Well we can set hard upper and lower bounds on the ages of people they fund (let's call them 18 and 80). And if we further assume that 18-22 is more common than 65-80, that gives us fair bit of information to start building a likely distribution.
(I'm actually pretty sure there is enough information about on how many companies YC is funded and the average team size to have a decent go at estimating the absolute numbers too, but that is less interesting for the purposes of this discussion)
Edit: (since you seem interested in the numbers) based on [1] and [2] we can guess lower bounds of (a) 550 companies funded, (b) 2 founders per company and taking my previous "(c)80% less than 40 estimate" (based on a normal distribution) that gives around 220 older than 40.
Given than we don't know (a),(b) or the standard deviation needed to accurately estimate (c) it is actually more useful to talk about the estimate (c) than to increase the inaccuracy by using the numbers (a) and (b).
Edit 2: [3] gives us more accurate numbers. 842 companies funded, max age of founder 66 (I guessed a max of 70). Using that to adjust the standard deviation, and sticking with the (b)=2 estimate above we get around 294 founders 40 or older. Call it 300 in round numbers.
To verify this, someone could run the numbers to check if this gives the average age (30.27) along with the median of 29 (although guessing the shape of the lower-side curve is probably challenging).
Given that the actual value is known we don't really need to calculate anything, just get those that know to tell us.
I suspect that the distribution is approximately normal and that your calculations are close to the true value. From my perspective the really interesting thing to know is how close the true numbers are to expected. This will tell us something useful about the selection process.
