n-ice.org OpenTTD Forum
Administration => Suggestions => Topic started by: TechNoir on July 31, 2012, 11:37:30 am
-
After playing a long game and winning only 12 points I would like to suggest scoring points for other things than achieving a goal and achievements.
So points would be possible to get from such things (these are only suggestions after some brainstorming):
-most passengers transported
-most trees planted
-most coal transported
-highest profit on vehicle
-highest average income per vehicle
-most primary industries funded
-highest average rating per station
-most vehicles in the game
-crashed trains = -1 pt deducted
-caught by regional investigators after bribery = -1 pt deduced
-highest number of tracks
-biggest production of goods
Discuss ;)
-
Hey TechNoir
As an answer to your suggestions:
You can get even 5 points for a game...depends on what u did in that game. The score algorythm is secret but if you think about it, it`s pretty logic what u get points for and what u don`t get points for.Therefore, i`ll answer to your suggestions:
1. most passengers transported - (u already get points for the quantity of cargo that u transported so the more u transport, the more u get)
2. most trees planted - if players would know that we give points that, everybody would try to grab some points by planting
3. most coal transported- the same as passengers
4. highest profit on vehicle - you already get points for that via performance score
5. highest average income - the same as passenger number
6. Most primary industries funded - personally, i don`t agree with that since i consider that funding industries is just a replacement for real work (making transfer rail systems)
7. highest average rating per station - i agree with this proposal.....it would make players learn about a good rating for a station
8. crashed trains(points deducted) - yes, i agree with that too, that would make players more responsible
9. caught by regional investigators after bribery - i don`t agree because not being able to work near that town for a year is a hard enough punishment, it would be too much
10. highest number of tracks - if we would do that, players would start making thousands of tiles of tracks, and i think it`s already included in the scoring somewhere but no directly
11. biggest production of goods - the reward for that is a higher profit and finishing the server faster, therefore getting more points:)
I hope i was of help....personally, i think a discussion about that list of proposals would be useful:)
-
Wow. I didnt know that points depends on secret alghorythm ;o I am now very curious how it works. I guess the main thing is about speed. So it only promotes "speeders" ;o
Mainly i suggested this because sometimes you play a lot and you can get small number of points ;/
Your justification was good. Yea some people can think like that.
-
u should give alex 1 karma if the answer helped u :D
well, ofc there is an algorythm, we don't give points based on looks :P
-
well, I can point to where's some hints :P
-->> http://www.n-ice.org/openttd/highscorecalculation.php (http://www.n-ice.org/openttd/highscorecalculation.php)
-
i agree with alex in every point. and yes, 7. and 8. really sound like good ideas with some slight modifications i'd propose:
7. highest average rating per station :
that sounds like you want to give only the person with the highest average station rating a bonus? that's too punctual and only awards one single player, i'd say we rather make it more fuzzy and just say "the algorithm should take the average station rating of each player into account for calculation of his score". so the higher the avg. station rating of a company is the more points the player gets.
8. crashed trains(points deducted)
if you deduct 10 points from a player on server 7 it could mean you take all the points he gets for this game. do the same on server 5 and the player won't even notice the missing points.
so again let's not take hard values here. we'd do something with percentages, taking the game length into account and give it a weighting so it only affects the total resulting score by a certain degree (the exact values again will be secret :P)
a general note on changes to the score system:
such changes as above depend on certain data being stored in the first place. e.g. right now we don't record the number of crashes a player has to the database, that means i cannot just change the scores that already exist.
- in step 1 we have to add the feature that all necessary data is collected
- in step 2 we would reset all scores to 0 and archive the TOP 10 score table
- in step 3 we would change the score algorithm and now a new round starts
we've already done that once, so the process is not new. but we have to plan it, pick a reset date, announce it on the website early enough (maybe 2-4 weeks before) so players don't get a bad surprise...
for this reason i'd like to let this discussion go on more before we do such changes and collect more ideas so it's really worth the effort.
