First of all, as Rob Herman pointed out in the comments last week, a simple payoff based on point differential leads to conservative game-play. In other words, if you have a chance of getting 9 points easily, or risking a chance between 8 or 10 points, you will only risk to get 10 points if there is a disproportionate extra payoff for doing so.
Second of all, I didn't resolve the problem of when two experts play each other. If two people each play a great game, why should one be the "winner" and the other be the "loser"? Shouldn't they both be winners compared to another game where two newbies are playing each other? Conversely, shouldn't even a newbie who plays well be a winner? Don't we want games to reflect that, rather than having to implicitly understand it?
Third, I neglected the huge state change that still happens between winning and losing. In Puerto Rico, a score of 1 to 44 and a barrel is a loss. 2 to 44 and a barrel is a loss ... 44 to 44 and a barrel is a loss. But 45 to 44 and a barrel is a win, and now the 44 and a barrel is a loss. Is this sudden changeover really a good thing?
Risk vs Reward
I'll start with the first problem, the lack of risk taking.
Rewards should be commensurate with risk. Or rather, players should be able to evaluate the proportionality of reward for risk along each path available to them, so that they can choose paths accordingly. A low-risk taker may choose a lower reward, even a proportionally lower reward, for lower risk, while a high-risk taker may choose a higher reward for a higher risk.
Paths with high-risk and little reward should not be considered paths at all, unless they are disguised in such a way as to hide this fact, or the only path left in a desperate situation. And the same is true for the inverse. If a path too obviously leads to high rewards for little risk, there's not much point in considering anything else.
In the case of rewards commensurate with points, rather than all or nothing for the best score, the trick is to ensure that all players at all times have multiple meaningful options from which to choose.
When a player is ahead by 100 points in a game that scores around 120 points, the leading player has no important choices left to make; neither, in fact do the losers. By adding rewards based on points, all players still have reasons to continue making decisions.
To provide tension, and to prevent dull, conservative play, we want to make a greater risk for a greater reward just a little more enticing. That is why the answer has to be some sort of proportionality, where incremental advances yield greater rewards.
In other words, in a typical all-or-nothing game, advancing 1 or 10 points is about the same as far as reward goes. You beat player A but you haven't beat player B. It didn't matter if you beat player A by 1 or 10 points. Only 12 points matters, because then you beat player B. And then it doesn't matter again until you get to 25 points in order to beat player C.
A proportionally rewarding system would make a gain of 1 point yield payoff X, 2 points yield payoff 3x, 3 points 6x, and so on, making both conservative play and risk-taking both viable options at any point in the game. In fact, this is the ideal solution to the "runaway winner" problem in games.
All good players know that we don't play to win or lose, but just to, uh, beat my previous score, ... I mean, to, uh ... feel good about my own play.
We all know it. And yet we sound so defenseless when we say it, because the language of games treats this as outside the game experience. The games says "win" and "lose", not "nice try". Wouldn't it be good to put this into the game itself, so that we don't have to search around for it?
When we say "good play", what we mean is "good play for someone of your experience", or "good play for someone who encountered the challenges that you faced". In other words, rather than comparing ourselves to our opponents, or to the scoreboard, we are playing mental duplicate and comparing ourselves to a hypothetical average gamer of our level playing the same game that we are. If we perform better than that guy, we played well. If we perform worse, we played poorly.
The first way to address this is to create duplicate game experiences. Great databases of games can be stored on the Internet, tracking the experience of the players and the order of random events in games, and providing rough comparisons between games played in similar or identical contexts. Advanced A.I. could eventually evaluate performance based on certain positioning, such as how well a player does against a similar occurrence.
Players could download "Shuffle #232a" for Torres and compare their results against other players who played the same shuffle. Unless we assign a neutral party to order the deck before the game starts, this type of thing may only be possible for online games or once we achieve integrated electronic game tables.
A second way to address this is to create standard lists of hypothetical scores to beat for each game. The lists could be either absolute expected scores, or a percentage of score relative to other players. A look on the chart might cross check a novice player playing with two intermediate players and a guru for an expected value of X, or X relative to the guru. Any score close to X is average, less than X minus delta is poor, and greater than X plus delta is good; of course, the greater the granularity, the greater the incentive to score higher.
A third way to address this is to record specific instances of good or bad play. For instance, better moves gain you one or more tokens, and poor moves cost you one or more tokens. This would be the hardest idea to implement, because good and bad play is very hard to pin down in many circumstances. On whose judgment does it depend, yours or your opponents? Or a third party judge after you submit the game to gamejudges.org? How do you evaluate the quality of a good play? What about lucky plays - were they good? And so on.
Each of these methods depends heavily on the type of game, as some games will easily lend themselves to this type of methodology, while others will be harder to fit.
Winningness and Losingness
Now we come to the possible elimination of winning and losing altogether. Can it be done, or is this as bad an idea as eliminating dice from combat games (he asked, sarcastically)?
Let me quote you the rules from Puerto Rico:
Each player gets blah blah blah and you choose a role blah blah and in the Captain phase blah blah lets you store an additional blah blah blah wakki wakki wakki and the third way the game ends is if you need to fill the colonist ship and there are not enough colonists left to do so. The game ends at the end of the round that this happens, that is, after all players have taken a phase, and just before the Governor would pass to a new player.That was about twelve pages of rules. The rules don't actually stop there, however. They add two more sentences:
The winner of the game is the one with the most points. In case of a tie, the winner is the player with the most barrels plus doubloons among tied players.Now, what would happen to this game if we simply eliminated the last two sentences? Would the game be less challenging? Less fun to play?
Don't we know that the object of the game is to get the most points? Or, most points versus your opponents? I suppose we might have thought that the winner is the one who can produce the most tobacco at the end of the game, but we can specify: the goal of the game is to get the most points, and in any case more points than your opponents. What changes by eliminating winningness and losingness from the game?
You might have less need to end the game quickly before your opponents can achieve a higher score. But no, not if the goal is still to get the highest score relative to your opponents.
You might be tempted to turn the game into a cooperative game. Imagine Puerto Rico as a cooperative game, where players try to achieve the maximum overall score on all boards before the game ends. As in all cooperative games, you would have to limit or eliminate player discussion in order to prevent the same dominant players from running every game. But again, this violates the goal.
Isn't this what we do anyway, when we "don't really play to beat our opponents but to play against ourselves?" Why not make it official? Try rewriting the ending of your games and see what happens.
I admit that this may be a radical change for board game makers to introduce, so as an alternative for the more conservative among you, I will retreat to something a little less radical: partial winning and losing.
Imagine that instead of completely "winning" or "losing" you are assigned a percentage of winning and losing. For instance, beating your opponent by one point nets you a 50.1% win, while your opponent nets a 49.9% win. By two points is 50.3% to 49.7%. And so on. All we're really doing here is keeping the score percentages at the end of the game, instead of translating them into the binary notion of 100% win vs 100% loss that games prescribe.
Could it really be that simple?
Most games we have don't really need much changing to implement these ideas. They only need to reword the last paragraph in the game. Harder will be the education campaign for gamers and game designers, to convince them that competition can still have a point even if you don't absolutely crush your opponent at the end of it.