SGOTM 13 - Pre-game discussions

[mabraham]

One or two Teams might get the Quest that provides the free City Raider I for all Swordsmen for building some Barracks. The other teams may or may not get a different quest. My point is for a War agenda, a Team might get a huge boost by completing the free City Raider I Quest, especially if they are running Hereditary Rule which provides another free promotion (if I recall correctly). Getting a free CR1 promotion for all Swordsmen built is a huge bonus for a war effort and I doubt that all Teams would even get the opportunity to complete this valuable War Monger Quest.

Sure. The playing field is not level, and cannot be level. Even if you take out huts, barbs and events, a good team that pops a T20 gems will beat every other team. Likewise the team that gets lucky with no AI rush-building a wonder critical to its strategy, but which they delayed to achieve something else. Likewise the team that doesn't prioritize Alphabet and an AI randomly chooses to tech and trade it. Likewise the team that has an early forest growth that generates a useful bonus or extra chopping-. Likewise the team that correctly guesses which way to circumnavigate on an archipelago map. If you want a level playing field, play succession chess-of-the-month

In each game, I doubt that more then a few quests will be offered and I doubt all teams will be offered the same quests. I believe that which quest offered is randomly chosen.

Yes, but the set of available quests is fixed at the time the map is generated, like shuye said. The Python source code is available, and can be read by those with enough interest.

The Quests are also Era Centric, so a Team that passes through the Classical Era without this quest being offered will never get it in the future ever.

Shrug. That was the choice of that team. They knew the consequences (or should have, and can't complain if they didn't do enough homework). The game is about managing risk and reward with incomplete information. Experience, knowledge and judgement are critical to doing well. Sometimes they'll be enough to dominate the randomness, sometimes they won't.

 

[LowtherCastle]

Quests favor less skilled teams which balances the playing field in a sense.

 

[mabraham]

Quests favor less skilled teams which balances the playing field in a sense.

That is, quests (and events, etc.) generate randomness, which generates a wider spread of results, and gives a greater chance that the final results of different teams is not reflective of the skill difference.

Ask a pro game player how much of a bankroll he needs to make a living playing against opponents of a given strength. It is well known that you need a larger bankroll to play bridge, than backgammon, than poker, and that reflects the relative role of skill and randomness in the games. At bridge, where the random component is largest, top seeds have been known to lose all-day head-to-head knockout contests against bottom-seed opponents. That would never happen at poker under the same kind of conditions (i.e. very slow blind increments), and certainly never happens at, say, Grand Slam tennis tournaments (even though Grand Slam bottom-seed entries are pro players, unlike poker and bridge tournaments).

That lesson applies to GOTM contests too. More randomness makes the contest one less likely to be determined by skill. Hence LC's point.

 

[Fluroscent]

I am very happy by the decision to turn off random events.
But I also would like to apologize to DS if my ranting felt like undue pressure. It was pointed out to me that certain parts of my posts could be considered threatening (about leaving if random events were included), which was never my intention. I (obviously) feel strongly about the issue, and that was all I was trying to convey. I am happy to see the reasoning for turning off events being the general consensus of the SGOTM crowd.

Now let's have a great game everyone!

 

[BLubmuz]

Are Bad Events really so bad that a team of strategic thinkers cannot find some way to deal with them, much like they would any other screw that an AI throws into the inner workings of your well-oiled machine?

Are Good Events REALLY so good that they will catapult a team to a Gold Medal? Really? Do you honestly believe so?


"Well, if teams are close anyway and if one team gets a really bad streak of luck and the other team gets a good streak of luck, then yeah, maybe it might make a difference in one placing in the standings..." Maybe. Or are we just fear-mongering?

I think that in the just-finished SG, this kind of opposite event happened to the Gold and Silver winners in the early game could have reversed the final placement. So, far to be game-breaking, those events can make a difference in a competition like this. I'm happy they are left out, even if i have some fun with them in my solo games.

Same for goody huts: team A pops a map, team B pops a tech. Don't tell me that the wheel or BW or any other early tech in the early game won't affect the final result. It costs twice than Education in terms of turns.

I'm happy those are removed too. Fun, sure, but in solo games. This competition lasts many months and many people is putting great effort on it. Better let the field evened.

You have enough randomness with events and goody huts off, i will (probably) never forget that GLH on turn 61 in my last SG.