-
I think the scoring system is quite well balanced as it is now. It still might benefit from some *small* tweaks, like some of those proposed in the original post. And I absolutely don't like the idea of resetiing all scores and starting a new round. If the changes are small (and I vote for them being small and gradual if needed), resetting will not be necessary.
-
Maybe instead of adding new points we should make new statistics. Like monthy, weekly and yearly. I also want to suggest table with average points per game ;p that would be interesting to see
-
And I absolutely don't like the idea of resetiing all scores and starting a new round. If the changes are small (and I vote for them being small and gradual if needed), resetting will not be necessary.
in the beginning we did some changes to fine-tune the formula but it was always applied to existing scores too so they all were consistent to each other. i don't like the idea of gradual changes that don't affect previous games, it means you look at your personal scores and got 2 same games but you got different scores for them. will lead to a lot of confusion among players and probably complaints.
so what i'd rather do is apply the changes to all old game records too and recalculate all scores. but as you can imagine even a very small change will affect especially the TOP 10 players that had a lot of games very much, because it multiplies itself with every game played.
imagine 2 players that both won 100 games and are at almost the same rank. but one didn't care for vehicle efficiency and ran a lot of small vehicles while the other one used fewer big trains. now if i do only a very small modification to the weighting of the vehicle amount it might boost one player 10 ranks up and the other one 10 ranks down.
i can't think of any change that matters and brings in new ideas (like the station rating suggestion) on one hand but on the other hand wouldn't completely turn the current score board upside down (which will lead to complaints from our most faithful players.
this is why i'd always want to do it using a score reset.
also i think it's not that bad to start anew every X months (the current scores are already connected since 8 months). probably we take a lot more time to find good changes and then we'd do the next reset at the end of the year. what's so bad about restarting once a year? many players will be happy for a new chance. it's not motivating for new players to see how they get 10 points for a game and even the guy at rank 20 has 600.
and if a TOP 10 really is good he will get back to this stage on the new highscore board soon enough, why should they fear their standing?
-
And I absolutely don't like the idea of resetiing all scores and starting a new round. If the changes are small (and I vote for them being small and gradual if needed), resetting will not be necessary.
in the beginning we did some changes to fine-tune the formula but it was always applied to existing scores too so they all were consistent to each other. i don't like the idea of gradual changes that don't affect previous games, it means you look at your personal scores and got 2 same games but you got different scores for them. will lead to a lot of confusion among players and probably complaints.
so what i'd rather do is apply the changes to all old game records too and recalculate all scores. but as you can imagine even a very small change will affect especially the TOP 10 players that had a lot of games very much, because it multiplies itself with every game played.
imagine 2 players that both won 100 games and are at almost the same rank. but one didn't care for vehicle efficiency and ran a lot of small vehicles while the other one used fewer big trains. now if i do only a very small modification to the weighting of the vehicle amount it might boost one player 10 ranks up and the other one 10 ranks down.
i can't think of any change that matters and brings in new ideas (like the station rating suggestion) on one hand but on the other hand wouldn't completely turn the current score board upside down (which will lead to complaints from our most faithful players.
this is why i'd always want to do it using a score reset.
also i think it's not that bad to start anew every X months (the current scores are already connected since 8 months). probably we take a lot more time to find good changes and then we'd do the next reset at the end of the year. what's so bad about restarting once a year? many players will be happy for a new chance. it's not motivating for new players to see how they get 10 points for a game and even the guy at rank 20 has 600.
and if a TOP 10 really is good he will get back to this stage on the new highscore board soon enough, why should they fear their standing?
I don`t agree with the fact that 2 games having different scoring is a reason for reseting scores cause the same happens...even if you reset scores, they`re still our scores so 2 games have a different score anyway u take it. However , i do agree that a new chance is not bad from time to time, and starting clear and fresh is very important. It might make certain players show ambition. I`m sure everyone wants to reach 1st place and for some reseting scores is just a catalyzer. But we need to wait until more modifications come up before we can reset the scores.
-
I'm not sure how score calculation is implemented technically in your system, but I believe it will not be difficult to keep points for past games intact, while updated scoring rules would be applied only to newly played games. This approach will allow introducing changes without any complaints from players, because it does not affect current leaderboard standings.
