Fun fact: The term bubble originally came from SCP-CN-1875 to satirize and deconstruct coined nouns and meta writing skill in pataphysis. In the kalpas series, though, the concept is used to describe narrative structures with more logos.
Out-universe, we write stories. Because of some thoughts or prearranged, these stories will more or less overlap. We call this prearrangement canon. Sometimes the same story by the same author is in different canons at the same time; Or there are those who continue the story of the original author, but put it in another canon. At this time, the "canon" that this part of the story belongs to is both A and B, or, C and not C, which is both mutually exclusive and compatible. At this time, it is difficult to describe the structure of the world view built by co-creation using simple "narrative layer" and "authors' headcanons".
In narratology, scholars call the personified image of a narrative text as the implied author. Like a novel that describes the feelings of a female protagonist in detail, the reader imagines the work as an inner monologue of author and imagines the author as such a woman. But it's possible that the real author wasn't a woman, or even that the story was written to mock about women like stereotypes . In addition, by different readers' interpretation, there are different implied authors in different readers' minds. In addition, since the creative ideas of the real authors are constantly changing, the image of the implied authors and the Settings they write are also constantly changing, which is a dynamic process.
This concept can be adapted to describe the situation mentioned earlier. The setting presented in a certain part of each story we called them lores. Then, the lores are regarded as some scattering pointd, and a part is taken as the anchor points, and a regression line of dynamic linear regression is drawn. This regression line is the story line to which some of these stories point, or canon. This setting can be called the implied canonline, written by the implied author.
In this case, the correlation concept of linear regression can be appropriated to describe the weights of different lore. The lore, which has a highly correlation, is naturally the core story, the main setting of that canon. Stories that are low-correlation but still relevant are unreliable narrations. As in some canons hub, there is an additional list for tales that fall within that setting but are not part of the main stolyline. It's basically like saying "you can refer to how these stories are written, but it's best not to use the plot that appears in them, which may conflict with the main stories."
Now, imagine you're playing a visual novel, and these unreliable narration are like the "what if" easter eggs in the game. Then let's turn the camera back to the main line, where there is an option to branch the story, which can only be taken one way in a round. After we choose one path, we can't see the story of the other routes, and we can only continue to follow the path shown on the screen, except for some meta games. But in this case, what should we call these different scenarios, parallel worlds? No.
The Almighty Demon in the game claims to have destroyed all parallel worlds, but what about a world in there that the heros has saved the entire universe if the players choose the right option button at pass? Either the devil's narrative is unreliable, or the devil is not almighty. A lot of times there is a situation where the Almighty in one story is not almighty in other story1, and this contradiction occurs if the other story is seen as a parallel world.
So in order to solve this problem, add a narrative concept to the universe concept , and the narrative priority is higher than the universe. Let's start with the concept of "narrative focalization".
The narrative lens (narrative perspective) is controlled by the narrator, and the narrative world under the focus of the lens will become the "actual world", which seems to be the predestined world. The world under the camera can all be classified into the same canon. Which one is the sci-fi world view, the other is the DND world view. In other words, magic in the sci-fi world, which is something that can happen but doesn't necessarily happen, exists the a whole layer. The narrative world that is not focused is the narrative blank, which can be filled arbitrarily. For example, the sci-fi world never mentions magic, but a magical era can be crammed into the blank of sci-fi world's history, only by the time the main story takes place, all magic is gone, and the record completely disappears, as if it did not exist at all.
The narrative lens will focus on the implied canonline, incorporating nearby lore. When you take a picture, the picture is a tale that contains these things. Successive tales are collectively called "muthos". The implied canonline then crosses into two lines at some point, both possibilly being placed into the lens. When the lens is turned here, select one of the switch lines and the other switch line is dropped. The abandoned branch has the potential to become the actual world, but it has not yet fully become the actual world.
At this point we name it dynamis world, calling this dynamis(potential) "possibility". Thence introduce the theory of "possible worlds" and modal logic. Each branching line from the implied canonline can be regarded as a possible world of that route relative to the one already chosen by the narrator. All possible worlds add up to a phantom-bubble.
The phantom-bubble is the equivalent of stretching the narrative lens to the maximum to build a complete implicit canon. Such as the true ending of a meta game, all the lines of the round, and even all the clues and plots of the series of games, are presented to the larger perspective in this true ending. Relatively speaking, the world in the story that is completely impossible to happen under this canon is called the impossible worlds, the irreal worlds, or just another phantom-bubble.
