A WORKED EXAMPLE
The bet you only make once
A worked example – Why ranking the risks of an irreversible, one-time bet is not prudence but the move most likely to sink it – and what takes its place.
Start with sound advice, because this is where it goes wrong. When a decision can be made many times – a price tested, a campaign run again next quarter, a process tuned over a hundred cycles – the right move is to learn the odds and back the likeliest outcome. Run it often enough and the average becomes real: the bad results are paid for by the good ones, and ranking your risks to arm against the largest is exactly prudence. Most of what is known about deciding well assumes this world, because most decisions live in it.
Then there is the other kind of bet, and it resembles the first closely enough that the same advice gets carried across without anyone noticing the join. A company commits itself to a product the market has not asked for, in a market that does not quite exist yet. The decisive thing here is not that the odds are poor. It is that the bet is made once, and that failure is a door that does not reopen.
In that kind of bet the odds describe a crowd you are not in. Ventures like this succeed one time in five is a true sentence about five hundred ventures; it says nothing about the single one being run once, because there is no second and third and fourth attempt across which the average can come true. The number is real, and it is not about you. Treating it as though it were – backing the likeliest path, bracing for the most probable failure – quietly assumes a repetition the bet does not contain.
The move that follows from that assumption is the one that does the damage. Faced with a list of risks, the natural thing is to rank them and point the enterprise at not dying of the largest. But to aim the whole of a venture at the one failure you can name is to narrow what it can still see to a single hazard – and the failures that decide a genuinely new bet are, by their nature, the ones not on the list, because they come from the part of the situation that has not formed yet and so cannot be priced.
A bet braced against its likeliest death is rigid in precisely the direction the unpriced death will come from.
There is one device worth the name, and it is not prediction, which is not on offer. It is to write the failure in advance: to take the venture as already dead and set down, in detail, why. Done without flinching it does one specific thing – it tends to reveal that the many different-looking risks are the same risk: a single buried assumption nobody had written down, breaking in several places at once.
Ten dangers resolve into one, and the one is often so foundational it had never been stated, only assumed. That is worth more than any ranking, because it finds the load-bearing belief the whole bet rests on while there is still time to test it.
What it cannot find is the failure that has not formed. Nothing can; that is what makes the bet a bet. So the discipline is not to predict better but to stay able to turn – to build the venture so that when the unpriced thing arrives, and something unpriced always arrives, there is still room to move, rather than everything having been spent bracing for the wreck that did not come.
You can tell the two kinds of bet apart with one question. If this fails, can it be run again? If yes, learn the odds and back the likeliest; the average is on your side. If no – if failure is the door that does not reopen – then the odds are describing a crowd you are not in, and the question is no longer which failure is most likely but which failure could not be survived, and whether there is still room to turn when it comes.
The first question keeps you arming against the wrong death. The second is the only one you can afford – if it’s a bet you make only once.
This is one of the moves the programme makes visible. The full account of what it is and how it works is here.