Discussion in 'Civ4 - Hall of Fame Discussion' started by Denniz, Feb 6, 2008.
I like it, but people should always get more than 0.
Ya, have it go from 100 down to 5 (or 1 at least)
Hey, I probably don't even have much clue about what Qscore means But here is my 2
Do the following to balance out all entries and your problems will be resolved:
Say we are looking at Deity Qscores for EQM:
1) Look at how many entries are made by each leader.
2) The more entries made with a specific leader, the more it singals a potentially abusive early UU or potentially abusive advantage at Deity difficulty.
3) Distribute a weighting coefficient to all leaders depending on how many entries are made by that specific leader... The more entries made by a specific leader, the lower the weighting coefficient.
4) If there is only one entry made by a specific leader, then this particular leaders gets the highest coefficient. I don't know exact number of leaders in this game, but say Huayna Capac has most entries. Then if a brand new deity entry by a new leader is submitted, compared to the many Huayna Capac entries, his Qscore should be adjusted such that it is still better than the best Huayna entry Yea you read it right, better than the best Huayna entry, because this system will encourage new entries by new leaders...
5) Recalculate Qscores for all previous HOF submissions etc and rerank...
6) problem solved, but many will not be happy that their Huayna victories do not score as high They must try the new leaders or the underdog leaders to get higher Qscores. Once enough submittions are received for the underdog leaders, then their weighting will also decrease and Qscores will be balanced in the long term.
@Denniz - I think the revised scoring system is great! It proportionally fixes tables with only a few entries and also those with a lot of entries by only a few players. And it still appears to give reasonable scores for well-played games with non-ideal map/leader choices that finish later than the #1 submission. The only thing it won't do is give a score to a Future conq/dom if there's a decent ancient game in the table, but I suppose we can simply use Future Space for the two RoA entries.
EDIT - Just noticed that removing the modifiers has a huge effect on the Gauntlet scores in ALL diff level tables since currently any level Gauntlet can be used for the 2 entries. Might need a tweak here.
So far I like what I have seen (I prefer the old method but let me get used to this method a bit more before I make a final call).
Will these new scores be used for EQM? Because if it is then Inca should not be allowed as it is too hard to compete against (and if it's not allowed in the EQM then we shouldn't have to compete against it).
The further down we stretch the range the bigger the gap between each turn. I was trying to get the best distribution of points for both small samples and large. I tired it alot of different ways. The turn numbers are on the query page. If anyone can suggest refinements to the formula...
Are you suggesting a minimum of 1 point for any entry with 1 or less?
I was thinking of removing Ancient from RoA and isolating those non-ancient starts for scoring purposes. Either that or drop the RoA as a few have suggested. I would rather have a consistant scoring system than RoA.
I still have to implement the EQM restrictions and difficult levels for Gauntlets in the EQM.
Yes, they are just for the EQM. I plan to apply the EQM restrictions to determine the best date as well. I was trying to get a feel for the numbers before I spent too much time refining it. As it was tried a couple dozen variations on the formula before I found something that looked half decent.
The downside of using score penalties is that some people will be happy with being on the list. Score would not be an issue for them. I am fairly convinced that the only way to promote balance is to make it a requirement to get on the list in the first place. (Not that I have been able to figure out a way to do it yet... )
The 25% rule is nice. I've looked at some of the results and they look pretty fair. If you want to make the EQM truly elite, I would apply that formula and say any game that regsisters a 0 doesn't count for EQM purposes. So to be EQM, all of your entries have to be within 25% of the turn count of the best score. I wouldn't apply this to QM though as it's a little too harsh for everyone...
The downside of the 25% rule is that it has the harshest effects on military type wins. But if you don't count Inca wins at the higher levels, then I think it still works.
what do you think of my suggestion?
This could be used to go both ways...some people don't care about scores and only want on the list, and they'll stay n the list with this, but not be counted for ranking at that level. You could also add in that if they stay 'unbalanced' for a certain number of updates(say, 2 or 3 updates), then they get removed until they do get balanced.
The thought had crossed my mind.
Why do you say that? It should be pretty consistent across all the tables.
That was what I talked about doing originally but I don't think it will have the desired result. We already have QM for people who like to check of the boxes. My hope is that people will branch out a little bit, stretch themselves, try new things.
I already really like the new EliteQM. The old one soon turned me off, when I realized that the best SettlerEliteQM would be Future Immortal games with 5 times more score than anything possible on Settler.
The new one seems already really awesome. Now it will actually hurt if somebody else submits a better game.
The 25% rule would make it even more awesome. I just fear that Rock of Ages will be tough. Having a game with 25%+ could prove difficult
The new spark of competition really fired me up, now it's really time to submit a shipload of games
I agree with this. It would be pretty demotivating to submit (or even start) a game that you know you will get zero points for, even after putting in a week's effort on it.
This would pretty much discourage me from ever bothering to submit a HoF game ever again, so safe to say that I disagree with this strongly. For example, it would be virtually impossible for me to ever submit a conquest game within 25% of the superhuman 2900BC dates that some people manage to get by using freaky super ideal leaders, mapfinder locations, and other factors. If I choose to use a sup-optimal leader for a particular setting, then I want to get some credit for having made the effort to successfully complete a game. I don't mind if I don't get a high score for it, but to get absolutely no credit at all for it would make it pointless to bother.
Exactly Harbourboy :-/ More entries must be encouraged. That should be the first priority... I understand you want to make EQM an elite community, but if you encourage underdog leader submissions, we will see more contributions to this beautiful game. Not exactly sure how you will do it, but a weighting system for leaders would be nice, imho, perhaps not as strict as I mentioned earlier, but still some weigthing on leaders could be nice... For example, leaders with more than 50 submissions receive a low weighting, while leaders in the 50-25 range receive a normal weigthing, leaders with 25-5 range receive even higher, while leaders with 0-5 submission range receive the highest weighting. Who cares if you submit the 3203th deity win by Incas when you are the first to submit a Charlamange deity win
I like this idea but the truth is that if Inca's are allowed then it will be very hard to get within that 25%; and on some maps virtually impossible.
I don't understand this argument. The proposed change is only for EQM, not for QM. If you are happy with submitting games to compete in QM, why not just ignore the fact that EQM has been introduced and continue as before?
I'm assuming that the Incan dates will be ignored for purposes of the 25% rule.
As to why it's harshest for military dates, it's because they have the largest spread in finish dates. SColony/SS are obviously the slowest VC so it's easier to get a better score (relative to the fastest finish) on those under the proposed system. Next smallest spread is probably UN diplo. I think domination and conquest have very large spreads on finish dates. I was looking at the emperor marathon setting game (in preparation for the new major). The fastest finish is 2590 BC (turn 94) with the Incas. The average finish date for that entry is 694 BC (roughly turn 200?). Currently the highest non-Incan slot is 190 BC (turn 281) with a QScore of 35 on the old system, 37 on the new system.
Because nobody is really going to look at QM any more. EQM is where it's at, and if I am going to get zero points in that (or have no chance of getting there because I can't match the superfast conquest dates, or domination victories using the silly settler spam strategy) then the motivation to submit falls to zero as well.
I understand your point. Anyway, if that is your concern, you could support my proposal described in posts #197 and #236 which doesn't have the zero score problem.
Your proposals are sure better than what we have now, but i prefer the phantom date: less complicated, and better, IMO.
I'd prefer the range of phantom dates that was suggested, as I too feel that in many tables(especially gauntlet tables) there is no need for dramatic changes, and that no game should be worth 0, but especially not a gauntlet game.
Separate names with a comma.