Danyl McLauchlan visits the Reserve Bank to inspect Bill Phillip’s MONIAC.
It looks like an artifact from an alternate timeline. MONIAC is about two metres tall and one metre wide; like an oversized perspex coffin. It consists of tanks and tubes and pumps and columns and electrodes and pipes, all interconnected, with labels like ‘Domestic Expenditure’ and ‘Income After Taxes’. When it is activated – by switching on the pumps and turning a tap on the side – water cascades through it, bubbling along pipes and weirs and tubes until it reaches the large tank at the bottom labeled ‘National Income’ where it is pumped all the way back to the top. As it flows around the system the torrent of water solves nine differential equations simultaneously. MONIAC is a computer.
It was built in 1949, the dawn of the information age. Standing in the Reserve Bank Museum in downtown Wellington watching it operate you can imagine, for a moment, a world where technology took a different turn. Instead of digital computation – instead of PCs, Unix, Apple, tablets, social media, mobiles – we’d have a world driven by fluidic logic machines descended from MONIAC: eerily silent computer centres built beneath dams; the world’s knowledge stored in vast capillary beds; burst water-mains spraying bits of data everywhere.
MONIAC was designed and built by Bill Phillips, a New Zealand economist who grew up on a dairy farm near Dannevirke, spent most of his career as a professor at the London School of Economics and who flatters New Zealand’s conception of ourselves as humble but ingenious folk solving the world’s problems with can-do and bits of fencing wire to such an outrageous degree it’s a miracle there aren’t giant statues of him looming over our cities.
Phillips was born in 1914. His father was a farmer and a self-taught engineer who electrified his farm with a water-wheel decades before rural areas were connected to the grid, and who taught his son electronics by opening up their crystal radio set and showing him how it worked. This was long before the days of solid state transistors: circuits were soldered wires and glowing valves; electrons flowed through the system in simple, comprehensible paths. Like water.
Phillips’ parents planned to send their conspicuously brilliant son to university but then the Great Depression hit, devastating the agricultural sector and wiping out their savings. Instead Phillips became an apprentice engineer at the Tuai hydroelectric station in the Hawke’s Bay. From there he moved to Australia where he became a crocodile hunter, left for Russia in 1937 on a Japanese boat that was diverted to Tokyo when Japan declared war on China mid-voyage, was briefly arrested as a suspected spy, then traveled across Russia at the height of Stalin’s terror, where he tried to get a job as a miner but couldn’t because the Siberian labour market was flooded with “a plentiful supply of political prisoners”. He wound up in London. When the war came he volunteered for the Royal Air Force.
The Air Force transferred Phillips to Singapore. When the Japanese army overran the city he evacuated on a cargo ship which he defended from bombers with an improvised mounted machine gun; he fled Japanese paratroopers in Sumatra; attempted to turn an abandoned bus into a raft and sail it to Australia, but was caught and spent three years in a Prisoner of War camp. There he constructed secret radios out of pilfered telegraph components, and water heaters that tapped into the mains power, before being transferred to a death camp where the prisoners were ordered to dig their own mass graves at gunpoint. The war ended before they finished digging. When Phillips was released he weighed forty-five kilograms; he spent the rest of his life chain-smoking cigarettes with trembling hands and refusing to talk about the camps. He returned to London, studied sociology then economics at the London School of Economics, and in his final graduate year he built the MONIAC.
When the Great Depression struck, John Maynard Keynes, the founder of modern macroeconomics, declared that the economy had “magneto trouble”. Liberal economists in the late 19th and early 20th century liked to think of market economies as rational, self-correcting systems that operated best without government intervention. Keynes talked about the paradox of thrift. Say you’re a business owner. There’s a downturn in the economy: profits are low, you might even be losing money. So you buy less stock, maybe lay off some staff or close some unprofitable branches to reduce overheads. All totally rational, and yet if every sensible, rational business does the same thing, consumption will slump, unemployment will soar, and the economy will crash. Individual rational decisions can aggregate into irrational catastrophes.
The panic of ‘29 and great depression of the 30s seemed to lead directly to the rise of totalitarianism, global war, people being machine-gunned into mass graves so there was intense interest in preventing or mitigating future crashes. The economy was like an engine, Keynes insisted; it got into trouble when part of it was malfunctioning. There was no self-correcting system: you needed a good mechanic to intervene and fix it.
