Beta Gauntlet VII

superslug said:
At the conclusion of the Gauntlet, games will be sorted into the various Victory Conditions and the average turn number finish of each VC will be calculated. Then, the fastest turn finish from each VC will be taken and divided by the average turn finish for it's VC. The number one game farthest above it's class average will be declared the winner.

Division is horribly biased in favor of early victory conditions. You should use subtraction instead.
 
Since i am very bad at high research games, I gave it a try.
Diplomatic victory in 1675 AD. This time will be probably beaten by some hundred years.
 
To heck will "winning" this guantlet.

Must. Make. Boorish. AI. Become. Cultured.

Monte + Louis + Hatie + PA + Free Speech + Eiffel + Sistine + Statue of Liberty + Merchantilism + Warfare == Culture win?


We shall see... Or not, actually. I leave on a week long trip on Friday and won't have much time to play, but I want to see if a Culture win is possible. The plan is to rush to fascism for PAs then to all the +50% culture wonder techs so that the allies build em. Once everything is build, start attacking people to trigger war weariness which hopefully makes the AI use the slider, but making sure the foe won't get killed to trigger domination. Grab the Chapel and Eiffel along the way, get the allies to convert to the right civics, and hope for the best. Gift Great People and hope for the best. Max civs so everything is built in the AI capitols, which will hopefully become legendary. Crazy enough to work?

On a somewhat related note, I wonder if a detour for fascism for the research boost from an alliance is ever worth it in a tech game. You can change the AI's tech selections to what you want.
 
I thought your beakers per tech double when in a PA. So, this would never be a tech boost unless you're playing on Diety.
 
BlueRenner said:
Of course, you could use the idea of dropping the lower 50% of the games... but that just adds another layer of complexity into a system that's already a bit awkward. I think a new scoring metric is needed here.
- Bill

You nailed it. Keep it simple. However, the rest of your post was so complex I couldn't even follow you. Maybe you were you being facetious.

Anyway, Take a step back and forget about all the formulas and percentages for a minute. Ask yourself, who should be declared the winner.

The winner is the person who can achieve a victory condition and do it farther in front of the others in the same category.

Now, if there were 100 entries in each category, taking the top 50% or top 10 might be good idea, but there are not that many. So, keep it simple. Just count how many turns ahead of #2 you are. Now let's look at how elegant this plays out.

1. As people post their best times you can easily figure out how many turns you are ahead.

2. Submitting really bad (or even bogus) victory years doesn't effect anything.

3. If, for example, you're ahead 5 turns in the Space race, you are encouraged to play other victories and get within 4 turns of conquest, etc.

4. This rewards players for playing the harder win conditions. (Fewer competing entries)

5. It rewards players for playing the longer win conditions. (If conquest can be won in 40 turns, it can probably be won in 41 turns. Neither of these games took much thought or effort and don't deserve to be the ultimate winner. They probably won't with only a 1 turn lead.)

6. If you're the only one to win a culture victory in say 2000 AD. You're ahead of #2 by 50 turns (2050 AD). You will win this gauntlet (and should) unless someone else finishes a culture win too. Let's say someone is currently winning with a 10 turn lead in Diplomatic. It would be smart for them to try to get a culture win.
Now, if only one person achieves a Domination win in 1990AD too, they would win over the single culture person because the year is father ahead of the next (2050).

7. If you're ahead in one victory, you are now encouraged to share your strategy on the other win conditions so that players #1 and #2 finish closely. If someone pulls off a culture win, I can see the message board rallying to get someone else on the score board in culture. It's horrible to say "If you're the only one to win culture, you lose cus you're zero turns ahead of the average."

This could be the most fun/active gauntlet ever if we get it right.
I think this scoring system not only declares the correct winner, but encourages people to play all the win conditions. Sound good? Simple? or did I confuse you?
 
simple and encourages you to attempt all vc's, especially the harder ones like cultural. thumbs up for that, but how would you rank the rest of the submitted games?
 
pooLarized said:
how would you rank the rest of the submitted games?

Simple, all the fastest finishes (sorted by how many turns they beat #2), then all the #2's sorted by how much they beat #3, etc.
 
WastinTime,

My problem with your system is essentially the same problem I have with the experimental system in this gauntlet -- it takes the critical win condition out of your hands and places it in the hands of your competitors.

You don't win because you do very well in your best victory condition. You win because your best win condition has no players approaching your skill.

This may or may not have any relation to the skill level of the other victory conditions. I mean, imagine a situation where you and I are duking it out for the OCC Space Race crown, and finish up within two turns of each other. Then this other guy waltzs by both of us because neither of us did Conquest (anarchy exploit land as it is) and the field is comparitively much less skilled.

