Quality is Fractal

The downward-transmission mechanism

"It's very rare that you see an organization where standards increase as you get lower and lower down," Alexandr Wang says. "Most of the time, when people realize their manager or their director don't really care, that removes the deep desire to need to care."¹ He built an explicit organizational value out of that observation: "Quality is fractal. High standards trickle down within an organization."¹ The word fractal is doing specific work in his account, because the same pattern of care or indifference reproduces at every level regardless of scale. The mechanism he describes is observational: people model the level of care they see in those above them, so the standard set at the top becomes both the floor and the ceiling for everyone below.

How Wang operationalizes it

Wang describes several ways he pays the cost of holding the standard personally. He says he still approves or rejects every single hire at the company, which he treats as the most direct signal of how seriously the top of the org takes quality because the marginal cost of each review is real and he pays it anyway.¹ He also describes personally reviewing data quality being sent to partner companies, acting as a final checkpoint, and frames customer unhappiness as "personally very painful" rather than as a business metric.¹ The personal gatekeeping is itself the message: the standard is visible because the founder is visibly willing to spend attention defending it.

The supply-chain and time-dynamics version

Todd Graves supplies a restaurant-operations account of the same principle, with two additions. The first is that the fractal runs upstream into the inputs. Because Raising Cane's sells essentially one item, he describes the care propagating down to the rawest level: exact bird weight and species, rigor-mortis timing on the bone, a 24-hour brine, pull-apart dough-ball bread, and tea sourced from three countries, plus a Middle East launch he says took two years purely to make the supply chain taste identical.² "It's not simple. It's focused," he says, which frames the standard as reaching every grower and bakery in the network rather than only the org chart.

The second addition is a time dynamic he calls death by a thousand cuts. The business runs on thin margins, "we're doing well if we make 10 cents on a dollar" in his own accounting, so he says there is perennial pressure to shave quality "just a little bit," and his refusal is that "if you start cutting a little bit here to save a penny, and a little bit here, it's death by a thousand cuts. Then your food one day is not craveable."² Indifference, in this telling, does not announce itself with one bad decision but accumulates invisibly until the product crosses a line. He describes setting care visibly from the top by running what he calls a restaurant support office rather than a corporate office, wearing a Cane's T-shirt to suit-and-tie conferences rather than separating himself from crew and customers, and taking failures personally, which is the same top-of-org signal Wang sends by reviewing every hire.

Relationship to adjacent standards concepts

Quality is fractal is closely tied to Standards Over Rules, which describes how an internalized bar rather than an enumerated rulebook is what actually carries a standard across a large org. It also connects to No Spare Customers, the customer-facing expression of the same refusal to let care lapse anywhere in the chain. Wang describes the org-chart transmission and Graves describes the supply-chain and time-decay versions, and both converge on the claim that care or indifference reproduces at every level and every stage.

Where the principle is contestable

The clearest tension is scale. Personal review of every hire and every data batch is a founder-level intervention, and the principle does not by itself specify what replaces that personal attention once the founder can no longer touch every decision. The mechanism explains why indifference spreads, but the operational cost of holding the standard rises with size, which is precisely where the founder's direct sampling becomes hardest to sustain.