Fifty years of a deck of cards that engineers actually use

$ eldr --cite eno-1975

In 1975 Brian Eno and Peter Schmidt printed a box of cards, each carrying one instruction. "Honour thy error as a hidden intention." "Use an old idea." "Work at a different speed." You draw one when you are stuck and you do what it says.

Fifty years later it is still in print, still on the desks of producers and architects and software engineers who would walk out of a room if you described it as anything other than a tool. No belief system was ever attached to it. Nobody has to think the cards know anything, because they obviously do not, and the absence of that claim is the reason the thing survived.

That is the precedent worth studying here, and it is more instructive than any amount of market research.

The claim

The value of an external instruction comes from the constraint, not from the content — and a man who accepts that sentence has already accepted the mechanism this product runs on.

The cards contain no information about your specific problem. They cannot. What they do is remove one degree of freedom, and removing a degree of freedom is what breaks a loop: with "work at a different speed" imposed, you can no longer keep re-running the comparison you have been re-running, because the comparison has been ruled out of the search space.

This is the same result Levitt measured across 22,511 usable coin tosses, arrived at from the other end. The coin has no information either. On the decisions participants themselves rated as important, quitting and ending a relationship among them, those who got "change" acted about 11 percentage points more often and reported significantly higher well-being at six months. On the less important questions the action effect was larger and the well-being effect did not appear at all, which is the qualification worth carrying: the mechanism pays where the decision is heavy, and that is the only place this product claims to operate.

Falsifiable form: if the content mattered more than the constraint, then a deck with better instructions would outperform a deck with arbitrary ones, and Eno's specific 100-odd aphorisms would have been superseded decades ago by a researched version. They have not been. Multiple better-researched decks exist. The 1975 box is the one still in use.

Where the analogy breaks, and it breaks hard

Here is the part a marketing page would leave out.

Oblique Strategies operates in a domain where being wrong is nearly free. If the card says "use an old idea" and it produces a worse take, you delete the take. The iteration cost is one afternoon, the decision is reversible, and the deck is therefore free to be arbitrary because arbitrariness has almost no downside and a large upside in variance.

Your decisions are not like that. Quitting is not a take you can delete. Neither is leaving, or spending $40,000, or telling your co-founder what you actually want. A card oracle on an irreversible decision with a five-figure downside is not an instrument, it is a shrug with good typography, and any product that borrows the deck's charm while ignoring the difference in reversibility is quietly selling you something worse than it claims.

Which is exactly why this product refuses to be a deck. Fourteen inputs, a position, a confidence number, a window, and an argument against itself that has to name something checkable. The constraint mechanism is Eno's. The obligation to show work is not, and it is the obligation that the reversibility gap demands.

The case against the whole borrowing

Pointing at a fifty-year-old artists' card deck as your intellectual ancestor is a selection effect with a nice story. Thousands of creative tools from 1975 are gone; one survived, and survivorship tells you approximately nothing about mechanism. Worse: the people who use the cards are working inside a craft with a feedback loop measured in hours. There is no evidence whatsoever that an external constraint helps a man decide whether to leave his marriage, and the honest version of this page would say so in the first paragraph instead of the seventh.

Accepted, most of it. The survivorship point is correct and the domain-transfer point is the one that keeps me honest: the coin study is the only real evidence here for consequential decisions, it covers a self-selected population of already-stuck people, and everything else on this page is mechanism-level reasoning rather than measurement.

So the defensible statement is narrow. For a decision you have already been circling for months, an externally generated constraint reliably ends the circling, and ending the circling is measurably better than not ending it for that population. For a decision you encountered on Tuesday, none of this applies, and you should just think about it like an adult.

The operational version

If you want the deck's effect without the deck's irresponsibility: take the position this instrument returns and treat it as a constraint for exactly one week rather than as a verdict. Act as if it were settled, watch what you do, and note what you resist. A week of behaving as though the question is closed will tell you more about what you want than another month of behaving as though it is open.

That is Eno's instruction, restated for decisions that cost something: honor the constraint long enough to find out what you do under it.

The deck has one more lesson that has nothing to do with mechanism. It was sold as an object, once, for a price, with no claim attached and no relationship implied. You buy the box, you use it for thirty years, nobody sends you a notification. Half of what has gone wrong in the category this product sits next to comes from the opposite instinct — from needing you back tomorrow.

Reference: Eno, B., & Schmidt, P. (1975). Oblique Strategies: Over One Hundred Worthwhile Dilemmas. Still in print.