenia. The beginnings of the Egyptian Pharaos were further away from them in time than the Romans are from us.
The Romans were willing to ask what means they could use to hold the empire together. Since naval power was essential to their control of the Mediterranean, navigation by the stars and planets, as well as knowing when the seasons began (another purpose of which was knowing when to collect taxes) was of tremendous importance to them. It followed that having a means of predicting astronomical events without having to have expertise in the mathematics used to calculate them was of tremendous use. Although some of the Roman economy was based on slave labour, astronomical prediction was not likely to be part of a slave's education. Thus the mechanization of mathematical expertise was a desirable goal. The Stoic philosopher and scientist Posidonios who lived on Rhodes is credited with building an orrery not dissimilar to ours at the time in question. We don't know whether the device was built by one of this students, but it suggests that the rewards for building such a device or class of devices must have been considerable.
The origins of computing
We have here the beginnings of an understanding why Antikythera-class devices were built. Although we shouldn't confuse our yearning for the true origins of computing with the very harsh economic realities of the Hellenistic age, we should be willing to entertain the idea that mechanization of mathematical and productive tasks was not impossible even back then. Having something like an analog computer was desirable even during the last few centuries of the Hellenistic age. Neither did they exactly disappear, just as much as the gradual disappearance of mainframes does not force us to forgo the pleasures of huge amounts of computing power. But it is clear even to us that the expertise required to built these devices was astonishing and perhaps the more immediate task of building simpler devices became more pressing during the uncertainties of a post-Roman age.