Discarding a failed sequence is a courtesy to Evolution, in order to err on the side of probability. Otherwise, if the first character out of any of the data sources was in error (or as soon as the first error occured in any source), then the rest of the entire output string would be invalid.
Rather than discard an entire source of data, the math presumes to restart searching for the valid sequence at the first point after an error.
Further, the math in this article deals with sequencing. At no point does it presume that the data all "forms at once".
It does not matter if groups of coins are flipped together, or if one coin is flipped multiple times. What does matter is the sequence of the resulting output, and for that, the math in this thread calculates it perfectly.
I refuted this in post 373. Your suggestion would be a waste of time because unless the error was in the first link of the chain, adding a single link to the end will not matter at all. According to Jay L. Devore on page 92 of Probability and Statistics for Engineering and the Sciences, 4th ed. (1995, Wadsworth, Inc.),
What does matter is the sequence of the resulting output, and for that, the math in this thread calculates it perfectly.
You are correct on the second part. Your analogy fails on the first. Who has ever suggested discussing an unending chain of base pairs? That would be silly. The math is valid only for discrete trials. The author says so. I have show you where. And statistician will tell so so. Not that it really matters because
Period, the end. Until you recognize that I'm wasting my time.