Re: [Freeciv-Dev] (PR#37592) Project directions
URL: http://bugs.freeciv.org/Ticket/Display.html?id=37592 Per I. Mathisen [EMAIL PROTECTED] wrote: URL: http://bugs.freeciv.org/Ticket/Display.html?id=37592 On Mon, 5 Mar 2007, (Eddie_Anderson) wrote: a) Reduce the government corruption penalty for Despotism and Anarchy to 10%. Both Despotism and Anarchy impose penalties at the tile level. For tiles like fish, that represents a 33% penalty of its food production. I don't understand why there is an additional 37% penalty (plus a distance penalty) on top of that. I agree that the recent corruption changes may have penalized the early governments a little bit too much. I've changed my mind about this one. Part of the beauty of the original proposal (IMO) was that it was simple (a one line change in a ruleset, IIRC). But I think that I have a better idea now. Please tell me what you think of it. Here it is: Get rid of the per tile penalties for Despotism and Anarchy. E.g. as it is now, if you put a citizen of a city on a Spice tile (while your government type is Despotism), then that tile produces 2 food, 0 shields, and 3 trade (2-0-3). Yet, under any other gov't type, that tile would produce 3-0-4. I propose making that tile's output 3-0-4 at all times (regardless of your gov't type). Here are some of the reasons for eliminating the per tile penalty: 1) It is a newbie trap. 2) It would remove the need for putting parenthetical values on the tile info displays. 3) It would concentrate the waste/corruption penalties in one place (the ruleset) which would make it easier to tune those penalties in the future. 4) It increases the benefits of choosing good city sites. If this is implemented, then there are some other things that may be worth doing too: 1) Restore Whales tiles back to 2-2-3 (Food-Shields-Trade). AIUI part of why a Whales tile (at 2-2-3) was disproportionately valuable was because most of its production flew under the radar of Despotism's per tile penalties. But with the per tile penalties removed, other special tiles would be more competitive with the Whales tile's original values. 2) Make Anarchy significantly more wasteful than Despotism. Either make Anarchy's waste greater or Despotism's less. IIRC, as it is now, there is little difference between them. IMO that seems counterintuitive. 3) Remove the free food and free shield for city centers. This is sort of a newbie trap in reverse. But it helps some tile types and not others - which means that what appears on tile info displays is sometimes not true. FWIW, I implemented this on my copy of Freeciv a while ago. AFAICS there were no ill effects on game balance. What do you think? -Eddie ___ Freeciv-dev mailing list Freeciv-dev@gna.org https://mail.gna.org/listinfo/freeciv-dev
Re: [Freeciv-Dev] (PR#37592) Project directions
URL: http://bugs.freeciv.org/Ticket/Display.html?id=37592 On Sunday 18 March 2007 02:02 pm, Peter Schaefer wrote: There always will be newbie traps and Im not sure whether it is a good idea to remove complexity from the game. After all, you want some learning curve. +1 Randy Kramer ___ Freeciv-dev mailing list Freeciv-dev@gna.org https://mail.gna.org/listinfo/freeciv-dev
Re: [Freeciv-Dev] (PR#37592) Project directions
URL: http://bugs.freeciv.org/Ticket/Display.html?id=37592 Per I. Mathisen [EMAIL PROTECTED] wrote: Good idea. Many times the designers of European board/card games do a good job of creating well balanced games. Don't forget the Australians! They produce some of the best board games there are. I wasn't aware of that. But I didn't intend to slight any particular continent. :-) I meant the word European in the sense that it is (was?) frequently used in rec.games.board. In a similar vein, I think all possibilities to rush at victory conditions should be high risk to opponent's counter-strategies. However - this depends on other players being able to see what you are doing, which is very hard in Freeciv at the moment. There needs to be a way to correct that. As a starting point, how about a scoreboard for each victory point? I'm thinking of reports similar to the greatest cities report, except that they would cover other things. There could be a report showing the top 5 contenders for longest road, another showing the top 5 contenders for largest palace, etc. I sympathize with this idea, but I suspect this would be a performance killer with the way city production is calculated currently. I expected that too, but I didn't see much difference. Granted, this was on small maps. OTOH, my code was pretty crude and it could probably be made more efficient by a better programmer than me. This is made very complicated by the way CM(A) calculates benefits. We will need to cache it somehow, or things will get rather ugly. Or things will get inaccurate. IIRC, it turned out to be simpler than I had anticipated. That's because it was a one for one replacement of the distance from the capital. PF distance was substituted for real map distance. IIRC, everything else (in the CM(A)) stayed the same. I would be interested in seeing how you programmed it, though. So please post a patch, even if it is not pretty ;) I'm will do that. I've been working on finding and assembling the pieces today. I should warn you that my patches were applied to trunk version 11656. Further, Benedict Adamson's AI patch (RT#14923?) was applied to that. Then my patches were applied on top of that. That said, I don't remember any overlap between my code and Adamson's. But the line number offsets might be substantially different in some files. -Eddie ___ Freeciv-dev mailing list Freeciv-dev@gna.org https://mail.gna.org/listinfo/freeciv-dev
