Which Development Gates Earn Their Place

How to read this sheet

Three numbers are the answer and they fit on a card. Every gate has an order quantity above which it pays and below which it does not, and once you have those three quantities the sheet has done its work - a development arrives, somebody looks at the bulk behind it, and the gates follow. That is a different thing from a procedure, because a procedure is the same for everything and this is not. Why it is a question about size. What a problem costs at the bench is what the bench cost. What it costs at first bulk is mostly the cloth, and the cloth is the order. So the escalation everybody quotes as one to ten to a hundred is not a rule of thumb here, it falls out of the arithmetic, and on a large order it is much steeper than a hundred. Nothing in this sheet asks you to believe the rule; it asks what a metre costs to put right. What a gate is actually worth. Not what it catches. A gate that finds nearly everything, one day before a gate that would have found it almost as cheaply, is worth close to nothing - and a gate that rarely finds anything but finds it before forty-six thousand metres exist can be worth its cost many times over. That is why the sheet solves the whole sequence rather than scoring each gate: the value of a gate depends entirely on what comes after it, and removing one changes the worth of all the others. Read the second reading before arguing about the first. Days, not money, are why gates get dropped - a pilot costs six days and the customer is waiting, and the six days are charged to nobody. The sheet prices the same book with the calendar taken out of it and reports how many gates would come back. If that number is not nought then the argument in your building is not about testing at all, it is about the development calendar, and it should be had there. What is not modelled. Problems are treated as independent, which flatters the sheet: in practice an unlevel dyeing and a shade failure often have one cause, and finding one finds the other. A gate is either run or not, where in reality it has a depth - half a panel is a real option and is not on this sheet. Iteration is charged as a full repeat from the beginning, which is right for a recipe change and harsh for a small correction. And nothing here values what a gate teaches: a pilot that finds nothing still tells the plant how the quality runs, and that knowledge is real and is not counted.