I never stated that a constant or near constant number of Bitcoins would be lost per year or per unit time. What I said was this:
M = total number of Bitcoins ever minted
C = total number of Bitcoins not lost
L = total number of Bitcoins truly lost
t = time
Per any unit of time you wish to use, L at t + 1 will be >= L at t.
Logically, then, L will get bigger over time. From that, it follows L will eventually approach M.
Let's assume that you agree that a certain percentage of unlost Bitcoins will be lost every year. Is that not an unreasonable assumption? No matter what number you choose, if you continue with the following calculation, the end result is the same:
C at t = M
C at t + 1 = (C at t) * x where x = some constant number less than 1.
Bzzzzt. You fail to factor in the value of a bitcoin. As the value increases, the
rate at which they are lost will decrease, as people will take additional measures to secure them. If R
L approaches 0, L will approach some value less than M.