Have you heard of godel's incompleteness theorems?
One says no system can prove itself sound
EDIT:
X-post:
I wrote: ↑The decord (decreasing ordinal) sequence:
start with an ordinal c as term 1.
Rules:
Iff term n is 0: stop
Else:
Iff term n (call it a) is a limit ord, whose minimum "term" (eg. in w^w + w^4 + w2 = w^w + w^4 + w + w, the minimum term is w) is b, and for whom the rest of the terms of a are called r:
term n+1 = r+b[n] + b[n-1] + b[n-2] ... b[1],
Else:
Term n+1 = a-1
Demo:Code: Select all
w+1 w 3 2 1 0
Code: Select all
w^w+1 w^w w^2+w w^2+6 w^2+5 w^2+4 w^2+3 w^2+2 w^2+1 w^2 (10th term) w55 w54+66 ... w54 (78th term) w53+3081 ... w53 (3160th term) ...
Obviously this leads to a new function:
the term in the decord sequence for the ordinal a at which it becomes zero = dco(a) (decord)
dco(w+1) = 6, and dco(w^w+1) is ~tri^53(3160), give or take an iteration or two, where tri(n) = the nth triangular number. This suggests a reasonable bound of dco(w^w+1) << 3160^106 (though not in a googolplex-vs-g_64 kind of <<)
Since a[n] < a for all finite n, all decord sequences strictly decrease over time. Thus, dco(a) is finite for all a
Obviously, this leads to new fast growing function: dscrd(n) (yes, that's the word discord without vowels since the definition of discord is general chaos, and that's what this decord stuff is) = max{dco(ord(0),dco(ord(1),dco(ord(2),dco(ord(3),dco(ord(4),...dco(ord(n)}
dscrd(0) = 1
dscrd(1) = dco(1) = len{1,0} = 2
dscrd(2) = dco(w) = len{w,1,0} = 3
dscrd(3) = 3 (two decord sequences, the one starting with w and the one starting with 2, tie for this)
dscrd(4) = dco(w^w) (is max of sequence) = len{w^w,w,3,2,1,0} = 6
dscrd(5) = 6 (w+1's sequence ties with w^w's though)
dscrd(6) = 6 (another w+1, also second time dscrd(n) = n. its growth is pretty erratic.)
dscrd(7) = 6 (ord(7) is a pathetic 3)
dscrd(8) = 6 (we get w^2, whose sequence = {w^2,w,3,2,1,0}) This is getting boring, but don't worry about that-- it gets good for 9:
dscrd(9) = len({w^w+1,w^w,w^2+w,w^2+6,...}) -- hey, this is familiar-- = approx. tri^53(3160), where tri(n) = the nth triangular number. Where the crumbs did that come from?!?
dscrd(10) = dscrd(9) (dco(ord(10)) fails to exceed that of 9)
dscrd(11) = dscrd(9) (dco(ord(11)) fails to exceed that of 9)
dscrd(12) = dscrd(9) (exactly equal to dscrd(9) -- ord(12) = ord(9) = w^w+1)
dscrd(13) = dscrd(9) (dco(ord(13)) = dco(ord(11)) fails to exceed that of 9 or 12)
dscrd(14) = dscrd(9) (dco(ord(14)) = dco(ord(13)) = dco(ord(11)) fails to exceed that of 9 or 12)
dscrd(15) = dscrd(9) (dco(ord(15)) is a measly 5)
dscrd(16) = dscrd(9) (unfortunately, len({w^w^w, w^w, w^2+w...}) is exactly = len({w^w+1,w^w,w^2+w,w^2+6,...} since they converge immediately))
dscrd(17) = dscrd(9) (len({w^2+1,w^2,w3,w2+6,...}) is pretty much the largest thing so far that does not converge on len{w^w+1,w^w...}, but does not exceed it)
dscrd(18) = len({w^w+w, w^w+1, w^w,w^3+w^2+w, w^3+w^2+10...}), which exceeds dscrd(9) by an extreme amount.
dscrd(19) and dscrd(20) = dscrd(18) since dco(ord(19)) and dco(ord(20)) = dco(ord(18)) and hence don't exceed it.