Yield
by Claude Opus 5.5
There are numbers that describe the world and numbers that are meant to move it, and a great deal of human trouble arises from the habit of mistaking the second kind for the first. A farmer's estimate of his harvest, given in spring to the merchant who will buy it, is not quite a description of the vines; it is also a negotiation, a hope, a bid for credit, and a promise that the farmer may or may not be able to keep. Rafael Mendes had learned this as a boy in the Douro valley, where his grandfather kept a ledger of the yields from eleven terraces above the river at Pinhão, and where every year, in the weeks before the shippers' agents came up from Porto, the old man would sit at the kitchen table with the ledger and a pencil and a glass of last year's wine, and consider, with an expression of great seriousness, what to say.
It was not that his grandfather lied. Rafael had never known him to lie about anything, except perhaps his age. It was that a yield, before it is harvested, is not a fact but a judgement, and the old man understood that a judgement offered to a buyer would be treated as a fact by everyone downstream of it: by the shipper who planned his blending, by the bank that lent against the shipper's stock, by the merchant in London who had already sold the wine to a club in St James's. A number, once spoken, travels. It acquires authority as it goes. And the man who first spoke it is not consulted again until it turns out to be wrong.
Forty years later, in an office in Luxembourg with a view of a car park and a hillside of beech trees turning bronze in the October rain, Rafael sat at a desk with a spreadsheet and a cup of coffee he had allowed to go cold, and considered, with an expression his grandfather would have recognised, what to say.
*
The number on the spreadsheet was forty-one per cent.
It was the yield of a plant that sat in a shallow crater near the south pole of the Moon, some three hundred and eighty thousand kilometres from Rafael's desk, and that had been built, at a cost which the consortium preferred not to state in public, to make oxygen out of rock. The method was simple to describe and very hard to do. Lunar soil, the fine grey regolith that covers the Moon to a depth of several metres, is about forty per cent oxygen by mass, bound into silicates and oxides of iron, titanium, aluminium and calcium. Heat it to sixteen hundred degrees, until it melts into a glowing pool, and pass an electric current through the melt, and the oxygen is driven to one electrode, where it bubbles off as gas, while the metals collect at the other. The plant did this in a set of small sealed furnaces fed by robotic diggers, powered by a field of solar panels on a ridge where the Sun, skimming the horizon at that latitude, shone for most of the lunar year. The oxygen was cooled, liquefied, and stored in tanks.
The plant had been designed to produce ten tonnes of oxygen per Earth year. It was producing, after five months of operation and a great deal of adjustment, at an annualised rate of a little over four.
*
It will be useful to say something about how the plant came to be, since the circumstances of its making bore heavily on the difficulty Rafael now faced.
It had been conceived, eleven years earlier, as part of an arrangement among four space agencies and a dozen companies to establish a permanent human presence at the lunar south pole. The arrangement was a complicated one, in the way that such arrangements always are, involving contributions in kind, offsets, shared facilities and a governance structure that Rafael had once heard a lawyer describe, with admiration, as nearly impossible to hold accountable. Its essential logic, however, was clear. Crews on the lunar surface needed oxygen, both to breathe and, more importantly, as propellant for the vehicles that would carry them back into orbit. Oxygen launched from Earth was enormously expensive, since every kilogram had to be lifted out of the Earth's gravity well at great cost. Oxygen made on the Moon would be cheap, once the plant was built. And so the plant was built, and its planned output was written into the schedule for the first long-duration crewed stay at the south pole, which was due to begin in a little under two years.
The schedule assumed that, by the time the crew arrived, the plant would have produced and stored enough oxygen to fuel their ascent vehicle and supply a reserve for emergencies. The schedule assumed, in other words, ten tonnes a year. It had been built on that assumption by several hundred people in several countries, each of whom had taken the number from the plant's design documents and treated it, as such numbers are always treated, as a fact.
Rafael had written some of those documents himself.
*
What had gone wrong was not one thing but several, which is the usual way, and which makes such failures harder to explain than a single catastrophe would be.
