6.1The claim
Mackay's chapter contains three things a reader can turn against its own verdicts: a concession that some schemes were sound and failed by their moment, an admission that buyers' reasons were not what they seemed, and a footnote in the later edition that amends the chapter's ending after an event later than the first edition. Read with the records' two forward-looking remedies and their revisit conditions, they add to the three-way debate in one direction. A reason written before an outcome is the only material that resists being rewritten by the outcome, but it does not by itself say whether the decision was good, and the records' own revisit conditions return the judgment to the question of what was "knowable" or "clear" at the time, which is the question hindsight distorts.
6.2Sound schemes at the wrong time
In the middle of his account of the minor companies, which he calls "Bubbles", Mackay pauses:
Some of these schemes were plausible enough, and, had they been undertaken at a time when the public mind was unexcited, might have been pursued with advantage to all concerned. But they were established merely with the view of raising the shares in the market.
The first sentence separates the quality of a scheme from its outcome: a narrator in the middle of a history of folly admits that a collapse does not prove a project foolish. The second takes the admission back by moving the judgment to motive, founded "merely" to raise shares, a motive stated, not shown, and resting like the directors' imagined confidence on the outcome. The voyage record asks for exactly the separation of "a bad outcome" from "a bad decision"; Mackay makes it for one sentence and lets it go.
6.3Reasons that were not what they seemed
Another passage records that buyers' reasons could not be read from their purchases:
But it did not follow that all these people believed in the feasibility of the schemes to which they subscribed; it was enough for their purpose that their shares would, by stock-jobbing arts, be soon raised to a premium, when they got rid of them with all expedition to the really credulous.
Mackay's point is moral: some buyers were sharpers, not dupes. The passage also shows that two identical purchases may have had opposite reasons, and that the outcome cannot tell them apart; only a record of the reason made at the time could. Mackay infers the difference from general knowledge, not from any such record.
This bears on the voyage record's third option, "the cumulative outcomes of similar decisions". Voyages that look alike may have been sent for different reasons, and cumulative outcomes do not say which decisions belong together. Any scoring of the kind Brier proposed works on forecasts stated before the event, and a forecast is a reason in numerical form. A series of outcomes without the reasons that produced them is only a series of outcomes, and Mackay's chapter shows what a teller makes of one: a crowd, and a madness.
6.4The footnote that amends the ending
Mackay's chapter ends with a reflection on what came after:
But it was a long time before public credit was thoroughly restored. Enterprise, like Icarus, had soared too high, and melted the wax of her wings; like Icarus, she had fallen into a sea, and learned, while floundering in its waves, that her proper element was the solid ground. She has never since attempted so high a flight.
The last sentence is a forecast disguised as history. The next paragraph qualifies it, noting later schemes that threatened "results as disastrous" but "were happily averted before it was too late". To that paragraph the edition read here attaches a footnote:
The South-Sea project remained until 1845 the greatest example in British history of the infatuation of the people for commercial gambling. The first edition of these volumes was published some time before the outbreak of the Great Railway Mania of that and the following year.
The words "remained until 1845" imply that after that year the South Sea project was no longer the greatest example, which sits uneasily with "She has never since attempted so high a flight". The transcription does not say whether that sentence was left standing from the first edition or revised, and the article assumes neither. What the page shows is that the later knowledge was not worked silently into the body. It was added in a note marked by its own reference to the first edition, so a reader can see that the author's view changed after a later event.
That is, in form, what the voyage record's example judgment asks: information available at the time and information learned afterward "are written down separately". The device is modest, and it is what the Broderick letters lacked and the company's altered books had lost.
6.5The records' remedies
Each record offers a remedy that does not depend on the single outcome. The voyage record's third option is "Evaluation Based on Cumulative Outcomes". The post-mortem record's third option is to "Mandate a record before the decision". The first corresponds to the second position in the literature; the second is the practice on which all three positions depend, and which the third doubts.
The cumulative option does not judge a voyage by one storm. It has the limit found above. Outcomes in a series need reasons to be grouped, and the scene keeps the reasons in the log; a judgment that ignored the log would discard the evidence of which voyages were similar decisions and which only looked similar. The record's second and third options are then less rivals than stages: the log tells which decisions belong in the series, and the series tells how good that kind of decision has been.
The mandate meets the third position head on. If defensive documentation can take the place of judgment (Power 2004), and documents can be written to persuade that risk is under control (Clarke 1999), a rule requiring a record before each trade invites both: a trader who knows a losing trade will be reviewed has reason to write reasons for the review rather than for the trade. The record names a different cost of the mandate, in its revisit condition.
6.6What would make the judgment be reviewed
Each record ends with a condition on which its example judgment would be reviewed. The voyage record's reads:
If the log contains a record of ignoring risk signals that were knowable at the time, it is reviewed again.
The condition turns the log from a defence into evidence that might convict: if the trader wrote down a signal and sailed anyway, the partner's charge gains ground. It is reasonable, and it contains the word that makes it hard to apply, "knowable". Whether a signal was knowable at the time is the very judgment that knowledge of the outcome distorts, on Fischhoff's account. A reviewer who knows the ship was lost to a storm, reading a log entry about weather, will tend to find the storm knowable. The condition does not say who decides that, or from what material, and so hands the question back to the reviewer the example judgment was written to guard against.
The post-mortem record's condition has two parts:
If the reconstruction confirms that the signal was clear at the time, or if the recording obligation substantially harms decision speed, it is reviewed again.
The first part repeats the review's claim that "the signal was clear at the time" and makes the judgment depend on a reconstruction confirming it. Using only material from before the decision is the reconstruction's strength; the word "clear" is its weakness, since clarity is what the example judgment says "follows a result". A reconstruction by people who know the result is still made by people who know the result, and the condition does not say how the result would be kept out.
The second part concedes that a mandate has a cost, and names speed, not defensive writing. A mandate that slowed decisions would show itself in time taken; one that turned records into defences would not show itself until a review found reasons written for the review. The condition watches for the first only.
Neither condition gives a number, threshold or time by which an observer could tell it had been met, as is true of most conditions in the study's records. Here it has an edge: the conditions turn on "knowable" and "clear", the two words the records' own thread treats as unreliable once the outcome is known.
6.7What the case adds
Read against its own verdicts, Mackay's chapter shows the value of a reason written before: only it tells a sound scheme launched at a bad moment from one founded to deceive, and a believer from a sharper. His later footnote shows the form the voyage record recommends at work in a printed book. For the three-way debate, the case adds a point about where the positions meet. Judging by outcomes in a series needs prior reasons to know which decisions form a series. Judging from what was known at the time needs a judge who can say what was knowable without being moved by the outcome, and the records' revisit conditions require such a judge without supplying one. The doubt about records written to defend applies most directly to the post-mortem record's mandate, whose revisit condition watches for a different cost.