Board Thread:General Discussion/@comment-17054615-20150516215111/@comment-1298029-20150517001506

You know what, unlike many of the theories floating around, this actually sounds plausible. I'm not saying I'm convinced, much less that I think it's a sure thing, but it's interesting. Most appealing is, it has actual facts behind it, not just speculation upon speculation. Granted, the facts are collected to fit the theory, but even so that's more than most of these proposals can say for themselves.

I guess I might as well recap the facts the video mentions:


 * Heimdall's armor in Thor 2 has a small round ornament inset on the chest. It's a dull amber colour and actually looks like amber (the material) or glass, possibly a polished crystal or metal. The theory purports this is an Infinity Gem.

HEIMDALL: Still shining. From here I can see nine realms and ten trillion souls." While at first glance most people would probably (at least I did) take the term souls in this use as figurative, meaning people'', there's nothing says it can't be interpreted literally, as in see souls or into them. The video thus takes this as indication that the gem in Heimdall's armor (assuming previous point is valid) is the Soul gem.
 * Heimdall has the ability to see into the souls of everyone in range of the Bifrost (at least). Quote from Thor 2: "''THOR [laughs]: Perhaps. How fare the stars?


 * Heimdall's eyes are a striking amber colour, unlike those of any other Asgardian seen, and the same colour as the gem in his chest. In Thor's second vision in AoU, Heimdall's eyes are white (of the no pupils kind). The video purports both facts are related to his ownership and presumed future loss of the Soul Gem. Furthermore, Heimdall's future eye change is taken to mean that the Soul gem will be revealed in Thor: Ragnarok, which is where the vision in AoU presumably flashes forward to.


 * The "H" in Heimdall's name would fit the T.H.A.N.O.S. initials theory (which I won't introduce because this forum is chock full of it already).

Now, as for my comments on it: As for the colour matching one of the currently vacant gem colours (green and orange, the latter of which could quite easily be interpreted as amber, too), I would just point out that Heimdall's entire colour scheme is an amber-gold. His armor, sword hilt, everything. It's just his thing. Then again, this might have been intentional, at least in Thor 2. As for his eyes, I always assumed they just matched his outfit plus made him look cool, but I have to admit that there were quite a lot of close-up shots of Heimdall's eyeballs, compared to the total screen time afforded the character. However, I have to point out that Heimdall's character design was created for Thor 1, at which point I rather doubt the whole Infinity War storyline had been developed, making it rather unlikely that Heimdall's eye and gear colour were purposely picked to suit a gem. Then again, they might have been, or the writers just caught a lucky break.

I don't really buy into the white-eyes thing from the vision very much, as I kinda think the white eyes were meant to convey that they're all zombies or something. Then again, out of all the "zombie" people in the vision, only Heimdall actually had white eyes, so maybe there's something to it as well.

I won't comment on the initials thing because it's completely gratuitous, but I'll contribute the mathematical odds that it's correct given the 4 gem names we now know can be reasonably made out as matching (Tesseract, Aether, Orb, Sceptre): Let "match is intended" be H1 (the hypothesis), and "match is coincidental" be the null [hypothesis] (H0). Under H1, obviously P(H1|H1)=1 (i.e., if H1 holds, there is no randomness in the naming). If instead the null holds, then the probability of four random letters of the English alphabet (which has 26 standard letters) matching a unique 6-digit key is P(H0|({A,O,S,T}e(S0))=(Cr(2,26)-1)/Cr(6,26), where e stands in for the element sign, S0 is the solution set, and Cr(x,y) denotes combination with repetition (multicombination) of length x from a set of y elements.

In words, the probability of the match being random given that we know that four letters match is the number of combinations (with repetition) of the remaining two free letters, less the one that is correct (assuming that being correct by coincidence would still count as correct, hence, the subtraction of this possibility), over the total number of possible combinations with repetition of 6 letters from a 26-letter alphabet. This comes out to P=((26::2)-1)/(26::6) = 0.000475 = 0.0475% = 4.75 parts in 10,000 =~ 1 in 2105. (The :: stands in for the multichoose operator.) So, infinitesimal, i.e., probabilistically speaking four letters out of 6 matching virtually guarantees that the match is not a coincidence.

''Note that this is ex-ante probability, whereas the idea to fit the letters to the particular word Thanos was chsoen ex post, thereby making the information content of this probability calculation rather limited (by not taking into account that different letters might have conceivably instead been fitted to some other word that would've presented itself). In other words, there is likely massive selection bias in this, making the "actual probability" that the names will spell Thanos much lower. Unfortunately, this kind of problem is next to impossible to quantify mathematically (actually, it'd be possible with a dictionary of candidate acronyms and reasonable rules for their selection based on known letter sets, but it'd still be bloody difficult to compute).''

''Even disregarding the previous note, the result I give above may be significantly underestimated, as not all letters have the same likelihood of being the initial of a random word, and all these letters are among the most common (the sum of their probabilies of occurrence as initials makes up 42% of total, despite the four letters only making up 15% of the alphabet by count). However, weighting probabilities of individual letters' occurrence by their occurrence as initials in English (which is available from Project Gutenberg, reprinted e.g. here) would require a computation beyond my ability to easily perform on a hand calculator, or in Excel for that matter.''

Even so, that would be rather far from a true representation of how likely a given letter is to be an initial of a gem "container" (only nouns would apply, and then only nouns that would "sound good" for that purpose, with possibly proper names in the mix which the frequency analyses do not track, with made up names being potentially infinite, or at least very numerous; however, the "good-sounding" nouns could be reasonably assumed to have initials distributed the same way as the set of all nouns, and the presence of proper names, especially made up ones, would actually probably tend to balance the initial frequencies towards equal probability distribution).