-
The amount of points awarded for a Hard Game seems small and out of proportion compared to the Desert S#3 specially since it is harder, and even points awarded to non winners should have more weighting, since they can spend a lot of effort and hours.
Example: I got 10 points for coming 2nd in a hard game which took 4 hours. If I did 5 Quickie cames in that time I'd get more points whether I won or got 2 or 3 points a game.
-
The amount of points awarded for a Hard Game seems small and out of proportion compared to the Desert S#3 specially since it is harder, and even points awarded to non winners should have more weighting, since they can spend a lot of effort and hours.
Example: I got 10 points for coming 2nd in a hard game which took 4 hours. If I did 5 Quickie cames in that time I'd get more points whether I won or got 2 or 3 points a game.
imho that doesn't have to do anything with this thread.
such things we can always discuss and if it turns out to be true we can always raise the score multiplier of server 6, even for existing scores. but please make a post about that on the special board for server 6, i want to hear more examples from others too and also find out whether you can really do 5 quickie games in the same time ;)
-
I'm not sure how score calculation is implemented technically in your system, but I believe it will not be difficult to keep points for past games intact, while updated scoring rules would be applied only to newly played games. This approach will allow introducing changes without any complaints from players, because it does not affect current leaderboard standings.
if i'd win a game twice on the same server within 30 game years and almost same performance points and i get different points for those two games i would complain. and some time later it will be hard to remember when we did which changes so nobody can explain anymore how points at specific times had been calculated.
also you lose any way to track your improvements - right now you can compare and see that 3 months before you only made 50 points per game, but today you already make 70 points. that wouldn't be possible anymore if you change the score algorithm from today to tomorrow.
scores of games from yesterday would be different scores than those from today, so you cannot compare them anymore. it is some kind of reset but you'd hide that fact from players, confusing anyone who tries to compare them.
such a score system lives from the trust of players in it to some degree. if players get the feeling it is constantly adapted by measures they don't know (but might assume its to someone elses advantage and not theirs) he might lose trust into it and get frustrated.
a reset on the other side is a clean cut, it's official, obvious for anyone.
-
In complementation to the proposal of TechNoir.
I have another proposal, but with different orientation.
In the actual system, the only way of get point is wait to the goal reached. If you not are loged WHEN the goal is reached, you not get points.
I think in an aditional system that assign points to the user EVERY MONTH for stay conected and make things. I think in relationate this system with the performance score. In that form, if you not have performance, you not get points. If you have a performance below 100 or 200, you not get point. But the main orientation is... STAY CONNECTED AND MAKE THINGS. And not in AFK, because if you are in AFK, you are making nothing, and you not get points while you are in AFK. The points can be asigned to the user when they disconnect, or when the goal are reached.
With this system, i think that the servers are promoted, because the user get points for make things, and not for wait to the goal reached, and we can get more players, and more logins.
Wait for your comments.
-
Well, I must admit that performance under 100 or 200 or.... a certain value meaning that u don`t get points sounds very good. I would agree to that. About the points giving, first i suggest xor or knogle to give one opinion....would it be possible to make? and in that case....easy ...? hard to make?
-
The points giving can be another problem, becouse 1 point for each month are 12 point every year, and some servers not give 10 for points for goal.
I think in a low rate at month, from 0.1 to 0.5 in base of performance. If performance is from 100 to 200 give 0.1 each month, and a formula that increase 0.1 for each 100 of performace to a max top of 0.5 point for performance equal or greater 500. With this formula, the system can give a top of 6 points for year. And, of course, if this is excesive, can be lowered to a max of 0.25 or 3 points for year. The amount of points is a decision of the masters. This is only an idea.
-
the very basic idea sounds good, because even on a long run server players can get scores when they only want to play an hour or so.
in a next step we should be more clear about what "making things" would mean.
something that was also discussed between admins on IRC a very long time ago was some kind of XP system, like other games have it. so you get XP for certain things, you level up with certain XP amounts, maybe also make achievements part of that.
the hard part will be deciding what to give XP for. delta of CV to the CV you had last quarter? same for income? or for transported cargo amounts? giving it only for building wouldn't be a good idea, people could start to abuse it by spamming unused structures everywhere.
