If I understand correctly the score calculation description (http://www.n-ice.org/openttd/highscorecalculation.php), the more trains a company has, the less the score. This implies that it is better to have a small number of long trains. At the same time as Sassafrass noted here: http://www.n-ice.org/openttd/forum/index.php?topic=73.0, there are some valuable and viable strategies involving high volume of short trains. But unfortunately, for some reason (?) the scoring system doesn't favour them, and therefore we don't see such systems on n-ice servers often.
In connection with the new servers being launched, may I propose the following setup:
- long run goal (5...10 billion)
- big map (up to 1024 or 2048)
- short trains only (max trainlength=5...7)
- station spread=14...20
Large stations are intended to have many small platforms for bigger loads (the usual rule about unused station parts of course applies). This setup can be combined with FIRS too, since short trains are good for distributing supplies.
And by the way, what is the justification behind encouraging small number of long trains (with score calculation)? Speaking about assessing network efficiency, i'd name a number of wagons instead of a number of trains as a factor to consider.
we don't explicitely favour a small number of long trains. also none of the people who thought up the score system (including me) have long trains included in their usual play style. when i was more active i even never used trains longer than 5 tiles (or 10 engines/cars), just because i personally liked this length and i just don't like to play with long trains.
we had the problem that at some point players figured they can give their company value a big boost by building lots of
unused engines, using this cheat they not only won the game but also boosted their scores. so the algorithm was modified to at least give a score penalty for each train by giving more scores for "train efficiency". the idea is that if you don't actually use the train (because it was only built for cheating) you would fully take that penalty. but if the train is generating a lot of money this will give again scores (because CV and Perf raises) compensating for the penalty. now in OpenTTD a train can make more money the longer it is, this is not a rule we have made, so as a side-effect longer trains don't only make more money but also give more score and compensate for the penalty better.
so longer trains giving you an advantage is just something that comes from the game anyway and our train efficiency penalty happens to make you benefit with longer trains as a side-effect.
now that i think you will understand better why the system is like it is you are still right that simply considering wagons could lead to a better solution for the above mentioned problem. but when we thought up the algorithm it wasn't technically possible to see the number of engines and wagons, the total number of trains per company was everything we had.
with introduction of gamescripts to OpenTTD this has changed. i haven't checked on the technical details so far but i am quite sure it would be possible now, for example we could calculate an "engines to wagons" ratio and only give a penalty when this falls below a certain limit (e.g. a player has more engines than wagons).
but we collect such ideas for a future change of the score system
here so please suggest there. maybe you also have good ideas about what would be good ratios to still sanction cheaters while not penalizing "short train lovers" too much.
now that was for long-term improvements. but the score algorithm also uses a score multiplier that is different for every server, e.g. server 6 has the highest score multiplier as the landscape configuration makes getting CV and Perf harder and the scores on the server could never compete with those from other servers without that.
so your suggestion got my full support and when Kadar should create such a server i will make sure it has a high enough score multiplier so it can fully compete with long train servers about the scores you get.
however, when we got the server i'd ask you to help with doing some test games for us to find a balanced multiplier value (i don't want to overcompensate either...).
(Read 14792 times)