As mentioned earlier, a story can be divided into different canons or interpreted separately under different canons. At this time, different lore can be used to draw different canonlines and construct different phantom bubbles.
It should be noted that between the bubble and the bubble, it is nested, squeezed, overlapped and blended. The bubble cannot be specified, its boundaries are vague, has no form or idea. For example, concepts are often used as the authority of conceptual gods in creation, but in reality, the boundary between one concept and another concept is fuzzy, just like the paradox of the grain mound, which is bound by language and words and creates a specious phantoms. And a definable God will be killed by wordplay.
If writers hypostatize the concepts of narrative as the actual stuff, it would immerse readers in phantoms of terms. For example, we use "Foucault" as a unit to describe the concentration of substantialized particle which reflects the intensity of control between the narrative layers, but hence we limit our imagine in that discourse for searching what far-fetched stuff can match such term.
The narrator can create the illusion that many stories are in one big canon, like metafoundation. The modes of the world will change with the choices of the narrator, and people in the worldview will also enter different Settings. For example, in one game, the protagonist chooses to focus on the cultists, and if the protagonist doesn't prevent them, the world will be destroyed by the Cthulhu. But if you don't pay attention, then the world continues to function as if the cultists were just mentally ill.
It is like The Garden of Forking Paths, except that the path is called muthos, the fork is called the phantom bubbles, and the garden is called 妄域. The bubble will continue to birth and die with the change of narrative focus, leaving a phantom under the flash.
At the end, the Hume index is the degree of reality bending. Hume was a skeptic, so his name is used here as a unit of narrative reliablity of baseline reality which has the acknowledged laws of physics2 , or the suspicion of acknowledged axioms system. Back to the examples of the sci-fi world and the DND world given above. If the magic stuff appears in the sci-fi world, then magic itself is an "anomaly" in the sci-fi world, because it is contrary to the acknowledged axioms of science of the world. The more anti-science magic is, the more it is suspected that magic should not come up in this sci-fi world at all. That is to say, the Hume index at this point is essentially the suspicion, a measure of how far reality deviates from acknowledged reality.
