Fair point, but I do believe that you and Mr. Messenger have distributed the prizes in my example in different manners, and both you and Mr. Messenger have stated (or at least implied) that Mr. Goichberg would agree with the prize distribution in each case.
I understand that the prize fund and restriction is not typical of CCA events; it was designed to stimulate discussion of the application of a particular rule. (And i must say it succeeded beyond my wildest hope, especially with Mr. Perks coming up with a distribution I did not anticipate.) On the other hand, I also firmly believe that not all of chessdom is CCA.
I also admit that I’m not sure how I would distribute the prizes in my own example. There is a definite appeal to Mr. Reed’s method, which essentially creates a new “virtual prize” out of the remainder of the prize to which the restriction has been applied. On the other hand, Mr. Doan’s argument (that the top three players expect to be awarded the sum of $200, $100, and $50, or $350 and should take as much of that as possible) is also compelling.
If I were to summarize the two approaches, I would say that Mr. Reed’s approach creates a new prize, while Mr. Doan’s approach (which Mr. Messenger favors) essentially awards the unrated player prize money from the least prize that is at least as much as the restricted amount.
To make that clearer, let’s suppose the prize fund includes a fourth place prize of $25, so the prizes are $$200-100-50-25. (And, to address Mr. Mulford’s concern, let’s take 401 rating points away from everybody and call it an “under 1800” section. 
) In this case, Mr. Reed’s method would award Alice $200, Bob $25, Carol $75, Dave $50, and the fourth place prize would be paid out to someone else (who scored 3.0 or less). If I understand Mr. Doan’s method, I believe it would award Alice $200, Bob $25, Carol $100, and Dave $50. Effectively, Bob would win the fourth place prize, Carol the second place prize, and Dave the third place prize. By introducing a new prize from the remainder of the second place prize, Mr. Reed’s method awards prizes to more players than Mr. Doan’s method.
When I posted the hypothetical, I originally had in mind Mr. Reed’s distribution as “the correct answer.” Now, I’m really not sure, although I do find Mr. Perks’s prize distribution troubling. One thing that troubles me about Mr. Perks’s distribution (and, to be honest, about Mr. Doan’s distribution as well) is that Carol finished in third place. If Bob were not unrated, this would be an utterly trivial prize distribution question, and Carol would win the $50 third place prize. I’m troubled that Carol wins more money for her performance based on who won the first and second place prizes. (I know this also happens with Mr. Reed’s distribution, but I am less troubled because Mr. Reed’s method creates a new “somewhere between second and third prize” from the remainder of the second place prize. I also realize I’m splitting semantic hairs.) Also quite troubling to me is that, in both Mr. Perks’s and Mr. Doan’s distributions is that Dave ends up with less than the fourth place prize of $50, even though Dave is clearly in fourth place with no tie. (Hm, with more thought, I’m beginning to see that as a compelling reason to believe Mr. Reed’s distribution is best.)
We had a prize distribution question several years ago that involved four players tied for two place prizes and two “under” prizes. Of the four, three players were eligible for the “under” prize; one was not. The two “under” prizes were considerably larger than the third and fourth place prizes, but not large enough that if the three eligible players divided the two “under” prizes they would receive more than just summing all the prizes involved and dividing evenly. However, this had a bizarre effect: By summing the prizes involved and dividing the sum evenly (rule 32B3), the player who was not eligible for the “under” prizes would win more than the third place prize (which is what that player would have received had there been no tie). It doesn’t make sense for a player to win more money in the presence of a tie than if there were no tie. As a result, rule 32B3 was amended in 2014 to include an explicit statement that no player may win more as the result of a tie than that player would win if the player were alone in that score group.
Now, I fully understand that this situation is different and the analogy is flawed. Here, we do not have a tie, but rather an artificial restriction on a particular player’s prize winnings. But I still have a feeling of disquiet caused by Carol winning more than the “natural prize” (to use Mr. Reed’s term). Maybe the clearest thing I can say is that I really don’t like limits on unrated players’ prize winnings, although I understand why organizers do this.
The rule governing restricted prize amounts was added while I was a delegate but before I was on the rules committee. I was also not on the rules committee when the motion from Mr. Messenger and Mr. Immitt was considered in 2011. I would certainly hope the rules committee considered various “what if” scenarios such as the one I presented when considering this rule, but I’m concerned it may not have. In any case, I can’t say that I’m a fan of having ambiguous prize distribution rules, and the fact that two NTDs and a very highly experienced ANTD come up with different prize distributions. If we’re going to have rules governing prize distribution, I would greatly prefer that the rules be deterministic in all cases. (Otherwise, what’s the point?)
