Here's the quote:
[ QUOTE ]
The chances of making two pair or better if drawing three cards to a pair is about 44% or about (1-(39/47 * 38/46 * 37/45) = 1-(.8298 * .8261 * .8222) = 1-.5636 =43.64%) or odds of about 9-4. So by keeping that ace you are significantly hurting your chances to improve your hands.
[/ QUOTE ]
The calculation appears to be "how often will 8 outs hit" which doesn't make much sense to me.
I think if you dig through the archives you can find examples calculating the number exactly; Caro's number is correct.
Quick and dirty estimate #1: use twodimes.net calculator in 2-7 lowball mode. AAKQJ pat loses to 88/743 drawing three 29.7% of the time in lowball so the pair of eights wins 29.7% of the time playing for high. This is in line with the Caro number (not exact because we specified 5 more cards), obviously not anywhere near 44%.
http://twodimes.net/h/?z=2941094
Quick and dirty estimate #2: 47 unknown cards
On the first card you have 2 outs to improve.
On the second card you have 4 or 5 outs to improve.
On the third card you have 6, 7, or 8 outs to improve.
Assuming the best case you will brick 45/47 on the first card, 41/46 on the second card, and 37/45 on the third card. So your chances of improving must be less than 1-45/47*41/46*37/45 = 29.83%.
The worst case is bricking 45/47 on the first drawn card, 42/46 on the second card, and 39/45 on the third card, or no less than 1-45/47*42/46*39/45 = 24.34% to improve.