Re: [Freeciv-Dev] (PR#37592) Project directions
URL: http://bugs.freeciv.org/Ticket/Display.html?id=37592 Per I. Mathisen [EMAIL PROTECTED] wrote: URL: http://bugs.freeciv.org/Ticket/Display.html?id=37592 On Mon, 5 Mar 2007, (Eddie_Anderson) wrote: Since this discussion last summer, I've been thinking some more Thanks for opening up this topic again. I will soon have more time to look at such ideas. You're welcome. It seemed like the right time. 1) Add a new victory condition to Freeciv that is based on points. ... I propose that a fourth victory condition be added to Freeciv: d) The first player to X points wins. ... One of the biggest advantages of this is that it makes Freeciv games scalable. E.g. if you want a short game, play to 500 points; if you want a longer game, play 2000 points; etc. I have been thinking along these lines myself, inspired a great deal by the way the board game Settlers of Catan manages the victory condition. Good idea. Many times the designers of European board/card games do a good job of creating well balanced games. However, I think the number of points should be very low, preferably 10 by default, to make the number seem more meaningful and close, and it should in most cases be possible to steal points from opponents to create counter-strategies (just like in Settlers of Catan). To quote from a design document I never made public (some parts may not apply): ... The following may give victory points (and other effects) in default ruleset: * Largest space station, 1 point. * Longest continuous road, 1 point. * Longest continuous wall, 1 point. Only need to pay half as much to barbarians to appease them. * Each defeated enemy civilization, 1 point. * Defeated barbarians, 1 point each time. * Largest palace, 1 point. There are several versions of the Palace building, in different sizes, just like Settlers of Catan. * First circumnavigation, 1 point. * First to philosophy, 1 point. * First to cure for cancer, 1 point. * Possession of Eiffel Tower, 1 point. * Greatest Expedition, 1 point. Magellan's Darwin's Apollo Project. An early victory with five victory points is possible by getting for example longest road, defeating one civilization, getting first to philosophy, producing Magellan's and building the largest palace. This gives a natural way of doing shorter games, without having an entirely artificial end year and the artificial and very hard to sabotage game scores as a way of appointing a winner. That sounds good. It offers players the opportunity to earn points in a variety of ways (so that everybody does not have to pursue the same strategy in every game). I assume that that means that the points for longest road, largest palace, etc. could change hands many times during the course of a game. Is that correct? If so, then a player's Worker/Settler/Engineer might be able to gain the last needed victory point by completing a road on a tile that links two existing roads together. Likewise, a player could conceivably push himself over the top by rush buying a bigger palace. Is that correct? Another potential advantage is that the AI can be programmed to focus on accumulating points rather than trying to engineer a conquest. It would certainly give the AI a shorter future horizon to focus on, which could be a good thing, and war in the game would become more a means to various ends (primarily to sabotage victory points) rather than all-out conquest, although the latter would still be possible. Exactly. The conquer everybody option is still in the rules, but this gives the AI some more practical, short term goals to pursue instead. d) Eliminate unlimited movement on railroads. I would be in favour, but it is not something I would consider important. I agree; it's not critical. Plus, (I assume that) teaching the AI to use unlimited movement wisely is difficult. Not at all. I figured that it was the difference between choosing from among ~30 potential destination tiles and choosing from among 100, 150, or more (for each unit that is on a long railroad). Then multiply that by the number of turns and by the number of units that enter (or remain on) that railroad during those turns. e) Normalize the effect of most happiness buildings and effects (all except martial law?). E.g. If a temple can turn an unhappy citizen into a content one, then maybe that temple should also be able to turn a content citizen into a happy one. The way it works now is somewhat counter-intuitive. If you did that, you could just as well remove all the unhappy/happy effects, and replace them with buildings that produced luxuries. The non-transitiveness (if I can call it that) of happy-buildings is the only thing that separates them from luxury production. (Well, that, and more esoterically, at what step in
Re: [Freeciv-Dev] (PR#37592) Project directions