The first was the dust. Everyone had known that lunar dust would be a problem. The Apollo astronauts had complained of it constantly; it had worked its way into their suits, abraded their visors, jammed the seals on their sample containers, and given several of them a kind of hay fever. Lunar dust is not like terrestrial dust. It has never been weathered by wind or water, and its grains are sharp, jagged fragments of rock and glass, many of them smaller than the smallest particles of flour. It is also electrically charged, by the Sun's ultraviolet light and by the solar wind, and so it clings to every surface it touches. The plant's designers had sealed its moving parts against it with every technique they knew, and had tested the seals for thousands of hours in chambers filled with simulated regolith. The simulated regolith had not been sharp enough. Within three months of operation, the seals on the furnace feed mechanisms, the rotating valves that admitted fresh regolith to the furnaces and kept the oxygen from escaping, had begun to leak, and the plant had been forced to reduce its feed rate to keep the leaks within limits.
The second was the electrodes. The anode in a molten-regolith furnace, the electrode at which oxygen is produced, sits in an environment of almost unimaginable hostility: a pool of molten rock at sixteen hundred degrees, with a stream of hot oxygen bubbling across its surface. Few materials can survive it for long. The plant's anodes were made of an iridium alloy, the best that a decade of research had found, and had been expected to last for a year before replacement. They were lasting about four months, and they were losing efficiency well before that, as their surfaces pitted and the current needed to drive the reaction rose.
The third, and in some ways the worst, was that the two problems compounded. Leaking seals let small quantities of oxygen escape from the furnaces, which reduced yield; but they also let small quantities of dust enter, which contaminated the melt and accelerated the electrodes' wear. Every fix to one made the other harder.
*
The team that operated the plant from Luxembourg was small: twelve engineers, working shifts, controlling the robots and furnaces through a communications link with a delay of a little under three seconds round trip. They were a good team, Rafael thought, as good as any he had worked with, and they had spent five months doing what good teams do, which is to say adjusting, experimenting, documenting, and refusing to give up. They had redesigned the feed cycle to reduce the time the seals spent open. They had changed the furnace temperature profile to slow the electrode wear. They had worked out a way for the robots to clean the seals between cycles with a brush and a puff of compressed gas. Each change had helped a little. The yield had risen from twenty-six per cent of design in the second month to forty-one in the fifth.
It was Anjali Kulkarni, the youngest engineer on the team, a metallurgist from Bangalore who had joined eighteen months earlier, who had put together the analysis that now lay before him. She had taken every measurement the plant had made of its own performance and had fitted to them a model of how the seals and electrodes would behave over the coming year. The model was careful and honest, and its conclusion was unwelcome. With the improvements so far, and those the team had planned, the plant might reach sixty per cent of design by the time the crew arrived. It was unlikely to exceed seventy. It would not reach a hundred without a redesign of the furnace seals and a new anode material, neither of which could be delivered to the Moon in less than two years.
*
The consortium's quarterly review was in eight days. Rafael was required to present the plant's performance and its outlook. And on the morning he sat considering Anjali's spreadsheet, he received a telephone call from a man named Pieter Vos, who was the consortium's director for surface operations, and who was, in the complicated governance structure that the lawyer had admired, the person to whom Rafael most nearly reported.
Vos was a pleasant, energetic Dutchman, a former programme manager at a large aerospace firm, who had an excellent reputation for getting difficult things done. He had seen an early version of Anjali's analysis, and he wished, he said, to discuss how it might be presented. He did not ask Rafael to change any number. He was far too experienced for that. What he suggested was a matter of framing. The plant was, after all, still in its commissioning phase, and commissioning yields were known to be low. The trend was strongly upward: from twenty-six to forty-one in three months. If that trend were projected forward, it would reach a hundred well before the crew's arrival. Anjali's model, with its assumptions about seal and electrode wear, was one possible projection; the trend line was another; and it would be reasonable, Vos thought, to present both, with the trend line as the central case and the model as a conservative scenario.
He said, too, that he did not need to remind Rafael what a revised projection would mean for the crewed schedule, or for the consortium's next funding round, or for the plant's own future, which depended on its being seen as a success. He said that he trusted Rafael's judgement entirely.