In that situation, you would lose the gauntlet not because you didn't put out supreme effort -- you'd lose because I also went all-in. This results in a system where direct confrontation between good players is actively discouraged, as it would generally ensure that both would lose.

The alternative would be to near-require participation in all victory types. I don't think anyone wants this. Its hard enough maximizing the given victory condition in the given timespan.

Now, as for my idea, I'll give another shot at explaining it. It only deals with comparing the fastest finish across Victory Conditions, so only getting the top spot in a VC ensures you have a chance of winning the overall gauntlet.

1) Take the turn number of the fastest finishes of each VC.
2) Multiply that turn number by a modifier that would normalize the scores.
3) Compare the composite scores, the lowest one wins the Gauntlet.

The modifiers would be assigned fairly arbitrarily at first, but would be refined as data comes in from the Gauntlets themselves. The modifiers would be very high for Domination to reflect its early win dates, and very low for Space Race to reflect its late win dates. All this does is to attempt to accurately compare win dates across the victory conditions.

I threw in a little aside at the end, detailing that you could encourage harder settings (barbarians on, playing with Alexander, etc) by giving special modifier bonuses for these wins. These would essentially be Special Victory Conditions which could change with each gauntlet, and be treated as seperate with its own modifier.

- Bill
 
I wanted to knock out a competitive conquest before doing space. I don't think I want to mess with diplo, culture, domination tho (perm alliances are a pain).

Pangaea, pressed/tropical, 4 enemies (hatty, asoka, gandi, izzy). I got a 460 BC conquest. Techs were mining/BW/IW/Wheel.

I did want to check that Ronald was doing a quick speed game. I'd estimate that it took an extra ~15 turns because of the geometry, but 1500 BC is only 50 turns into a quick game. What map type can you use that will even let you reach all 4 enemies in 50 moves? To put it another way, a 1500 BC quick finish would be equivalent to a 2000 BC normal and...well, a lot earlier at epic or marathon. I should be able to push it to 1000 BC with a decent set of starting AI positions, but if you really have to get to 1500...
 
Don't forget. We're not talking about all gauntlets, just a multiple vc one. The whole point of this gauntlet is to play all or most vc's!
The leader has been chosen for you. As well and size, era, etc. Many of the iterations are eliminated. AND its quick speed.

There should be plenty of good competition in each vc.

If I only beat you by 2 turns in the space race, that's not all that spectacular. If you beat everyone by 5 turns in Diplomatic, you win this gauntlet. Seems fair to me.

BlueRenner, your idea isn't bad. However, there is no way to come up with the multipliers. Even after several gauntlets of data and you come up with reasonably good multipliers, those probably won't work for the next gauntlet because the size or speed has been changed. Think about it. If you said the next Gauntlet will be on Islands. Conquest will be MUCH slower. It's multiplier will eliminate it.
Or people will figure out "Oh, with the current multipliers, gauntlet X will be won with conquest. " Then it's just another Conquest Gauntlet. Again, the whole point is to have players go for all VC's. It was smart of them to make it quick speed and force a leader this time too IMHO.
 
I should point out that after this Gauntlet, most if not all future Gauntlets will be single victory conditions, not multiples. The purpose for testing the formula is not for Gauntlets, but for the Quatromaster's Challenge, where it's necessary to have a way to compare games from any two HOF tables to each other.
 
superslug said:
I should point out that after this Gauntlet, most if not all future Gauntlets will be single victory conditions, not multiples. The purpose for testing the formula is not for Gauntlets, but for the Quatromaster's Challenge, where it's necessary to have a way to compare games from any two HOF tables to each other.

:( With the proper system in place, I think multiple VC's could be the best gauntlet ever. You should reconsider.

superslug,
I'm assuming that even if you like some of the ideas presented here that you are going to let this one run to the end with "the formula" since it has started already. Can you confirm that? I'll probably be on the sidelines for this one...mostly because of a vacation. However, if you were to change the scoring to encourage attempts at cultural, domination, etc. I'd HAVE to find time for this.
 
Oh, I think it'd be very easy to come up with multipliers. The Hall of Fame community is already very good at estimating the finish time of various gauntlets, and this estimation ability will only improve as players (and admins) become more experienced. Finding the 'perfect' multiplier set might be flat-out impossible, but we will be able to get extremely close.

The multipliers would not be static, either. As you note, Islands Conquest takes longer than Pangaea Conquest. So, in a gauntlet where Islands are required, you lower the Conquest modifier a bit, based a the estimation of the fastest finish. You guess based on past experience. Perfection isn't the goal here -- just aligning them so the results are somewhat sane. It will be rapidly apparent what works and what doesn't.