The problem, Bill Phillips realised, was one of computational complexity. If you wanted to fix the economy you needed to understand the interactions between the constituent parts. If, say, interest rates went up, people would borrow less, save more and spend less. If you wanted to forecast the impact of a rise in interest rates on spending you could look up the historical data. Institutions like the LSE had ‘computers’, by which they meant rooms filled with women who performed complex calculations – women seemed to have a curious knack for calculus and discrete mathematics, it was felt at the time – and you could give them the variables and they could derive a function from them and you could use that to try and forecast what would happen to consumer spending if interest rates rose.
But how about all the components of the economy downstream from spending? And how would a change in the taxation rate impact those changes? That was another differential equation. What happened if export revenue dropped? Another equation. Imports? Yet another. And it was impossible to solve all these problems simultaneously. Not even the rare programmable electronic computers dotted around the world – ENIAC in the US, the EDSAC at Cambridge – could do it.
Some experts suspect that Phillips had access to a fully programmable electronic computer; that he needed one to solve some of the complex nonlinear calculations built into MONIAC’s architecture. They do know that he had access to a decommissioned Lancaster Bomber, because the first MONIAC was constructed from parts scavenged from one: fuel pumps, injectors, bulletproof perspex for the tanks. He filled it with ionised water dyed a shocking pink and demonstrated to the Robbins Seminar, one of the LSE’s most prestigious speaking events.
The crowd, dense with future Nobel laureates, regarded the machine with extreme scepticism, which deepened when Phillips – chain smoking; hands trembling – activated it and the pumps screamed and sprayed pink liquid around the system, like a wounded animal with a severed artery. The scepticism turned to astonishment as the machine reached an equilibrium, accurately displaying the relationships between economic variables, and they grasped what Phillips had accomplished: the first ever computational model of an economy.
Phillips was offered a teaching position. In less than ten years he was a full professor; he became one of the most distinguished macroeconomists of his generation. His paper on the link between inflation and unemployment – depicted by the Phillips curve describing an historical inverse relationship: when wage inflation was high, unemployment was low, and vice-versa – became the most frequently cited paper in the history of macroeconomics. Replicas of MONIAC were built and shipped to places like Harvard; the Ford Motor Company; the Central Bank of Guatemala. The problem of how market economies worked, and which parts to fix when they broke seemed largely to have been solved.
The water cycling around the MONIAC at the Reserve Bank is not pink. It’s not ionised either: instead they use distilled water to prevent corrosion of the parts. So some of the ingenious feedback mechanisms, like an electrode that closes a circuit if immersed, linking consumption with income, no longer function correctly. Even if they did, the device is designed to work for the British economy in the late 1940s. One of the bank’s economic forecasters demonstrated the machine to me, answering all of my questions no matter how obtuse with clarity and ease – until I stopped looking at the parts and considered the machine as a whole and asked, “And the water flowing through it symbolises money?”
“And what is money?”
“Ah.” She hesitated. “That’s a good question.”
The Reserve Bank is a central bank. It creates money. Sort of. They create our notes and coins (technically they don’t even do that anymore: currency creation is contracted out to an overseas supplier because that’s more, y’know, economic). But almost all of the money in our economy is electronic and most of it is created in the private banking system (simplified version: if you deposit $100 in a savings account your bank lends out some multiple of that; currently around 130%, so your bank creates $30 out of nowhere). About five billion dollars exists as actual notes and coins held by the public, compared to $290 billion in deposits existing as entries in financial databases. The Reserve Bank maintains the value of that money by setting the base interest rate it gets loaned out at.
Our lives are so circumscribed by money: some folks get upset when they find out it has no intrinsic worth and is created out of nothing. The Reserve Bank hears from them fairly frequently: people distraught to find there isn’t a vault somewhere – beneath The Terrace perhaps – filled with gold guaranteeing the value of their savings (the bank was very emphatic about the non-existence of a huge gold reserve beneath their offices.)
Paradoxically, it is this very lack of value and material existence that makes fiat money so valuable: it means it can be manufactured and exchanged at near-zero cost. It functions as a symbol for value, a way to communicate prices. The water spraying around MONIAC is metadata: information about information.
