So far I could identify the sequences up to the level 3 (from 0 to 3):
- Level 0: {1, 1, 1, 1, 1, ...} it's a boring sequence of 1s, nothing new (it has even an OEIS sequence: A000012)
- Level 1: {1, 2, 3, 4, 5, ...} the natural numbers (n ÷ 1), also nothing new (OEIS A000027)
- Level 2: {1, 5, 15, 34, 65, ...} a third-level polynomial, n × (n² + 1) ÷ 2 (A006003)
- Level 3: {1, 20, 210, 1309, 5720, ...} n × (n² + 1)((n² + 2)² - n²) ÷ 16 (A094311)
- Level 4: {1, 230, 26565, 1159774, 25170860, ...} doesn't exist on OEIS.
- Level 5 and above: also don't exist on OEIS.
There seems to be a pattern for the polynomials. There's a growing denominator. The denominators for levels 1, 2 and 3 (1, 2 and 16) suggest some candidates for what would be the next denominator:
- A374848 (1, 2, 16, 162, 3600, ...): it has some relationship to triangular numbers, as from its comments.
- A002416 (1, 2, 16, 512, 65536, ...): it has some relationship to the algorithmic exponential growth, so it makes sense to think in ramifications for the powers of 2.
- A016031 (1, 2, 16, 2048, 67108864, ...): paraphrasing OEIS, "De Bruijn's sequence: 2^(2^(n-1) - n): number of ways of arranging 2^n bits in circle so all 2^n consecutive strings of length n are distinct". Sounds like a candidate, too.
- A027871 (1, 2, 16, 416, 33280, ...): It's entitled as a product formula (the "multiplicative brother" of a math summation) of a exponentiation.