 

[Brian Shanahan]

I'm happy to. In a normal game or SG it doesn't matter, as you're playing for enjoyment and learning. But an SGOTM is compeditive first, and learning/enjoyment second. And for the game to be compeditive needs must have the playing field as equal as possible. And the way of doing that is by taking out all the optional random elements.

 

[Gumbolt]

Quests make things interesting. You can play a strategy to trigger certain events. It is not all down to dumb luck!

 

[Sun Tzu Wu]

Quests make things interesting. You can play a strategy to trigger certain events. It is not all down to dumb luck!

Do you mean trigger a quest or an event? I think you meant to say trigger certain quests, right?

How do you trigger an event that gives you a diplomatic bonus? I thought events were essentially random.

I agree that quests may not be quite so random; at least the quest you are given, usually makes sense for current Era of the game. Also, quests give you a choice, at least after they have been offered: Ignore or pursue the quest.

Too bad quests and events must be selected or rejected as a pair. It might be fun to play a game with quests, but no events.

Sun Tzu Wu

 

[Deckhand]

How do you trigger an event that gives you a diplomatic bonus?

as an example: build an airport or do whatever unlocks the airplane crash.
edit: they are random but they have preconditions (e.g. the discussion above about keeping cities small enough to avoid slave revolt)

If events were on, we would all learn a lot more about them.

If the feeling is the are "unfun". Then ok, leave them off.
But if the concern is just 1st & 2nd place be swapped or 3rd & 4th. Then leave them on.

 

[lymond]

My experience with events - when they are actually on - is getting 4 o 5 consecutive Slave Revolts and some random destroying of buildings.

 

[kcd_swede]

That is, quests (and events, etc.) generate randomness, which generates a wider spread of results, and gives a greater chance that the final results of different teams is not reflective of the skill difference.

Technically, increasing the number of random occurances would decrease the impact of any given random occurance.

But no random events and no huts is fine with me, as would be including them. Its just a game.

 

[neilmeister]

Its just a game.

Whoa ! Thats blasphemy, we are after all on the CivFanatics site.

Seriously though, I don't really care one way or another.

 

[mabraham]

Technically, increasing the number of random occurances would decrease the impact of any given random occurance.

Sure, but it's the comparison of

• the variance of the expected value of the result (i.e. SGOTM finish date) with events on, and the "expected value" of fun from those events, and
• that variance and value without events

that is the issue for whether people want to participate in a months-long competition. The former variance has to be larger, which decreases the chance that the results reflect the skill applied, and accordingly increases the chance that a less-skillful team can produce a good result. If people want "fun" events in their games there's nothing stopping them playing solo games That said, we want people to have fun in SGOTM well, but there is some conflict in the ways different people have fun. That's life

The underlying randomness of in-game occurrences is still there either way, but as kcd_swede observes, the consequences of losing your 98% CR3 maceman is lower with events on (because some big shiny event/quest might happen to compensate, or 15 slave revolts to make things much worse) than without events on. Events add some (relatively large) random effects to the game, which softens the impact of all the other random effects, while increasing the range over which the randomness affects the result.

 

[Sun Tzu Wu]

Technically, increasing the number of random occurances would decrease the impact of any given random occurance.

True, but then the signal (player skill) gets drowned out by all the random noise (uncontrollable, undeserved penalties & rewards).

In a SGOTM, you want your signal to noise ratio to be a high as possible so winning a SGOTM actually is a meaningful accomplishment, rather than which Team was lucky enough to Win.

Sun Tzu Wu

 

[Sun Tzu Wu]

Sure, but it's the comparison of

• the variance of the expected value of the result (i.e. SGOTM finish date) with events on, and the "expected value" of fun from those events, and
• that variance and value without events

that is the issue for whether people want to participate in a months-long competition. The former variance has to be larger, which decreases the chance that the results reflect the skill applied, and accordingly increases the chance that a less-skillful team can produce a good result. If people want "fun" events in their games there's nothing stopping them playing solo games That said, we want people to have fun in SGOTM well, but there is some conflict in the ways different people have fun. That's life

The underlying randomness of in-game occurrences is still there either way, but as kcd_swede observes, the consequences of losing your 98% CR3 maceman is lower with events on (because some big shiny event/quest might happen to compensate, or 15 slave revolts to make things much worse) than without events on. Events add some (relatively large) random effects to the game, which softens the impact of all the other random effects, while increasing the range over which the randomness affects the result.