-
it seems more and more like for the CityBuilder servers we will need a separate scoring system. so alex and me just discussed the idea that we could use this chance of starting over to test such a system there: let the CB servers have gradual scoring (e.g. tied to the growth of your city between quarters). as CB scoring would start at 0 we have the chance to play around with new scoring ideas in this sandbox, so we should use it ;)
-
I'm always recomend use performance for determine what make a player. One un-used estructure is excluded in the performance score. Then, spamless structure are not computed. Making things that not work or work bad are equivalent to a poor performance score.
I'm think that performance score is the key.
But a lot of things are not computed for performance score, as you and other players say, that delta of CV, or income, or station rating, that are very interesant calculate and compute, but i not know how complex can be compute this things in a script for geting a score.
-
yeah i guess we will end up using data we can get in a mixture and then use the delta between quarters to calculate scores given per quarter. we can take many things into account: transported CV, perf., income, transported cargo, number of vehicles, number of stations...plus the ideas TechNoir had on his initial post, remove XP if you had vehicle crashes and so on...
and then every quarter you get a message like:
*** You received [14 XP] for your efforts in the last quarter, you now got a total of [1,337 XP]
-
What What What?:P Does that sound like the idea for a new scoring system?:P Does it? Does it?:)
-
the right side shows what data is technically available and could be used for scoring, so you get an idea what's available. the left side is some first brainstorming based on several ideas that were posted here:
https://dl.dropbox.com/u/21345942/OpenTTD_GS_data.png
tell me your thoughts on it.
-
I dont get why you would use performance rating as a measurement for skill and exp reward.
Simple:
- If i forgot to pay off my loan i get 100 performance less, whereas i dont meet up to the requirements of certain server (in the heat of the battle its easy to forget your loan, as it is completely irrelevant after +- 1-2 year of playing)
- If i want a high performance i have to transport all kind of lousy items with one train just to initiate the different cargo reward
- I cannot use short feeders as the chance is high that one will get <10k income (this is 100 performance missing)
- All the other meetings are usually met by the time you finish the goal in any server, thus irrelevant
Trust me performance rating is way overrated :)
Tbh the best thing to do is have your score be highly (read: almost completely) dependant on the time it takes you to reach the goal, as that rewards the best players with the highest points.
I believe your servers are goal games right, not make the most stations or deliver all kinds of cargo's. Otherwise you could just ignore the CV and focus purely on 999 perf rating, which might be interesting aswell (read: also annoying to reach ;). )
-
I believe your servers are goal games right, not make the most stations or deliver all kinds of cargo's.
actually i like the idea of goals that make features attractive that otherwise nobody would care about. e.g. making various cargo types being delivered a necessity forces players to play the game with all its aspects instead of just playing game number 100 of type "who is the fastest to connect all oil wells".
but that's for sure a question of taste and at least for the goal part they could even be different between servers, so we could satisfy different tastes. as long as the score system is somehow universal...
Otherwise you could just ignore the CV and focus purely on 999 perf rating, which might be interesting aswell (read: also annoying to reach ;). )
systems that depend purely on a single value can be easily cheated. we started with only CV considered and how long the game takes and people won the games with high scores by buying 300 train engines unused in the depot as soon as they had the money for it.
as perf is already a value consisting of several other things it is not as vulnerable as CV but it's still not enough to depend on it alone. also it has some disadvantages that you have mentioned which also would be compensated if other values are relevant too.
the more values you consider for the score
- the more flaws some values might have on their own compensate each other when used in a mix
- the lower is the chance of something being cheated
Tbh the best thing to do is have your score be highly (read: almost completely) dependant on the time it takes you to reach the goal
apart from the problems mentioned above this would also exclude the idea to give XP monthly. and if the goal is only CV i can play a normal game and short before the end i create a new company and send all my money over to that new company. i wait a bit more until i got enough to reach the CV goal and send this over too, making my new company win a huge goal within only a year = crazy amount of points.
also, "time it takes you to reach the goal" in relation to what? and how would that work on a short server where you only need 1-2 game years to win it?
but yes, currently it IS part of the calculation so it DOES work, but again only as one part in a bigger mix.
basically there is 10 good ideas how to do a score system and 9 of them eventually suffer from all sorts of vulnerabilities.
