The ball that takes the wicket rarely does the work

Dismissals get credited to the delivery that beat the bat, when the setup happened three overs earlier and cost eleven runs.

An outswinger takes the edge, the slip cordon celebrates, and the replay package that runs at the interval shows that delivery from four angles. Beautiful ball, everyone agrees. What the package does not show is the six deliveries before it, all pitched a fraction fuller and straighter, all defended comfortably, all of which moved the batter's weight forward and made the one that left him unplayable.

The dismissal is the receipt. The setup is the purchase.

This causes real damage to how bowling is evaluated. Figures of one for fifty two in eighteen overs look mediocre and can describe a spell in which a bowler bowled superbly, made the good batter uncomfortable for an hour, and created the conditions for a partner at the other end to take three cheap wickets from batters who had been ground down. Cricket has never had a good way of crediting that, and averages will not do it, because the average punishes the runs conceded while making the setup.

The mechanism is mostly about the batter's decision-making rather than the ball itself. A batter who has been driving for twenty minutes is committing forward earlier, which is exactly what you want before you bowl one wider and shorter. A batter who has faced four bouncers is watching the length differently, which is why the yorker after them works. In both cases the bowler paid runs to shift where the batter is looking, and then collected.

So the spell has to be planned as a sequence and not as six individual attempts, and this is where a captain either helps or ruins it. Pulling a bowler off after three overs because they went for fifteen destroys the sequence, and rotating a bowler between ends interrupts it too. The bowler who has spent eleven runs setting up a plan needs the over that cashes it in.

The analytics are starting to catch this, sort of, with false-shot rates and control percentages, which are better than nothing and still miss the sequencing.

My rough test is simpler than any of that. Watch where the batter's front foot goes over ten balls. If it is going further and further forward, somebody is building something, and the wicket that follows will get credited to the wrong delivery.