有趣的事实：幻泡一词最初来自 SCP-CN-1875 ，用于讽刺和解构pataphysis中的生造名词和meta手法。虽然到了kalpas系列里，这个概念被用于描述带有更多逻格斯色彩的叙事结构。
在世界观外，我们写下一个个故事。因于某种思路或预先的安排，这些故事多多少少会有些交集。我们称这个安排叫做canon。有时候同一个作者写的同一个故事，会同时处于不同的canons；或者有的人续写了原作者的故事，但将其归入另一个canon中。这时候，这个故事部分所属的“canon”，即是A又是B，即是C又不是C，既互斥又可交融。这时候我们再用单纯的“narrative layer”和“authors' headcanons”就难以描述共同创作所构筑的世界观的结构了。
在叙事学中，学者们将一个叙事文本所展现的思想经人格化后的形象，称之为隐含作者。就如一个细腻地描写妇女主角感情的小说，读者将作者想象成这是作者内心独白，想象作者也是这么一个妇女。但可能真实的作者不是女性，甚至写这么一篇小说是为了嘲笑妇女的刻板印象。另外，经不同读者的解读，在不同读者心里就有不同的隐含作者。再加上由于真实作者的创作思路不断变化，隐含作者的形象及其所写的设定也不断变化，这是一个动态变化的过程。
这个概念我们可以挪用到描述之前所提及的情景中。我们将每个故事的某个部分所展示的设定，称之为lore。然后把lore视为一个个散点，然后取一部分作为锚点，并画一条动态线性回归的回归线。这条回归线便是这些故事某部分所共同指向的故事线，或者说，canon。可称这个设定为隐含canonline，由隐含作者所写成。
这时候便可挪用线性回归的相关性概念，来描述不同lore的权重。相关性很高的lore自然是核心故事，基础设定。偏离相关性但又有相关性的故事，则是“不可靠叙事（unreliable narrations）”。就像某些canon里面，会额外开设一处列表，用以放置处于该设定下但不属于主线的故事。基本上就像是说“你可以参考这些故事怎么写，但最好不要使用里面出现的剧情，这些剧情可能会和主线有冲突”。
想象一下你现在在玩一款视觉小说，这些“不可靠叙事”就像是游戏里的"what if"彩蛋。现在让我们把镜头转回主线上，这里出现了一个分支剧情的选项，在一周目中只能走一条路。我们选择其中一条路后，就看不了其他路线上的剧情了，只能随着屏幕所展示的道路继续走下去，当然有些游戏例外。不过在该例子中，我们该如何称呼这些不同走向的剧情，平行世界吗？答案并不是。
游戏里的全能魔王宣称毁灭了所有的平行世界，那么某个提供拯救全宇宙的选项里的世界，又该怎么说？要么魔王的叙述不可靠，要么魔王就不全能。很多时候会出现这么一个情况，跨世界观比战力的时候，一个故事里的全能者在另一个故事里不全能3，如果把另一个故事看成平行世界的话就会出现这种矛盾。
所以为了解决这个问题，在宇宙之外多加一层叙事概念，叙事优先级高于宇宙。我们先引用“叙事聚焦”概念。
叙事镜头由叙述者控制，镜头聚焦下的叙事世界，将成为“实在世界”，看似命中注定的世界。镜头下的世界，都可以归入到同一个canon里。哪篇一个是科幻世界观，一个是DND世界观。也就是说，魔法在科幻世界中，是可能出现但不一定出现的事物，在更高层面上是存在的。未被聚焦的叙事世界即是叙事空白，可以任意填充。如科幻世界从未提及魔法，但空白里可以塞入一个魔法时代，只是到了故事发生的时代所有魔法都没有了，记录也完全消失，好似根本不存在一样。
叙事镜头将聚焦在隐含canonline上，将附近的lore纳入其中。这时候拍个照，照片便是一个含有这些东西的tales。连续生成的tales合称“muthos”。然后隐含canonline会在某处分岔成两条线，都有潜在的可能被置入镜头中。镜头转到此处后，选择转入其中一条岔线，另一条岔线则被舍弃。被舍弃的岔线，具有成为实在世界的潜能，但尚未完全成为实在世界。
此时我们便称这种潜能为“可能性”，引入“可能世界”理论和模态逻辑。每一条由隐含canonline分出的岔线，相对于已被叙述者选择的路线，都可视为那条路线的可能世界。所有可能世界总和起来，就是一个幻泡。
幻泡相当于把叙事镜头拉到最大，构建出完整的隐含canon。如一个meta游戏的真结局，将所有周目所有线路，乃至所有系列游戏的线索和剧情，都呈示到更大视角的故事一样。而完全不可能发生在这个canonline里的设定，则是另一个幻泡了。
前文说到，一个故事可以分属不同的canons或者在不同canon下分别演绎。这时便可以取用不同lore，画出不同的canonlines，构筑出不同的幻泡。
要注意幻泡与幻泡之间，是嵌套、挤压、重叠、交融的。幻泡不可指定，其边界是模糊的，或者说它不具有 form或idea。就如创作中常常以概念作为概念性神明的权柄，但现实里，一个概念和另一个概念的边界是模糊的，就如谷堆悖论一样，被语言文字束缚而营造出似是而非的假象。而一个可定义的神又将被文字游戏所杀死。
而把涉及叙事的概念实在化，就如用“福柯”作为一种粒子，其浓度反映叙事层之间的控制强度一样，反而是令我们自己限制在话语里，只为了找到能和这个词牵强附会上的东西。
叙事者可以将营造出一个幻觉，好像很多故事都处于一个大设定之中，就像 metafoundation 一样。世界的模态会随着叙述者的选择而改变，世界观内的人也将走入不同的设定之中。就如一个游戏里，主角选择重点关注邪教徒，然后主角若不阻止他们，世界就会被邪神毁灭了。但如果不关注，那么世界依然在运转，仿佛邪教徒只是患有精神病一样。
就像《小径分岔的花园》，不过此时小径的名字叫做“muthos”，岔路口叫做幻泡，花园叫做妄域而已。幻泡会随着叙事聚焦的变化而不断生灭，在闪光灯下留下一个个幻影。
最后再说一下，休谟指数，即现实扭曲程度。休谟是怀疑论者，因而在此就取他的名字作为以公认基准现实4为对象的叙事可靠性单位，或者说对公理系统的怀疑度。回到上文所举的科幻世界和DND世界的例子。魔法的设定如果出现在科幻世界，那么魔法本身就是科幻世界的“异常”，因其与世界公理相背。魔法越是反科学公理，那么这个魔法越是被怀疑根本不应该出现在这个科幻世界上。也就是说，在此时的休谟指数，实质上就是怀疑度，用以描述现实背离基准现实的程度。