URL: http://bugs.freeciv.org/Ticket/Display.html?id=37592 On Wed, 7 Mar 2007, (Eddie_Anderson) wrote: I have been thinking along these lines myself, inspired a great deal by the way the board game Settlers of Catan manages the victory condition. Good idea. Many times the designers of European board/card games do a good job of creating well balanced games. Don't forget the Australians! They produce some of the best board games there are. This gives a natural way of doing shorter games, without having an entirely artificial end year and the artificial and very hard to sabotage game scores as a way of appointing a winner. That sounds good. It offers players the opportunity to earn points in a variety of ways (so that everybody does not have to pursue the same strategy in every game). I assume that that means that the points for longest road, largest palace, etc. could change hands many times during the course of a game. Is that correct? Of course, that is the point. If so, then a player's Worker/Settler/Engineer might be able to gain the last needed victory point by completing a road on a tile that links two existing roads together. Likewise, a player could conceivably push himself over the top by rush buying a bigger palace. Is that correct? Yes. This is a viable strategy in Settlers, but a high risk one, since it is easy for another player to sabotage it so effectively that you will not (perhaps ever) be able to complete the action. For example, if you build two road pieces that you intend to put together to produce the longest road, another player can block it by putting his own road or soldier in the middle. Or you could save up lots of resources, but if someone rolls the robber, you might lose most of them. In a similar vein, I think all possibilities to rush at victory conditions should be high risk to opponent's counter-strategies. However - this depends on other players being able to see what you are doing, which is very hard in Freeciv at the moment. There needs to be a way to correct that. I sympathize with this idea, but I suspect this would be a performance killer with the way city production is calculated currently. I expected that too, but I didn't see much difference. Granted, this was on small maps. OTOH, my code was pretty crude and it could probably be made more efficient by a better programmer than me. This is made very complicated by the way CM(A) calculates benefits. We will need to cache it somehow, or things will get rather ugly. Or things will get inaccurate. I would be interested in seeing how you programmed it, though. So please post a patch, even if it is not pretty ;) - Per ___ Freeciv-dev mailing list Freeciv-dev@gna.org https://mail.gna.org/listinfo/freeciv-dev
Re: [Freeciv-Dev] (PR#37592) Project directions
URL: http://bugs.freeciv.org/Ticket/Display.html?id=37592 On 3/5/07, (Eddie_Anderson) [EMAIL PROTECTED] wrote: URL: http://bugs.freeciv.org/Ticket/Display.html?id=37592 Since this discussion last summer, I've been thinking some more about what Freeciv could be. I've reached a few tentative conclusions. Some of the changes are simple; some are far-reaching. I've already done some work (on my own copy of Freeciv) to implement some of these ideas, but it has been several months since I last worked on it. So I figured that I would publish these ideas to see if anyone else was interested in them. ... Here are the ideas: 1) Add a new victory condition to Freeciv that is based on points. ... I propose that a fourth victory condition be added to Freeciv: d) The first player to X points wins. One of the biggest advantages of this is that it makes Freeciv games scalable. E.g. if you want a short game, play to 500 points; if you want a longer game, play 2000 points; etc. It is an interesting idea, but using a lower number would make many games end before any contact is made, so the point limit would have to be around the equivalent of two islands with fully built up cities plus 80 technologies. Another potential advantage is that the AI can be programmed to focus on accumulating points rather than trying to engineer a conquest. Presumably the AI can be programmed to compare numbers more effectively than it can be programmed to build and move units. This won't help. AI is not that simple. 2) Award points for things besides WoWs and techs. .. a) Award points for firsts like 50 points for the first player to get two size 3 cities. There could be multiple tiers. (E.g. size 3 cites, size 5 cities, size 8 cities, etc.) There could also be multiple bonuses at each tier. E.g. There could be two bonuses for size 3: .a bonus for the first civ with *one* size 3 city and .a separate bonus for the first civ with two size 3 cities - and the civ that wins one bonus on that tier is ineligible to win the other. This sounds like rules for a boardgame, not a computer game. You already receive points for your population. Players already hate the discrete advantages gained by players who complete wonders, they would surely hate this too. b) Award 1 point on every turn for each surplus happy citizen c) Award points on every turn to the player who has the largest city in the game. d) others a) Reduce the government corruption penalty for Despotism and Anarchy to 10%. Both Despotism and Anarchy impose penalties at the tile level. For tiles like fish, that represents a 33% penalty of its food production. I don't understand why there is an additional 37% penalty (plus a distance penalty) on top of that. Despotism is not meant to be a preferred government, just research the Republic. Despotism reduces upkeep for units, so it also offers an advantage over the republic in exchange. b) Revise the terrain improvement costs. ... c) Eliminate Rapture growth .. You can use the server settings rapturedelay to change that. This has been discussed before. Rapture Growth makes Granaries and Pyramids almost useless. Plus, (I