I agree, the skinny/fat "blob" track does not follow the format AND seems to have a lot lower entropy, so it could be a text or an "AND" mask (given a lot of '111' occurrences). In the analysis above I ignore that track as it's not a solid/reliable data source.

Skinny/Fat Pros:

• Paint-style-wise, some flames seem to have inner color intentionally "painted over" to achieve "skinny" look
• 0/1 ratio seems to be 50%/50%

Skinny/Fat Cons:

• There are around 14 places that are not 100% readable, some users read those as skinny, some as fat - this doesnt happen w/colors or heights.
• They do not follow the format of other tracks
• They have lower entropy (less "random" data), they might be a  text track.

Moreover, the pattern as described in the analysis, where first 7 bits of 3 tracks build a pattern that later fully appears in one of the tracks, has only 0,03% chance of happening at random. Now, on top of that add the fact that following 6 bits in each track are the same AND they are actual 011010 ribbon pattern AND the fact that "connector" bits also are (0)(1)(1)(0)(1)(0), you get a chance of around 0,00004%. I don't want to bother calculating a chance of same thing but with 4th piece added (like in the "guesswork" steps).

We didn't "just look" at the bits, we carefully calculated chances of those things to happen, and they are crazily low, and yet, the pattern/logic is visible with naked eye without any computation at all.