Two more live playtests, at the Los Altos Games Day! The game was well received, and I was quite pleased with the results.
The changes I made after last week's session worked well. Playtime came in at about two hours, and I suspect would be a bit shorter with experienced players. Players successfully built transcontinental networks and made coast-to-coast deliveries by the end of the game. A few switchyards were built in each game, but never all six. That all seemed just right.
On the down side, my long-standing worries that the scoring system is badly balanced were borne out. The reward for connecting all six cities is so high that a player who fails to do so is almost guaranteed to lose. This would be okay except that it's usually clear which players will fail by around halfway through the game. It's no fun being the goat in a game in which there is almost always one designated goat.
The way to fix that is probably to raise the VP reward for building switchyards. The trailing player has an advantage here, in a way: if he is flexible enough to give up on the six-city goal early enough, he will save several actions and a fair amount of cash. He can then devote those resources to switchyard building, and remain competitive with the six-city players.
I have also recently revised the simulator to play by the new rules, and run another couple hundred thousand simulations or so. These were, as before, mostly explorations of balance. It has become quite clear that an intelligent first player has a huge advantage over his opponents, because the choice of starting locations is not even remotely balanced. This isn't unusual for rail games. The simulations show that it can be balanced by giving the players differing amounts of starting cash. In the game rules, I expect I will express this in two ways: a "standard game" in which the starting cash is simply dictated by the rules, and an "advanced game" in which the players hold an initial auction for turn order. Players will use the standard rules until they feel qualified to judge fine differences in starting positions, and can then advance to the auction rules. (Unlike Age of Steam and Railroad Tycoon, I don't think this game needs a turn-order auction every round. One at the start of the game should be sufficient, and the turn order need not change after that.)
Finally, I came away with a clear understanding about the current inadequacies of the board graphics and the player aids. This is not part of game design (since I have no plans to self-publish), but a well-made prototype really helps newbies concentrate on the game instead of on decoding the board and remembering the rules. I have some ideas for improvements, and I will be playing around in Photoshop to try to turn those ideas into clear graphics.