assume that) teaching the AI to do Rapture Growth efficiently (with many cities) is difficult. d) Eliminate unlimited movement on railroads. ... e) Normalize the effect of most happiness buildings and effects (all except martial law?). ... This takes the edge off the Hanging Gardens and makes that wonder even less usefull in comparision. f) For each civ, only its top 3 cities count. ... This would make it useless for the player to conquer opponents. Players already prefer sitting on cozy islands on their own, this rule would make it worse. g) Use PF distances for calculating the distance-based corruption and waste penalties. I've experimented with this and it works. The code was copied from Per's Wonder City code and I don't understand all of it. But when you build roads, it *does* reduce the losses due to distance from the capital. Very interesting idea, but how does it affect server load if you have 20+ cities on one continent or 100+ cities in the game? h) Eliminate ZOC effects caused by units with 0 offense (e.g. Explorers, Diplomats, Caravans). ... I fully agree to this. i) Eliminate trade routes established by caravan. Caravans will still be used for building WoWs. But the entire trade route system (3 routes per city, caravan has to find a path to the target city, then both cities get a permanent benefit, etc.) should be scrapped, IMO. IMHO,
Re: [Freeciv-Dev] (PR#37592) Project directions
URL: http://bugs.freeciv.org/Ticket/Display.html?id=37592 On Mon, 5 Mar 2007, (Eddie_Anderson) wrote: Since this discussion last summer, I've been thinking some more about what Freeciv could be. I've reached a few tentative conclusions. Some of the changes are simple; some are far-reaching. Thanks for opening up this topic again. I will soon have more time to look at such ideas. 1) Add a new victory condition to Freeciv that is based on points. ... I propose that a fourth victory condition be added to Freeciv: d) The first player to X points wins. ... One of the biggest advantages of this is that it makes Freeciv games scalable. E.g. if you want a short game, play to 500 points; if you want a longer game, play 2000 points; etc. I have been thinking along these lines myself, inspired a great deal by the way the board game Settlers of Catan manages the victory condition. However, I think the number of points should be very low, preferably 10 by default, to make the number seem more meaningful and close, and it should in most cases be possible to steal points from opponents to create counter-strategies (just like in Settlers of Catan). To quote from a design document I never made public (some parts may not apply): When you start the game, the world of mankind is in its infancy. We have already spread across the whole planet, but only beginning to reap the benefits of true civilization. The first cities take form, and with the crowding of people, ideas about how we should live take on a new urgency. Can your chosen ideas for the best way of life stand the test of time? The winner of the game is the civilization whose ideas have become irresistible through the overwhelming success of the civilization that espouses them, be it through peaceful or military means. The game offers to end when one player has a given number of victory points, agreed in advance. The potentially victorious player can choose to end the game at this point, or at any other point that he/she/it still satisfies this condition, or until all other players are vanquished. The following may give victory points (and other effects) in default ruleset: * Largest space station, 1 point. * Longest continuous road, 1 point. * Longest continuous wall, 1 point. Only need to pay half as much to barbarians to appease them. * Each defeated enemy civilization, 1 point. * Defeated barbarians, 1 point each time. * Largest palace, 1 point. There are several versions of the Palace building, in different sizes, just like Settlers of Catan. * First circumnavigation, 1 point. * First to philosophy, 1 point. * First to cure for cancer, 1 point. * Possession of Eiffel Tower, 1 point. * Greatest Expedition, 1 point. Magellan's Darwin's Apollo Project. An early victory with five victory points is possible by getting for example longest road, defeating one civilization, getting first to philosophy, producing Magellan's and building the largest palace. This gives a natural way of doing shorter games, without having an entirely artificial end year and the artificial and very hard to sabotage game scores as a way of appointing a winner. Another potential advantage is that the AI can be programmed to focus on accumulating points rather than trying to engineer a conquest. It would certainly give the AI a shorter future horizon to focus on, which could be a good thing, and war in the game would become more a means to various ends (primarily to sabotage victory points) rather than all-out conquest, although the latter would still be possible. a) Reduce the government corruption penalty for Despotism and Anarchy to 10%. Both Despotism and Anarchy impose penalties at the tile level. For tiles like fish, that represents a 33% penalty of its food production. I don't understand why there is an additional 37% penalty (plus a distance penalty) on top of that. I agree that the recent corruption changes may have penalized the early governments a little bit too much. c) Eliminate Rapture growth I agree completely. d) Eliminate unlimited movement on railroads. I would be in favour, but it is not something I would consider important. Plus, (I assume that) teaching the AI to use unlimited movement wisely is difficult. Not at all. e) Normalize the effect of most happiness buildings and effects (all except martial law?). E.g. If a temple can turn an unhappy citizen into a content one, then maybe that temple should also be able to turn a content citizen into a happy one. The way it works now is somewhat counter-intuitive. If you did that, you could just as well remove all the unhappy/happy effects, and replace them with buildings that produced luxuries. The non-transitiveness (if I can call it that) of happy-buildings is the only thing
