The disaster that doesn't happen has no name.

The price of one fact

Take the single question this whole track has orbited. Is there rock under the northern half of the site?

Ask it before the bid and the answer costs $6,800, and it changes the price you tender. You either price the rock and lose the job to somebody who didn't, or you find clean ground and bid with confidence. Either way you are making a decision instead of hoping.

Ask it in Week 8, after award and before the rig mobilises, and the answer still costs $6,800. It no longer changes the price, but it changes the method, the sequence and the evidence file. Week 13 measured that at $5,033 at the eightieth percentile.

Ask it while the auger is turning and the answer is free. The ground tells you at no charge, one bore at a time, at $2,850 each.

Same fact. Three moments. Three completely different things you can do with it, and the useful ones are all at the start.

Information does not get more expensive as a project runs. It gets less useful, which is worse, because the invoice arrives regardless.

Why nobody acts early

So why does everybody know this and nobody do it?

The usual answers are that people are short-termist, or that project managers are optimists, or that there is never time. None of those is quite right, and none of them is useful, because you cannot fix any of them on a Monday.

The real answer is a scoreboard.

Think about what actually gets said afterwards. You spend $6,800, the north comes up clean, and what people remember is that you spent $6,800 on nothing. You spend it, you find rock, you manage it into the sequence and it costs $32,000 instead of $48,450 — and what people remember is that the rock turned out not to be that bad. You spend nothing, the ground is clean, and you were right. You spend nothing, you hit rock, and it was unforeseeable.

Four outcomes. Exactly one of them produces visible credit, and it is the one where you did nothing and got lucky.

That is the whole problem, and it is not a failure of anybody's character. A prevented disaster leaves no evidence behind it. There is no line in any cost report anywhere called money we did not spend on a thing that did not happen, and there never will be.

What a planner actually controls

Which brings us to the honest limit of this job, and it is worth being clear about because a great deal of risk training pretends otherwise.

You do not control the confidence level. Week 12 was blunt about that — choosing P50 or P80 is a statement about how much of the company's capital goes on a contract, and that belongs to somebody with a balance sheet.

You do not control whether the boreholes get drilled. You do not control the contract form, which was chosen before you arrived. You do not control whether the wayleave letter gets sent, because that is a commercial manager with his own week.

What you control is whether the choice is visible, priced and dated at the moment it is still cheap.

That sounds like a smaller job than it is. Almost every expensive thing in this track happened not because somebody decided wrongly, but because nobody decided at all. The boreholes were never declined. They were never proposed. The register was never judged incomplete. Nobody ever looked at it.

A decision that was never taken cannot be defended, and it also cannot be attributed. Turning a non-decision into a decision is the single highest-leverage thing anybody in project controls does, and it costs one page.

The page, and the three sentences

Eight lines. The risk. What we know today. What it would cost to know more. What that buys, at the mean and at the percentile the company actually funds. What we recommend. Then three blanks: the decision, who took it, and when.

The blanks are the point. They are not yours to fill in, and leaving them empty on somebody's desk is what converts a vague worry into an attributable choice.

And when you hand it over, three sentences do most of the work.

Here is what we know and here is what we do not. That separation is Week 8, and it is the one nobody makes.

Here is what it costs to find out, and here is what that is worth. Not a warning — a trade, with two numbers on it.

This is your call, and I will write down whatever you decide. That last one is not passive-aggressive if you mean it. It is the sentence that makes the record exist.

What this track was

Track 1 taught you to see a delay before it arrived. Track 2 taught you to see an overrun before it landed. Both of them ended in the same place: a number that moved, and a report explaining why.

This track was about the eighteen weeks before the number moves.

The contingency in Week 1 was $47,553 and the honest answer was $183,167. That gap was not created by bad luck in month six. It was created on the day somebody took five percent of a figure without checking the list underneath it, and every week after that it got harder to fix and more expensive to raise.

Risk work is not forecasting. Forecasting tells you where you are heading. Risk work is the argument you have while there is still time to head somewhere else — and the whole difficulty is that the argument has to happen when it looks least necessary, funded by somebody who will never see proof that it worked.

Do it anyway. Write it down. Put a date on it.

Key takeaways

✔ Every finding in eighteen weeks was available on day one. Nothing here failed on arithmetic.
✔ The same fact costs $6,800 before the bid, $6,800 before the rig, and nothing at all once it is too late to use.
✔ Of four possible outcomes to a risk decision, only one earns visible credit — doing nothing and being lucky.
✔ A prevented disaster leaves no evidence, and no cost report has ever had a line for it.
✔ Most expensive things happen because nobody decided, not because somebody decided wrongly.
✔ You do not control the confidence level, the contract or the drilling. You control whether the choice is visible, priced and dated while it is cheap.
✔ Separate what you know from what you do not, state the trade with two numbers, and leave the decision blank for somebody else to sign.

What's coming next

Week 15 left a question hanging on purpose. We found five risks with a contractual mechanism behind them and put clause numbers against each one — and then noted that an entitlement is not the same thing as an entitlement you can still use.

Every one of those clauses has a notice attached to it. A period, a form, a recipient, and a consequence for missing it that is considerably harsher in the 2017 editions than most people on site realise.

The rock is worth $48,450 and the contract may well say it is the employer's. Whether you ever see the money depends on a letter, a date, and a set of records that either exist or do not.

That is Track 4, and it starts where this one stops.