Healthy debate revolving around which Victory Condition & Multiplier combo would produce the highest final score will only help refine the process of selecting the modifiers. The system strengthens itself, so long as the person setting the modifiers is paying half attention.

If we really want to force players to go for all VCs, then just make it a requirement that they win with all conditions and then sum the total to determine the winner. This would drive me nuts, but dancing around the idea while trying to artificially stimulate it is worse.

Now, switching gears entirely: A hard-fought 2-turn Space Race victory is not inferior to a cakewalk 5-turn Diplomatic victory. The key concept here is the "Strength of Field", which comes up so often comparing sports teams against one another. If you have a perfect record, but have only beaten chumps, you are not rated as high as someone with a checkered record who goes up against the best of the best.

The problem with just a straight turn-comparison is that there is no way to evaluate the overall field strength. I suspect this is why the original concept delt with averages -- it is a way to get a sense of the median skill within a singular VC. However, I still dislike this for reasons previously mentioned.

- Bill
 
ohioastronomy said:
I wanted to knock out a competitive conquest before doing space. I don't think I want to mess with diplo, culture, domination tho (perm alliances are a pain).

Pangaea, pressed/tropical, 4 enemies (hatty, asoka, gandi, izzy). I got a 460 BC conquest. Techs were mining/BW/IW/Wheel.

I did want to check that Ronald was doing a quick speed game. I'd estimate that it took an extra ~15 turns because of the geometry, but 1500 BC is only 50 turns into a quick game. What map type can you use that will even let you reach all 4 enemies in 50 moves? To put it another way, a 1500 BC quick finish would be equivalent to a 2000 BC normal and...well, a lot earlier at epic or marathon. I should be able to push it to 1000 BC with a decent set of starting AI positions, but if you really have to get to 1500...

1500bc conquest shouldnt be impossible, i had a 800bc win despite screwing up a bit. i overestimated the AIs ability to defend and waited too long to attack.

i know i could have atleast shaved off 9-10 turns if i had played properly. i did have bronze next to river so i didnt have to wait for ironworking, but the AI were fairly far away and unfortunately the last one snuck away a settler which i didnt notice so that was 5-6 turns extra right there.

and the minimum amount of AI players on small is 3 right? i dont know why youd want to use 4 for conquest.
 
I'm a total noob.. how come my Jaguar Warriors keep getting killed by archers in my enemy's town? I give some of them +25% city attack and others I give +25% vs. archers. Do you usually wait for 3-4 warriors to pile up into one stack before attacking? Help! :)
 
WastinTime said:
Are you saying the multipliers are set at the start? or change as scores come in?

The scores will be defined at the start and then be static throughout the gauntlet's run. After the gauntlet is completed, they can be adjusted for the new gauntlet based on the new information, or the quirks of this new gauntlet.

Changing the modifiers mid-gauntlet wouldn't be the fairest thing to do, I'm thinking... perhaps let them be somewhat fluid for the first three days or so to let the players give input, but certainly no longer.

- Bill
 
We're going to stick with the originally stated formula. Input is of course welcome on other formulas, as after the fact analysis is inevitable.
 
I just got a 400 AD conquest win. :) Its not a 1650BC, but a big improvement for me, a noob. That pangea combo mentioned above worked great. I was pretty close to my neighbors, but unfortunately they kept spawning off cities and it became hard to kill them all off. I suppose maybe that is why a 1650 is actually possible: no extra cities.
 
Well, my conclusion is that PAs... suck. There's a 1 PA per civ limit that I didn't know about. That limit is rather annoying, but you can work around it. Or could except....

It seems your lone PA AI must have ALL THREE legendary cities. Illogical, and makes life much harder. At least the victory conditions doesn't list allied cities in the cultural win condition.

The real problem is the inability to ask the allied civ to change civics or religion. WTF. I can ask to change before the alliance, but afterwards I can't? I hope this gets changed in a patch. No forced Free Speach and Merchantilism hurts culturaly.

This makes a cultural win MUCH harder. I think the only way you can have a chance is a team battleground. Get lucky and have Louis on your side, then feed him tech so he builds all the wonders and pray they all end up in the same 3 cities. Have the other 2 AIs be Alex and Ghengis or whichever AIs are least likely to build wonders (hmm, that would actually be Monte, but we must play as Monte). You'd have to delay the PA till the very end so you can continue to ask Louis to swap to Free Speech and Merchantilism. Meh, makes Dom look positively easy by comparison.

If anyone tries a cultural win, good luck. You're gonna need it. At this point, it appears all but impossible.
 
Top Bottom