I dont get why you would use performance rating as a measurement for skill and exp reward.
it would be one aspect among many others so it wouldn't have a very high weight in the overall XP calculation (as everything else).
-
Ok you just said a few things that i dont understand, (i also dont know how to quote correctly but that aside :D ).
- This bug that you create x engines in a depot and shoot up your company value i heard of this before. Didnt know if that was true or not but thats a pain indeed. Isnt there a possibility to substract this in the CV-calculation formula? (something like: CV" = CV - (engines [complete trains] that have <=0 income this year * their value) ? Or is that to simply thought?
- The thing i mentioned about the different cargo types is also a fragile part because a few short trains with different types and i transported 8/8 types cargo.
- You must be joking that you have the option to send money to other companies on? ... really :/
- And if possible calculate the time taken on a monthly or quarterly base. Instead of only yearly
That said, im no coder or anything so i have no clue how easy/hard this stuff is
-
- This bug that you create x engines in a depot and shoot up your company value i heard of this before. Didnt know if that was true or not but thats a pain indeed. Isnt there a possibility to substract this in the CV-calculation formula? (something like: CV" = CV - (engines [complete trains] that have <=0 income this year * their value) ? Or is that to simply thought?
well in the mindmap linked above i tackled it with subtracting the total value of all stopped/crashed vehicles. but you are right, i have also added "total value of all vehicles without positive income is SUBTRACTED" now. should be technically possible but right now everything is still nothing but theory.
- The thing i mentioned about the different cargo types is also a fragile part because a few short trains with different types and i transported 8/8 types cargo.
yes, good point.
what would be also a way to circumvent the problems of perf. ranking while keeping its positive aspects would be to not use perf. but making the same criteria part of the score algorithm, just in a smarter way.
e.g. we give a reward for cargo diversity too (like +1 XP for every cargo type transported) but only cargo types that you transport at least 20% of the amount of your most used cargo type are counted. that means if you focus on coal and transport 100,000 tonnes of coal per quarter you must at least transport 20,000 litres of oil to have oil count as another cargo type you are seriously transporting.
this won't be 100% accurate as not all units of all cargo types work on the same scale and not all industries grow their production the same, but it should still do a much better job than only checking whether any single object of a specific cargo type was transported.
put that idea also into the mindmap.
- You must be joking that you have the option to send money to other companies on? ... really :/
players like to use it to boost each other and lend each other money, it often leads to a quite social gameplay i often wonder about. and as long as the score system isn't just depending on money or CV alone the advantages of money transfer aren't really big (except in the beginning, as i mentioned, initial boosting...). it has always been enabled and didn't cause any problems in the past.
circumventing weaknesses in the score algorithm by disabling game features is never an elegant solution, rather a last resort of there is no other way to mitigate a potential problematic effect. i still hear you that this poses a higher risk of getting abused than many other features and we should watch it carefully. but this we can still review again when the score idea is finished - then we'll see whether disabling it would be necessary or the benefit from it would be too small to have a considerable effect.
- And if possible calculate the time taken on a monthly or quarterly base. Instead of only yearly
unfortunately one technical limitation comes into play here: the CV value OpenTTD gives out to external applications over its interface is only updated quarterly, so that would be the lowest possible interval we can depend on (i know the OpenTTD client calculates it locally and real-time, but it can do this only because it has more data than our external application and has the exact same formula programmed into it as the server).
apart from that your suggestion is not really clear to me. calculate what against what? the time taken after a quarter...is a quarter. so i guess you mean see how much the CV has raised between quarters? and then give scores for it, more scores the more the CV increased within that quarter? or only check quarterly whether the goal was reached and at the end give higher score the lower the amount of years played is?
