Look, let's say I run it a few hundred times... meaning 300 times. 1-2*(.8^300)=1-2*8.452712499*10^-30. If I ran it hundred times, 1-2*2.037045976*10^-10... both come out astronomically high in terms of probability. So does .998019196 or 99.8019196%. On top of this who ran those probabilities? *Oystein* wrote the article on 2004-09-28. In comment # 24 he says
You are reading wrong somehow.
None-exp civs at deity will only get gold, map, warriors and barbs. The same for exp. civ at sid except you will not get barbs.
He comes from Oslo, Norway. Consequently, that his note sounds a little different might make some sense if you keep that in mind. On top of this look at the Deity table for non-expansionist tribes and compare to the Sid table... the Deity table for non-expansionist tribe table reads:
gold-5%
maps-5%
warriors-5%
barbs-85%. Now take out the barbs and distribute them evenly over gold, maps, and warriors. In other words, say 5% means 5. and 85% means 5. 85/3 (3 for gold, maps, and warriors)=28 1/3. 5+28 1/3=33 1/3=33.3333. So, if we take the Deity non-expansionist table and take out the barbs and distribute them evenly over all other categories which do not equal 0, we get the Sid expansionist table. Now, that's what his note says. Similarly, look at the Demi-god table for non-expansionist tribes. We have
5 for tech, gold, map, settlers, and warriors. We have 75 for barbs. Evenly distribute those 75 units over tech, gold, map, settlers, and warriors (75/5+5=20) and you have 20 for all those categories. That's the Deity expansionist table. Do the same with the other ones. His note makes sense if you keep this in mind.
To clarify the note it *means* "Expanision civs get the probabilities of 1 level lower *for the base values*, with all barbarians evenly distributed over categories already greater than 0 (except in the warlord case)." Please note his comment holds *for the base values* and *not* for the expansion values. That I didn't get a settler or a tech from those Sid level huts merely confirms this.
By all means test it yourself and then provide evidence of the otherwise. I understand the details of what he meant and his numbers line up with his comment, so you go ahead and do the test. It shouldn't take you too long.