
Inspired by last night’s #mathchat, I started exploring patterns. Specifically, I tried to create some patterns that could be “continued” in more than one way. For example, the sequence:
3, 5, 7, …
can have 9 as the next term (if we are listing odd numbers) or 11 (if we are listing odd prime numbers). Similarly, the sequence:
1, 11, 121, 1331, 14641 …
can have 161051 as the next term (if we are listing powers of 11) or 15101051 (if we are listing smashed together terms that form lines of Pascal’s Triangle).
But this only got me thinking: given any finite sequence of numbers, no matter how long, there are an infinite number of choices for the next number in the sequence, and all of them could fit into a plausible pattern! For example, you might think it’s obvious what the next number here is:
2, 4, 6, 8, 10, …
But I could just as easily claim the next number is 983, because the actual sequence I had in mind was:
2, 4, 6, 8, 10, 983, 985, 987, 989, 1001, 1974, 1976, 1978, 1980, 2953, …
Here’s a much more compelling question though, and it has close ties to some of the theory used in statistics and hypothesis testing. Consider this sequence again, but think about plausible choices for the next two terms:
2, 4, 6, 8, 10, …, …
You might believe, prior to all of this discussion, that the most likely choices are 12 and 14. If I reveal that the term right after 10 is not 12, then that completely undermines your belief. But what if I told you that it is 12?
2, 4, 6, 8, 10, 12, …
How much stronger is your belief now that 14 is the next number? Or put it another way, how does this new information affect your previously held belief?