That said, im no coder or anything so i have no clue how easy/hard this stuff is
that's why on the right side in my picture i included every piece of information the game is giving me as a coder, but explained as normal text, so everyone knowing OpenTTD should be able to understand it. everything you see there can theoretically be taken, summed up, averaged, calculated against other values, measured on different scales, set in relation to other values, get multiplied by other values, get added or subtracted...whatever helps. which values you consider, how much weight you give each value and the way you let it influence the final score is more a mathematical and logical question than one for coders.
-
Looking good,
What i would consider rather than giving a fixed 1xp bonus for transporting multiple cargos with a base value of >20% is use a multiplier on the total xp. somewhere between 1,03 and 1,06 for each cargo would suffice imo. I think the 20% is a correct value aswell.
I do remember now that i once entered a game later where 2 friends apparently were playing, i went over their value rather quickly and in the end when i was @ 98% goal i got resetted and apparently lost, because one decided to suicide to give the other one exactly enough to win the game. ANYWAY, your servers (mostly?) work with a 300k starting loan which seriously is waay more than enough if you ask me. But thats only my opinion i know. (however the fact remains that its a really fragile part of your reward)
Maybe theres an option to limit the amount of cash giving. (If you top it at say 10-50k each time players wouldnt even be bothered to transfer millions).
What i mean about the quarterly calculation is not to calculate in between months, but to calculate the total time required to reach the goal in a value of quarters. Then compare them with a fixed value that would be reachable by the average of the players. (this is hard to determine). Example: If you have a server where you get 10 base XP if you win the game in 10 quarters (this is what the average [wo]man would do), you get <10 base XP when you finish later than 10 quarters and >10 XP if you finish earlier. This fixed value should also be the start of your calculation where-as all the other factors get multiplied with that achieved base value.
This one doesnt come alone because i also think starting year has a great influence on the game, as starting in 1995-2005 is way way easier than starting in 1985-1995, because of the easy transition from normal into monorail and the +25% train speed.
-
- The thing i mentioned about the different cargo types is also a fragile part because a few short trains with different types and i transported 8/8 types cargo.
yes, good point.
what would be also a way to circumvent the problems of perf. ranking while keeping its positive aspects would be to not use perf. but making the same criteria part of the score algorithm, just in a smarter way.
e.g. we give a reward for cargo diversity too (like +1 XP for every cargo type transported) but only cargo types that you transport at least 20% of the amount of your most used cargo type are counted. that means if you focus on coal and transport 100,000 tonnes of coal per quarter you must at least transport 20,000 litres of oil to have oil count as another cargo type you are seriously transporting.
this won't be 100% accurate as not all units of all cargo types work on the same scale and not all industries grow their production the same, but it should still do a much better job than only checking whether any single object of a specific cargo type was transported.
put that idea also into the mindmap.
I also like the idea of including the good points of pref. score into the score aglorithm, as the cargo thing has also bothered me. Further I also thnik the current cargo/quarter needs some adjustment. The reason is that you can build about 10 busses in different towns, and just let them drive arround. They will give you about 20k of cargo delivered which is an additional 200 pref points. A possible counter solution for that is to lower the weight of pax, as it seems unbalanced in comparison too other cargoes. I think it will also be hard to reach the 20% part you were talking about if you have a decent passenger network. Especially in longer goals where passengers grow like crazy (certainly compared too other cargoes).
-
Sorry for the doublepost but I just thought of a little possible flaw of delta CV:
Would I get points for every quarter it goes up? As shown in the graph below, one month it goes up, the second it goes down. Does that mean points for delta cv every second month in a case like this?
-
thanks for pointing this out, we should make it "delta to last maximum CV" then.
-
I would like to point out that I slightly mistyped in the previous post. That was the income graph. For income I agree that delta to last max is a good idea, because it can fluctuate quite a bit and sometimes conversion to mono or maxlev might even reduce it to zero for a quarter.
For CV I don't think this should be applied, but just the quarter to quarter delta. Ofc there is a number of stuff that can decrease CV like funding industries, building statues and so on, but I do not think they should be of influence of the points awarded for the next quarter.