One of Phillips’ contemporaries at the London School of Economics was Friedrich von Hayek. Hayek was interested in spontaneous order, the natural emergence of complexity from simple systems: languages, societies, markets. Life itself emerged from inert compounds: simple chemicals self-assembling into multicellular organisms. No one designed any of these things, Hayek pointed out, yet they operated with astonishing intricacy and efficiency.
No one could predict the behaviour of a complex, chaotic system, he warned, especially not an economy made up of millions or billions of individuals, not even with the most powerful computer: because the information individuals used when making decisions was distributed, decentralised; impossible for any forecaster or planner to access. When he won the Nobel Prize in 1974, Hayek’s laureate speech cautioned his fellow social scientists against “the pretence of knowledge”. They might think they understood complex systems, he warned, and they might tell themselves they knew how to model them, improve them, fix them – but they didn’t.
By the 1970s the Keynesian model embodied in MONIAC, which the advanced economies of the West based their monetary and fiscal policies on was breaking down. They had low growth along with high unemployment and high inflation, in defiance of the inverse relationship between the factors predicted by the Phillips curve. Attempts to intervene, to fix the faltering economies only seemed to make things worse.
In 1976 the American economist Robert Lucas diagnosed the problem. When Bill Phillips modeled the historical variables he built into the MONIAC, or the link between unemployment and inflation in the Phillips curve, he wasn’t observing the operation of an engine, or a computer, but the aggregated results of the complex individual behaviour Hayek talked about.
People weren’t just tanks for money to flow through; they could change decisions based on their own forecasts, which could anticipate the government’s forecasts. If businesses knew the government was raising inflation to lower unemployment they could, completely rationally, reduce staff to offset the anticipated cost of inflation, leading the economy into another downward spiral, one that was harder for the government to fix because it was caused by the anticipation of the government trying to fix it.
By this time the MONIACs around the world had long been retired. The US military experimented with fluidic logic machines during the Cold War – unlike conventional computers, devices like MONIAC are immune to the electromagnetic pulse of nuclear weapons. But the future of computation was electronic, and after Lucas macroeconomic forecasting became more game-theoric: rational-agents with perfect information competing in multiplayer systems.
By this time Hayek’s vision of market economies as forms of emergent complexity had been embraced by political and business leaders across the West. It complimented their belief in capitalism as the system of reason, individuality and freedom. State-owned assets were privatised. Taxes were reduced. Ronald Reagan announced “Government isn’t the solution to the problem. Government is the problem.” In 1989 New Zealand’s Reserve Bank was granted political independence in inflation targeting; a governance model which was quickly copied in most developed economies. Financial markets were deregulated. Adversaries of the changes, recognising the similarity to the pre-Keynesian, liberal approach to markets, dubbed them ‘neoliberalism’. In 2003 Robert Lucas announced that the central problem of macroeconomics, that of depression prevention, had been solved. Five years later an international banking crisis crashed the global economy.
Back in 1967 Bill Phillips burned most of his macroeconomics papers. He moved to Canberra; suffered a massive stroke; returned to New Zealand; taught an undergraduate course on the Chinese economy at the University of Auckland; died of a second stroke in 1975; ashes returned to the lonely rural graveyard where his parents and brothers lay buried. He was only 60; friends and family felt he’d never really recovered from the war. He died before the Lucas critique; the neoliberal revolution; the financialisation of the world economy; the transformation of global markets by information technology.
The indirect descendents of Phillips’ MONIAC and the equations he encoded into it are today’s high performance computers and machine learning algorithms. They execute the majority of financial trades; human transactions account for only a tiny fraction of global commerce: AIs running on massively parallel architectures perform trillions of operations per second. The global derivatives market – hedges, speculations, forecasts – is packaged into complex financial products, their total value estimated at over a quadrillion dollars: ten times global GDP; constellations of data, self-assembling, proliferating: recondite complexity, impossible to forecast.
When the demonstration at the Reserve Bank Museum was over the demonstrator switched off the pumps. The water – or money, or information, or whatever it was – stopped churning. She said, “I like to leave it in a good condition for the next demonstrator” so we stood and waited for the system to empty. I was surprised by how quickly it all drained away.