You may believe that adding more randomness "levels the playing field", but it doesn't. Each Team doesn't simply collect roughly the same amount of bad luck and roughly the same amount of good luck. Very few Teams will receive roughly equal amounts of good and bad luck that evens out in the end. Many Teams will be subjected to more bad luck than good luck and about as Many Teams will be subjected to more good luck than good luck. One Team might be very unlucky and another Team might very lucky and largely due to that undeserved luck, Win SGOTM.

Take the simple example of flipping ten coins, each with a head on one side and tail on the other side (they are perfectly balanced = not loaded coins). Since these coin flips are independent of each other, one must consider any combination of flipped coins to be equally likely as any other. There are 2 to the 10 power (1024) such combinations to consider. On average, one expects to get 5 heads and 5 tails and while that is the most likely result, it happens far less often than one might expect, 24.6%. This is because 4 heads and 6 tails and 6 heads and 4 tails are not that much less likely, 20.5% each, but they subject a Team to either a net gain of two unlucky events or a net gain of two unlucky. See the probability table below and a reasonable expected placing of 20 Teams on that table EDIT: (actually 1 Team is left over and can't be placed, because the remaining percentages aren't large enough and due to round-off in the larger ones).

10H/0T = 0.1%
9H/1T = 01.0%
8H/2T = 04.4% 1 Team  +3 Luckiness
7H/3T = 11.7% 2 Teams +2 Luckiness
6H/4T = 20.5% 4 Teams +1 Luckiness
5H/5T = 24.6% 5 Teams 00 Luckiness
4H/6T = 20.5% 4 Teams -1 Luckiness
3H/7T = 11.7% 2 Teams -2 Luckiness
2H/8T = 04.4% 1 Team  -3 Luckiness
1H/9T = 01.0%
0H/10T = 0.1%

Consider the Heads results are rewards and the Tails results as penalties.

As you can see, more randomness doesn't level the playing field, it actually does the opposite, spreads them over the Luckiness axis in a bell curve.

An extremely skillful Team may overcome -3 Luckiness, but could they catch up with an almost as skillful Team with +3 Luckiness? They probably can't even make up a 3 Luckiness deficit, much less a 6 Luckiness deficit.

Why would we even want to put our teams through such a Luckiness/Unluckiness filter? That's what Tribal Villages and Events essentially do. I hope I've convinced you that having more of them doesn't help. Try a 20 coin analysis and you will see the Teams spread even more widely as the bell curve becomes shallower and wider and an even fewer percentage of all Teams will have 0 Luckiness (median).

Sun Tzu Wu

 

[mabraham]

You may believe that adding more randomness "levels the playing field", but it doesn't.

If you're going to use quotation marks (ie. adding randomness "levels the playing field"), please quote something I said, not the opposite...

Sure. The playing field is not level, and cannot be level... If you want a level playing field, play succession chess-of-the-month

That is, quests (and events, etc.) generate randomness, which generates a wider spread of results, and gives a greater chance that the final results of different teams is not reflective of the skill difference....

More randomness makes the contest one less likely to be determined by skill...

The extra randomness of events and quests doesn't "level the playing field" (to use your words), it adds a random positive or negative factor to everyone's result. See above where I talked about "wider spread", and below where you talk about "spreading them over the Luckiness axis".

As you can see, more randomness doesn't level the playing field, it actually does the opposite, spreads them over the Luckiness axis in a bell curve.

Even the person who did talk about "balancing the playing field" qualified that point of view...

Quests favor less skilled teams which balances the playing field in a sense.

If we were playing a large number of SGOTMs then it would all average out in the long run, but we don't have that luxury. So large sources of randomness (events) are not so desirable. Quests are similar, but not as large a source of randomness as you seemed to believe earlier in this thread, inasmuch as the same pool of quests are potentially available to all teams. However, we are both making the same general point...

 

[Sun Tzu Wu]

If you're going to use quotation marks (ie. adding randomness "levels the playing field"), please quote something I said, not the opposite...

Sorry, my use of quotation marks was simply to emphasize the phrase "levels the playing field". I probably should have italicized it instead. When I quote what someone is saying in these forums, I almost always use the "QUOTE" and "/QUOTE" meta commands.

In your original post, it seemed to me that your solution to having some bad (meaning instant rewards/penalties for doing nothing or something that requires near zero skill) random elements was to add more bad random elements. It seems that I probably inferred something in your posting that you did not intend or didn't even state.

Sorry, I must have completely missed the point of your post. So, we are actually in agreement?

