Chapter 01: Beyond Theory—The Residual Autopsy**
“当两条曲线都声称自己解释了宇宙，真正开口说话的，往往是它们没能解释掉的那一部分。” ”When two curves both claim to explain the universe, it is usually the unexplained remainder that speaks.”
“看这张图。如果只看上方的曲线，你会觉得 MOND 赢了；但如果你看下方的残差路径，你会发现 MOND 正在‘流血’——那是一个系统性的、不可修复的逻辑溃败。” "Look at this image. From the top curves, you might think MOND holds the ground. But look at the residuals: MOND is 'bleeding'—a systemic, incurable hemorrhaging of its logical foundation.”
在第一部分里，我一直在用各种方法论证的其实是：ACM 并不是靠逐星系调参，才在 SPARC 样本上显得有效。它有一条统一、刚性、可复现的主干函数，能够在大多数样本中稳定工作。
但在与观测样本拟合的过程并不是一帆风顺的。主干模型的基本定型大概只用了整个第一阶段建模和实验的20%。真正花时间的并不是那 102 个已经明显更支持 ACM 的星系，而是剩下那批“看起来更像 MOND 赢”的对象。
如果只停留在传统的模型比较层面，我们很容易得出一个过于粗糙的印象：
大多数星系 ACM 更好
但仍然有几十个星系像是 MOND 的保留地
问题在于，这种看法默认了一个前提：**这些 MOND-favored 样本属于同一种物理现象。**
In Part I, what I have been rigorously demonstrating is this: ACM is not effective across the SPARC sample because of galaxy-by-galaxy parameter tuning. Instead, it possesses a unified, rigid, and reproducible "Trunk Function" that operates consistently across the majority of samples.
However, the fitting process with observational samples was not always seamless. Defining the "Trunk" model took only about 20% of the modeling and experimental time in the first stage. What truly demanded our time were not the 102 galaxies that clearly favored ACM, but the remaining subset that "appeared" to favor MOND.
If we remain at the level of traditional model comparison, one might walk away with a crude impression:
The problem is that this view relies on a flawed premise: that these MOND-favored samples belong to the same physical phenomenon.