The game is based on predicting the deaths of celebrities in the coming year. Each player comes up with a list of twenty five predictions, and earns points for those which come true during the course of the game.
A complete submission includes:
No player may actively make any prediction come true.
The phone number and address are for administrative purposes only. Only your name and email address will be published on the web page. (And they can be withheld from the web page if you like. I'll just substitute initials or something quirky like that.)
Pick 25 people whom you predict will die during the course of the year. List at least two alternates in case one of your 25 predictions comes true between the time you send your list and the start of the game.
I've heard some complaints about choosing bogus names like Barney, so this year we're establishing that all predictions have to be real people. I think picking Barney is funny, too, but you can get the same chuckle by picking individual Spice Girls.
Payment must be received by October 31, 1997. Because of the new scoring rules, there will be absolutely no flexibility on the deadline.
A specific amount of points will be allocated to every prediction, the same number for each. If a prediction comes true, that point allocation is evenly split among all players, rounded to the nearest whole number.
The purpose is to give more points for coming up with predictions that other players don't also think of. As someone pointed out when William S. Burroughs died, predicting the death of a famous 90-year-old junkie doesn't require you to be all that clever. Dividing scores amongst the players who made a particular prediction means a good list will probably still have some obvious choices, but the real points will be for the more obscure predictions.
In a simple example, say there are 10 players and every prediction is worth 100 points total. That means that the per-player score would be:
| Predicted By # Players | Each Player Receives | Or |
| 1 | 100 / 1 | 100 points |
| 2 | 100 / 2 | 50 points |
| 3 | 100 / 3 | 34 points |
| 4 | 100 / 4 | 25 points |
| 5 | 100 / 5 | 20 points |
| 6 | 100 / 6 | 17 points |
| 7 | 100 / 7 | 15 points |
| 8 | 100 / 8 | 13 points |
| 9 | 100 / 9 | 12 points |
| 10 | 100 / 10 | 10 points |
The per-person score kind of depends on the number of players. We could just make sure the predictions' base score always divides evenly among players, but as the number of players increase, the scores get ridiculous. For example, if we were to use this alternate scoring for the 1997 game (which has 13 people), every prediction would be worth 360360 points, the lowest number that would be divisible by any number of players from 1 to 13.
Of course, the base score only needs to be divisible to support the most popular predictions, but even if that's 10, the base score could still be 2520 points. So we'll normalize the scores to more reasonable values, keeping all of the relative values the same except for rounding. So for example, depending on what the numbers work out to, we may very well assign 100 points for each prediction and work from there.
In the case of a tie, the winners split the cash and the glory.
Nope. Our Angel of Life has tenure. Sorry.