Re: [Freeciv-Dev] (PR#37592) Project directions
URL: http://bugs.freeciv.org/Ticket/Display.html?id=37592 Peter Schaefer [EMAIL PROTECTED] wrote: URL: http://bugs.freeciv.org/Ticket/Display.html?id=37592 On 3/5/07, (Eddie_Anderson) [EMAIL PROTECTED] wrote: Freeciv games scalable. E.g. if you want a short game, play to 500 points; if you want a longer game, play 2000 points; etc. It is an interesting idea, but using a lower number would make many games end before any contact is made, If that is a problem for players, then they could choose a larger total. The implementation could be similar to endyear - the players specify whatever value that they want at the beginning of the game. Personally, I think a no contact game might be interesting. It would be purely a race to see who could build enough WoWs, etc. to reach the target total. so the point limit would have to be around the equivalent of two islands with fully built up cities plus 80 technologies. I don't remember the current point values for techs (5?) and WoWs (20?) so I can't comment on your calculation. As I said elsewhere, the point values for things (as well as the victory threshhold) can easily be tweaked to balance the game to one's taste. That said, I think that the per turn points awards will add up quickly enough so that reaching the lower threshholds won't be a problem. Consider this: With an average of only two happy citizens you would accumulate 200 points by 1AD. Add some techs, a WoW, and maybe a few tier bonuses, and getting to 500 points won't take much longer. 2) Award points for things besides WoWs and techs. This sounds like rules for a boardgame, not a computer game. You already receive points for your population. Players already hate the discrete advantages gained by players who complete wonders, they would surely hate this too. The purpose is to provide alternative paths to a goal. Going for per turn points (happy citizens, largest city) is the slow and steady approach. Buildings WoWs or going for tier bonuses is the big score approach. The player can choose whichever approach suits him/her. But if the game is balanced right, then the characteristics of your cities will determine which approach is the best fit for a given game. a) Reduce the government corruption penalty for Despotism and Anarchy to 10%. Both Despotism and Anarchy impose penalties at the tile level. For tiles like fish, that represents a 33% penalty of its food production. I don't understand why there is an additional 37% penalty (plus a distance penalty) on top of that. Despotism is not meant to be a preferred government, just Agreed. And there are severe penalties ( 50% for shields and trade in many cases) for operating under it. research the Republic. Despotism reduces upkeep for units, so it also offers an advantage over the republic in exchange. IME Despotism's advantages are preferable to Republic's advantages only in a very limited set of circumstances. Are you saying that changing one's type of government should be the *only* way to escape the penalties of Despotism? If so, then we'll have to disagree. My problem with relying on the changing government type path is that it depends exclusively on research. Consequently, in the early stages of the game, everyone is doing the same thing (maximizing science output) to get to the same goal (Republic). Why not just start the game with everyone having Republic as their government type? Do you see what I mean? c) Eliminate Rapture growth You can use the server settings rapturedelay to change that. Indeed. I already do. But I've noticed that the AI civs seem to spend a lot on luxuries. So I wonder if the AI is still trying to induce rapture growth despite the rapturedelay being set to 99. e) Normalize the effect of most happiness buildings and effects (all except martial law?). ... This takes the edge off the Hanging Gardens and makes that wonder even less usefull in comparision. OK. How about if we adjust their build costs to make their costs more consistently proportional to their benefits? f) For each civ, only its top 3 cities count. ... This would make it useless for the player to conquer opponents. Do you mean conquering cities or entire civs? I can see how conquering an opponent's smaller cities would be pointless. But taking a city away from another civ that was a big points earner for him could help your cause a lot. Players already prefer sitting on cozy islands on their own, this rule would make it worse. Isn't this a starting position/mapgen problem? (i.e. some players get an island to themselves while others have to share/fight over one.) Or do you mean something else? g) Use PF distances for calculating the distance-based corruption and waste penalties. I've experimented with this and it works. The code