The current score formula doesn't reward using fewer vehicles to keep server load down. Of course that's a side effect but server load isn't an issue for us right now, we got more technical capacity than we got ideas for new servers. And in general there are more servers than players in the OpenTTD world anyway, no shortage of that.
Because usage of long or short trains is mostly a matter of preference or personal taste your next assumption might be that the reason for the current formula is that n-ice admins simply prefer long trains. While that might be true or not (I am not collecting statistics on that
) this is not the reason either. The most part of the calculation is coming from me and I definitely preferred short trains while I was still playing more actively and creating the initial formula.
The actual problem is that the calculations are still based on CV a lot for historical reasons: when we started with this there was no admin interface in OpenTTD. The only way to get values from a running game to do something with was the console. And that didn't show you anything useful about companies except Perf and CV.
At first this doesn't seem like much of a problem. CV is the obvious choice if you want to set a goal for a value that an automatic goal system could measure the progress on. And as a goal in terms of deciding when the game ends it is as good as everything else (e.g. end year reached). We're still depending on it for that and it doesn't work too bad.
But the concept of also taking CV for a (high)score system is flawed, because it can be manipulated by players. Players found out pretty fast that buying vehicles increases their CV, while at the same time any game will reach a point where money doesn't matter at all. Play long enough and you will always have enough money to max out on all vehicle limits, boosting your CV to heights you would otherwise never reach - and therefore this also applies to their score.
And even if the players didn't deliberately cheat the system: players with shorter trains would always have a higher CV than players with about the same network size but longer trains (because longer trains mean less engines that would increase the CV).
Fortunately from the console you could also see how many vehicles a company had, so even in the early days we were able to implement the obvious counter-measure: give a score penalty for having more vehicles. Now over time OpenTTD offered new ways to query more data from the game and our score calculation became more sophisticated, mitigating the effect of vehicles on the score a bit, but even today it's still strong enough that it could be used for cheating if there was no penalty for vehicles.
So any step in the direction of allowing players with short trains more points also means allowing cheating players more points.
That's why at some point it occurred to me that there won't be a way with the current system to improve much in this field. Also there are other issues with the system like the fact that (from score perspective) there is no motivation for players to play a partial game: If they play from the start but miss the end of the game (even when they reach the goal) they don't get anything. If they start late and finish a game they will probably not win it, hence also having a big disadvantage.
I must admit that I have been saying this for a long time and the system still is not here. But anyway: there is a whole new score system planned which will work entirely different. For starters, you will get scores every quarter so you can gain something even if you haven't got much time and join a 3 hour game for only the first hour. And you will get scores based on many many game parameters (and also score punishments, e.g. for vehicles with negative income, or crashes...) and can also get bonuses from certain events additionally to the quarterly points you get for progression in the current game.
The new scores are called "XP" (you might know the concept of experience points from other games) and are partially working like an ELO system, where you get more XP when playing against players that got more XP than you (while you get less XP for playing against "weaker" players).
You shouldn't forget that even the most sophisticated score system will never be able to act like a real jury and give you points for a "beautiful" or "aesthetic" or "well-thought" network. And even if it was technically possible: such things are a question of taste, just like the question whether you prefer long or short trains.
Instead you are stuck with taking the hard values that you can measure while at the same time these numbers can never tell the whole truth on their own, also specific numbers can always be cheated on. The sad truth I have learned is that it is really easy to find problems with a scoring system but infinitely hard to solve even only one without creating 3 new problems on the other side of the medal at the same time.
IMHO the best thing you can do (and that's the idea for the new score system) is to take as many game statistics into account as possible. That way you enforce a gameplay where players make use of many aspects of the game and by giving single values less weight remove the incentive to cheat on this value and at the same time more freedom (e.g. being able to leave the decision about longer or shorter trains to the player, because it wouldn't matter for the scoring).
To cut things short: I don't see any way how we could take the current system and solve the aforementioned problems by a monolithic change. The problem is the system itself.
Anyway, thank you for putting some thoughts into it and making this good (and correct) observation about one of the problems the current system has. As I have shown here creating a good score system is not an easy task and we certainly need as much input and ideas for it as we can get.
If you want to dive deeper into this topic I recommend reading the
additional points topic where most of the ideas for the new system originated from.
(Read 11363 times)