The extra randomness of events and quests doesn't "level the playing field" (to use your words), it adds a random positive or negative factor to everyone's result. See above where I talked about "wider spread", and below where you talk about "spreading them over the Luckiness axis".

Ok, so we do seem to be in agreement, assuming that your "wider spread" means essentially the same thing as my "spreading them [Teams] over the Luckiness axis".

I'm a firm believer in applying mathematics when doing so appropriately. I believe that my table was helpful in showing how bad randomness can subvert the Team rankings by effectively giving each Team a net random bonus or net random penalty. The best Team may not be able to overcome a series of bad random penalties whose combined expected occurrence is 5% (meaning that 1 team in 20 might really suffer such a net # of penalties), at least to the extent of catching up with a very lucky and almost as skillful Team.

If you agree that my application of mathematics is sound, fine. If you have no opinion on its validity, Ok. If believe that it is invalid, please show me how it is invalid or just explain your reservations concerning its alleged validity.

Even the person who did talk about "balancing the playing field" qualified that point of view...

And what point are you making here? That I should have responded to this person's post instead and taken into consideration their qualification?

If we were playing a large number of SGOTMs then it would all average out in the long run, but we don't have that luxury. So large sources of randomness (events) are not so desirable. Quests are similar, but not as large a source of randomness as you seemed to believe earlier in this thread, inasmuch as the same pool of quests are potentially available to all teams. However, we are both making the same general point...

Sorry, your statement above is where our viewpoints do (in an academic sense) diverge. I don't think it will matter how many times the SGOTMs are played, depending on the signal to noise ratio. The signal is player skill. The noise is bad randomness = randomness that just happens and has no real connection to player control. If signal <= noise (I don't for a second really believe this), we'd be better off not playing with these noisy options at all. I do believe in reducing noise in games as much as possible.

In a practical sense, I do agree with your statement. The best SGOTM Teams will win the most SGOTMs over a large number of SGOTM competitions despite bad randomness, because I don't believe Civ IV Beyond the Sword has enough bad randomness options to overwhelm their skill as players.

In a practical sense, we do appear to be in full agreement.

Sun Tzu Wu

P.S. Still waiting for a Proof of Fermi's Last Theorem, using methods available to him when he was still living. Perhaps his statement that the book's margin was too small for his proof was a bluff or perhaps his unwritten proof contained a flaw after all. I hope some day, a mathematics graduate student takes up this challenge as his dissertation and prevails! It would be nice to see Fermi vindicated in this way.

 

[Fluroscent]

If we were playing a large number of SGOTMs then it would all average out in the long run, but we don't have that luxury. So large sources of randomness (events) are not so desirable. Quests are similar, but not as large a source of randomness as you seemed to believe earlier in this thread, inasmuch as the same pool of quests are potentially available to all teams. However, we are both making the same general point...
Sorry, your statement above is where our viewpoints do (in an academic sense) diverge. I don't think it will matter how many times the SGOTMs are played, depending on the signal to noise ratio. The signal is player skill. The noise is bad randomness = randomness that just happens and has no real connection to player control. If signal <= noise (I don't for a second really believe this), we'd be better off not playing with these noisy options at all. I do believe in reducing noise in games as much as possible.

I am 99% sure mabraham meant the same thing mostly. What a large sample size will essentially eliminate is short term luck effects, and thus the higher skilled team would win the most, and the statistical material would be accurate in that sense. That is the "average out" part mabraham mentions. Basically, the effect of the luck factor (noise) will diminish, while the skill factor (signal) will remain constant the more iterations you go through. And actually, this is also the case, even if "signal <= noise" as you say (skill and luck are more instructive terms in my opinion). In fact even if luck >> skill, the same applies. You just need a lot more statistical material to determine who is best then.

Now, fortunately, the skill factor in Civ 4 is large enough compared to the luck factor that while a single game cannot with certainty establish the best team, the end result of a single game is still meaningful enough. Ie. SGOTM12 did not establish which of OSS and PD played the best game being only one turn apart, but the spread in results makes it extremely likely that two games they played were the best two played this time around.

I am also 99% sure mabraham plays poker

 

[LowtherCastle]

OSS played the better game. Fact.

;-)

 

[mabraham]

Sorry, I must have completely missed the point of your post. So, we are actually in agreement?

Yep, pretty much in agreement. No problem re: quotation marks.

I am also 99% sure mabraham plays poker

Not much, much more bridge and Civ