Two pieces of work,
and what each one changed

The question the client brought was not whether the government would survive. It was who would be running Venezuela in six months, and whether that government would keep paying the country’s debts and honour the contracts signed by the one before it.

Through the second half of November the same pattern kept arriving from sources that do not usually agree and were not talking to each other. Families of senior officials crossing the western border into Colombia. Ministries in Caracas destroying records. Senior figures preparing somewhere to go in the south, towards the Brazilian border. None of it would have cleared our bar alone. Together, from independent channels, it did.

What mattered was the gap between the three readings. US pressure on Caracas was climbing and jumped again in the last week of November. The circle around Maduro had held together all autumn and began to come apart at the start of December. The armed forces went the other way entirely, and on the day we made the call they looked more likely to stay with him than at any point we had measured.

Our judgement was that the army could not hold that line with Washington pressing on one side and Maduro’s own circle splitting on the other, and that when it broke it would break fast. That is why we called a sudden rupture rather than a slow decline.

The signs were on the table ninety days out. We made the final call on 2 December, thirty two days before it happened. Maduro was taken on 3 January. The successor came from inside the ruling coalition, the armed forces stayed intact, and the contract and counterparty assumptions the client had made held.

The client had asked one question: would Maduro still be in power by a given date. We gave them our answer on 2 December, they bought while the market was still pricing collapse, and they made a substantial margin on the debt when the succession held.

Ras Laffan is the size of a town. “An energy facility” there could mean a switchyard, a storage tank, an office block or one of the liquefaction trains that turn gas into the cargo a ship carries away. A switchyard is a news item. A train is a supply interruption. Nothing in the early reporting said which one had been hit.

Our joint venture partner had people on the ground who said a train had been damaged. Human reporting is the fastest thing available and the hardest to act on by itself. The person telling you is rarely positioned to see the whole plant, and in a week like that one, everybody on site has heard something.

So we went after the things that cannot be talked into existence. Thermal imagery showed where heat had stopped being produced and where it had not. Successive satellite passes showed the physical state of the structure rather than the claim about it. AIS tracking showed which carriers berthed, which loaded and which left, and on what schedule, because a train that is down eventually shows up as a ship that does not sail.

The three agreed with each other and with what the partner’s people had said. Within seventy two hours of the first report the company had the train identified, the damage established, and a line of sourcing written beside each claim rather than a single confident sentence.

That was the difference that mattered. The company was not deciding whether to believe a rumour. It was deciding what to do about a fact it could show to its own board, and later to anyone who asked how it had known.