起源故事 · 童子军营地里的一颗骰子
Origin Story · A Single Die at a Boy-Scout Camp
《目标》里，厂长 Alex 想让童子军们明白工厂为什么总是延期，于是发明了这个游戏：每个孩子是一道工序，掷骰子搬火柴。
孩子们都以为"平均每人 3.5、走十轮就该有 35 根"，结果末端只有二十几根，中间却堆了一大摞。
这一刻 Alex 彻底懂了 —— 平均产能从不等于实际产出。只要存在依赖和波动，链条越长，损失越大。这是约束理论最锋利的一课，也是 Six Sigma 减小波动、TOC 设置缓冲的共同出发点。
In The Goal, plant manager Alex wants the boy scouts to feel why his factory is always late, so he invents this game: each kid is a workstation, rolling a die and moving matches.
The kids assume "3.5 each on average, so ten rounds should deliver 35 matches" — but only twenty-something arrive at the end, while a heap of matches sits in the middle.
That is the moment Alex really gets it: average capacity is never the same as actual throughput. Once dependency and variability exist, the longer the chain, the bigger the loss. It is the sharpest lesson in the Theory of Constraints, and the shared starting point for Six Sigma (shrink the variation) and TOC (cover it with buffers).
1 一排碗传火柴：看它在中间堆起来 A Row of Bowls Passes Matches: Watch the Pile Build in the Middle
第 0 轮Round 02 累计产出：实际永远追不上理论平均 Cumulative Output: Actual Never Catches the Theoretical Average
蓝色虚线是理论累计（每轮 +3.5），红线是末端碗的实际累计产出。两线之间不断扩大的缺口，就是依赖事件 + 统计波动吃掉的产能。 The blue dashed line is the theoretical cumulative (+3.5 per round); the red line is what the last bowl actually delivers. The widening gap between them is the capacity eaten by dependent events plus statistical fluctuations.
3 现实里的火柴与碗 Matches and Bowls in the Real World
依赖事件：每道工序都依赖上一道交付，没料就只能空等 —— 上游的"不足"会一路传到末端。
Dependent events: every station depends on the previous one's hand-off. No material means idle time — any upstream shortfall flows all the way to the end of the line.
统计波动：再标准的工序也有快有慢。波动本身不可消除，只能减小或用缓冲吸收。
Statistical fluctuations: even the most standardized step runs hot one moment and cold the next. You can't eliminate variability — you can only shrink it or absorb it with a buffer.
为什么需要缓冲：在约束前留一堆"在制品缓冲"，让它永不断料 —— 这正是 DBR 排程的核心。
Why buffers exist: park a stack of WIP just upstream of the constraint so it never starves — that is the core idea of Drum-Buffer-Rope scheduling.
≠ 产出≠平均产能：拿平均产能去承诺交期，必然延误。排程要按瓶颈实际节拍，留出波动余量。
≠ Throughput ≠ average capacity: promise delivery dates off average capacity and you are guaranteed to miss them. Schedule to the bottleneck's real takt and leave room for variability.
一句话In One Line
这颗骰子量化了上一页"平衡线悖论"里那笔说不清的损失：当依赖事件遇上统计波动，系统的实际产出必然低于各环节平均产能。
一道工序的"多"帮不了全链（下游接不住、就堆成库存），但它的"少"会立刻向下传染。链条越长，缺口越大。
所以两条出路：用 Six Sigma 减小波动，或用 TOC 在瓶颈前设缓冲、按瓶颈节拍排程(DBR)。盯住约束、保护约束 —— 这就是约束理论给制造业的整套答案。
This one die quantifies the loss the previous page (the Balanced-Line Paradox) couldn't put a number on: when dependent events meet statistical fluctuations, system throughput must fall below the average capacity of any single step.
A station's "extra" never helps the whole chain — downstream can't catch it, so it piles up as WIP — but a station's "shortfall" infects the rest of the line instantly. The longer the chain, the wider the gap.
Two ways out: shrink the variability with Six Sigma (smaller σ), or buffer the bottleneck and schedule to its drumbeat with TOC's Drum-Buffer-Rope. Focus on the constraint, protect the constraint — that is TOC's full answer to manufacturing.
常见误用Common Mistakes
用平均产能算交期与产能。实际产出永远低于平均，排程要扣掉波动损失、留缓冲。
Sizing capacity and committing dates from average throughput. Actual output is always below the average — subtract the variability tax and budget a buffer when you schedule.
只盯每道工序"效率"。局部效率不等于系统产出；要看末端实收，不看中间忙碌。
Chasing "utilization" at every station. Local efficiency is not system throughput. Measure what ships at the end, not how busy the middle looks.
无视在制品堆积，只看产量。中间碗的积压就是损失的征兆，是该设缓冲、减波动的信号。
Ignoring WIP and watching only finished volume. A pile in the middle bowls is the symptom of the loss — and the signal that it is time to place a buffer or attack the variability.