William Poundstone's lay book on game theory, Prisoner's Dilemma, describes John von Neumann's dollar auction.
Von Neumann offered to sell a dollar, starting the bidding at some low value (a penny?), but with the condition that he would collect both the highest and second highest bid.
Once bidders begin participating, they have an incentive to continue bidding, or risk "throwing good money after bad", well past the point that the bids exceed one dollar.
The dollar auction is a prototype of many other issues hard to resolve rationally, [many of which] Poundstone describes.
In a BarnRaising context, the example of the dollar auction's mechanics illuminates why some individuals continue to participate in a project long after the point at which a naive analysis indicates they should exercise their RightToLeave or, if available, their RightToFork.
The only way to win a dollar auction is to bid 99 cents as the first bid. The payoff is low, one cent, but it's higher than zero. To extend the analogy (perhaps too far), AtheOS? is an [example] of this. Kurt does 99% of the work, and he gets 100% of the credit.
(Alternately, collaboration between bidders can be excluded by the game's set up. In von Neumann's dollar auctions, presumably any collaboration amongst bidders would have been apparent and perhaps prevented by implicit adoption of traditional auction protocols which disallow such collaboration. )
Communities form around projects in spite of any dollar-auction-like qualities accompanying involvement with those projects. Given that only one participant can win a dollar auction (either the auctioneer or the 99-cent bidder), the possibility of getting caught in a dollar-auction situation is a disincentive to participation.