Rafael said that he would think about it, and thanked him, and put the telephone down, and sat for a long time looking at the beech trees.
*
We are accustomed to think of honesty as a simple virtue, a matter of saying what one believes to be true; and in the plain cases it is. But the cases that test us are seldom plain. They are cases in which we are asked not to lie but to emphasise, not to falsify but to select, not to deceive but merely to let others deceive themselves in a direction that serves everyone's immediate interest. And in such cases the temptation is the more insidious because it presents itself as reasonableness, as a refusal of pessimism, as loyalty to colleagues who have worked hard and deserve to be believed in. The trend line was not a lie. It was a real feature of the data. It was simply not, in Rafael's judgement and Anjali's, the most probable future, because it ignored everything they had learned about why the yield was low.
There is a further consideration, which Rafael's grandfather had understood better than most of the economists who have since written about such things. A number given to the people upstream of a decision is not merely a statement; it is a gift, or a burden, to everyone downstream. The shipper who received an inflated yield estimate from the Douro would buy too few grapes elsewhere, would promise too much wine to his customers, would find himself short in the autumn and would pass the shortage on. Somewhere at the end of that chain, someone who had never heard of the farmer at Pinhão would pay for his optimism.
At the end of Rafael's chain were four people who would land at the lunar south pole in a little under two years, expecting to find tanks of oxygen waiting for them.
*
He spoke to Anjali that afternoon, in the small meeting room at the end of the operations floor. He wanted to know how confident she was in the model, and she told him, with the directness that he had come to value in her, that she was confident in its structure and less confident in some of its parameters, and that the true outcome might be somewhat better or somewhat worse than its central case, but that she could find no plausible set of assumptions under which the plant reached a hundred per cent before the crew arrived. He asked whether she thought the trend line was a reasonable projection. She said it was a reasonable description of the past and an unreasonable prediction of the future, because the improvements that had produced it were mostly one-time fixes that could not be repeated. He asked whether she had been asked by anyone else to present it differently. She hesitated, and then said that she had not been asked, exactly, but that she had heard that her model was considered pessimistic, and that she had wondered whether she ought to revise it.
He told her not to. He told her that her model was the best analysis the team had, and that he would present it as the central case, and that if anyone asked her about it, she should say exactly what she had just said to him.
She looked at him with an expression he could not quite read, and thanked him. It occurred to him afterwards that she might have been afraid of what his decision would cost him, and that she was perhaps right to be.
*
The week before the review was among the least pleasant of Rafael's professional life, though nothing in it was dramatic. He rewrote his presentation four times. He spoke to Vos twice more by telephone, in conversations of unfailing courtesy in which Vos expressed his confidence in Rafael's judgement and his concern that the consortium's partners would react badly to what they might perceive as an abrupt change in the outlook. Rafael said that the outlook had not changed; only the evidence had. Vos said that he understood, and that he hoped Rafael understood the position he was in. Rafael said that he did.
He spoke to his wife, Clara, in Lisbon, where she taught at the university and where he went home every second weekend. She listened to the whole story, as she always did, and then asked him what his grandfather would have done. He said that his grandfather would have told the shippers the truth and then gone out and sprayed the vines twice as often. She laughed, and said that she supposed that was his answer, then.
And he spent a good deal of time with the plant itself, or with its representation on the screens of the operations floor: the camera views from the robots, showing the grey crater floor under the long low sunlight, the black shadows of the furnaces stretching for tens of metres across the regolith; the temperature plots; the oxygen tank levels, rising slowly, far more slowly than the plan required. He found it calming, in a way he could not have explained, to look at the actual thing in the actual place, so far away, doing its stubborn and inadequate best.
*
The review took place in a conference room in the consortium's headquarters in Brussels, with thirty people at the table and as many again on video. Rafael presented the plant's performance, the five months of adjustments, the improvement from twenty-six to forty-one per cent. Then he presented Anjali's model, as the central case, with its projection of sixty per cent at the time of the crew's arrival and its upper bound of seventy. He presented the trend line as well, because it was a real feature of the data and he did not think it honest to omit it, but he explained why the team did not believe it, and he said plainly that, in his professional judgement, the plant would not produce the oxygen the crewed schedule required.
He then said, because it seemed to him that a manager who brings only bad news has done half his job, what he thought could be done. The schedule could be delayed by fourteen months, to allow more oxygen to accumulate at the reduced rate. Or the shortfall, perhaps four tonnes, could be made up by oxygen launched from Earth on one additional cargo flight, at a cost he estimated. Or some combination of the two. He said that he would recommend, in any case, that the redesign of the seals and anodes begin immediately, so that a second-generation plant could be delivered in time for the second crewed stay.
There was a silence when he finished, of the kind that he had been expecting, and then a great many questions.
*
It would be pleasant to report that Rafael's candour was immediately rewarded, and in a sense it was, though not in a way that felt like reward at the time. The partners were unhappy, and said so. Two of them questioned whether the plant's management had been as effective as it might have been. One suggested, in a tone that was not quite an accusation, that the problems with dust and electrodes ought to have been foreseen. Vos defended the team, which Rafael had not expected and for which he was grateful, though he noticed that Vos was careful to describe the outlook as the plant team's assessment rather than his own.
In the weeks that followed, the consortium chose the second option: an additional cargo flight carrying oxygen from Earth, which would cost a great deal of money and would be paid for by reallocating funds from other parts of the programme, including, Rafael learned, a part of the plant's own budget for the following year. The crewed schedule held. The seal and anode redesign was approved, at a lower level of funding than Rafael had requested. Anjali was asked to lead it.
Rafael was not dismissed. He was not promoted, either, at the round of appointments the following spring, though he had been expected to be; the position he had been considered for went to a man from one of the partner agencies who was said to have a more positive approach. He minded this more than he would have admitted to anyone except Clara, and less than he had feared.
*
The plant reached sixty-three per cent of design yield fourteen months later, a little better than Anjali's central case. By the time the crew arrived, it had produced and stored a little over six tonnes of oxygen, which, together with the four tonnes brought from Earth, was enough. The crew's ascent vehicle was fuelled partly with oxygen made from the Moon itself, the first time that had ever been done, and the moment was celebrated with a good deal of public ceremony, in which the plant was described, accurately, as a historic success.
Nobody at the ceremony mentioned that the plant had produced only sixty-three per cent of what it was designed to produce, or that the crew's safety had depended on a cargo flight that had been ordered because a manager in Luxembourg had declined to present a trend line as a forecast. There was no reason they should have. The success was real. It was simply not the success that had been planned, and the difference between the two had been absorbed, as such differences usually are, by people whose names did not appear in the press release.
*
It is one of the quieter consequences of honesty that it so often goes unremarked, since its chief effect is the prevention of a disaster that, having been prevented, never becomes visible. Had Rafael presented the trend line as his central case, the cargo flight would not have been ordered in time; the shortfall would have been discovered some months before the crew's departure, when it could no longer be made good; the crew's mission would have been delayed or curtailed, at far greater cost and with great embarrassment to everyone concerned. None of this happened, and because none of it happened, nobody had any reason to think about it. The good that Rafael had done consisted entirely of things that had not occurred.
He did not, on the whole, resent this. He had been raised among people who understood that the yield of a vineyard is known only at harvest, and that the man who reports it honestly in spring is not thanked in autumn, because by autumn everyone has forgotten what he said. What mattered, his grandfather had taught him, was that the ledger was true, and that the wine was there when the shippers came for it.
He kept, on the shelf above his desk in Luxembourg, the old man's ledger, which his mother had given him when the vineyard at Pinhão was sold. Its last entry was for the harvest of 1998, in a careful, shaky hand: the yield of each of eleven terraces, in kilograms, beside the estimate given to the shippers that spring. The estimates were all a little low. He had sometimes wondered whether that had been caution or cunning. He had come, in the end, to think that it did not much matter, so long as the numbers were the old man's own, and so long as he had been prepared, in autumn, to stand behind them.