Statistics

[kjn]

While we waited for the timer to run down, I did some statistical analysis of the the team performances so far.

Expansion

To quantify this, the first settled city of each new set of cities received a score of 0, while the rest of the teams received a score equal to the turn they settled their city in the set minus the turn of the first city in the set.

Example: Univers was the first to three cities. They receive a score of 0 for that. We settled our third city 8t later, and get a score of 8.

Then I summed all these values (teams with unsettled cities in the set get the current turn+1 as value here), and normalised the values by subtracting the value of the best team. Ie, the value is the number of turns that the team is ahead compared to slower-settling teams.

Univers: 0
CivFr: 5
Germans: 7
Apolyton: 18
RB: 25
CivPlayers: 30
CFC: 35
Spanish: 47
WPC: 53

We are solidly in the middle of the pack here.

Research

Here I simply used the score from tech, and divided it by the turn they last received a tech. Ie, for us it's 71 (12 ancient techs) divided by 60 (the turn when we had finished Writing). Then I normalised by dividing with the value for the poorest techer:

RB: 1.71
Apolyton: 1.47
CivPlayers: 1.47
CivFr: 1.4
CFC: 1.33
WPC: 1.25
Univers: 1.11
Spanish: 1.02
Germans: 1

We are pretty clearly the tech leader here. It also shows that Apolyton and CivFr has done a good job at balancing expansion and teching; the Germans and Univers, not so much. The Spanish are in the bottom of both teching and expansion, and are very clearly an also-ran.

City management

Assessing tile management quantitatively from C&D is probably hard, but there is a possible way to gauge it: the time between the founding of a city and its first growth. I'm including only cities founded on T63 or earlier in this comparison. Cities which haven't grown get the current turn plus one. In some cases (notably the Germans and CivFr) it's unclear if the most recent city has grown or not, due to whippings back to size 1, but having any city stuck at size 1 due to whippings isn't that good anyway.

City	RB	WPC	Spanish	Apolyton	Germans	CivFr	CivPlayers	CFC	Univers
City 1 turns to 2	4	7	8	6	6	5	4	9	4
City 2 turns to 2	6	8	11	8	7	7	7	6	7
City 3 turns to 2	4			5	17	14	6	5	6
Average	4.66	7.5	9.5	6.33	10	8.66	5.66	6.66	5.66

The only teams that has managed to grow a city from size 1 to 2 in 4 turns are Univers and CivPlayers. On the flip side, the city management of the Spanish, the Germans, and CivFr is very poor, while the rest of the teams are pretty mediocre.

Conclusions

Well, we're the clear leader in two categories while middle-of-the-pack in the third. Apolyton and CivPlayers appears to be our main contenders to win this thing from pure Civ ability. The team that manages to surprise me a bit is Univers: they've shown good management and expansion, and would be a strong contender if they hadn't made so poor macro decisions in their opening.

[Shoot the Moon]

Why for research not use total beakers? We know decently well what each team has researched right?

[kjn]

Why for research not use total beakers? We know decently well what each team has researched right?

No, we don't. There are lots of question marks on techs from T40 onwards. Also the beaker costs would mask the discounts. I think this way is simply much simpler with no real loss of accuracy.

[Fintourist]

I think this is all very interesting, thanks kjn!

[kjn]

I finally managed to beat at my spread sheets to produce a nice scatter plot. Here is one showing the score from tech as a function of the number of tiles worked (the sum of pop and cities, ie a size 1 city counts as 2 and a size 4 as 5) by each team.



The fun thing about the plot for this turn is that the only outliers on this turn are Univers and the Germans - every other team follows in a pretty linear fashion. But only last turn things looked quite differently.

I will probably do some more statistical analysis once T80 comes around.

[mostly_harmless]

Cool. Do this for each turn from the start. Then create an animation of how the points move.
How does the graph look like, if you include points from pop and wonders as well, because I think Univers is slightly undervalued.

And the engineer in me demands proper axis annotations!


mh

[kjn]

Cool. Do this for each turn from the start. Then create an animation of how the points move.
How does the graph look like, if you include points from pop and wonders as well, because I think Univers is slightly undervalued.

Ugh. I'm not sure Numbers (or my spare time!) is up to that.

Yes, I agree that Univers is easily underestimated here. I'd probably rank them above CFC right now, based on the degree of civ micro management they've shown. But they're still a fair bit away from CoL.

The reason I only make the graph a function of (pop+cities) and tech is that wonders skew things wildly, and having a function of (pop+cities) and (pop+tech) wouldn't be that interesting. Better to compare cities and tech in that case.

And the engineer in me demands proper axis annotations!

Now I found how to get those!

[antisocialmunky]

Can we get 10 turn resolution Why rely on the Civ graphs when we can make our own graphs?

[kjn]

Here's another scatter graph, showing the tech tempo (tech score adjusted over time[1]) as a function of the number of worked tiles (cities + pop) as of T80:



I think this graph gives the best view of the performance of each team of these I've made.

The Spanish, WPC, and the Germans are all clearly out of the race already - their tempo of teching is less than two thirds of the four tech leaders. CFC does a lot better, but are clearly outperformed by Univers, CivFr, and us in various metrics.

Apolyton has teched the fastest, but their growth is the slowest of the leading teams, unless they can grow back quickly and found new cities from their recent whip round.

So my ranking of the teams as of T80 would be:

1st: RB and CivFr
2nd: Apolyton and CivPlayers
3rd: Univers
4th: CFC
5th: WPC, the Spanish, and the Germans

[1] (Score from tech / Turn of last tech) / Worst tech tempo

[kjn]

My caveat about my raw data not being nicely regular proved to be correct. Disregard this analysis, since the numbers for several teams are incorrect.

The Whip

I sat down and made some functions in the C&D spread sheet for whippings, whip efficiency, and growth for the state of the game on T80. Here are the three key numbers I managed to compute:

Pop per whip: simply the average amount of pop that is whipped in each use of the whip. Generally, you would want a number slightly more than 2 in the early game.

Popwhip ratio: the amount of pop that has been whipped away divided by the current pop. No idea what a good number here would be, and it would probably be dependant on the map and the stage of the game.

Popwhip sum: the sum of whipped away pop and current pop. This shows the team that has managed to grow the fastest.

Team	Whips	Whipped pop	Pop per whip	Popwhip ratio	Popwhip sum
RB	7	16	2.29	0.67	40
WPC	2	3	1.5	0.2	18
Spanish	2	3	1.5	0.2	18
Apolyton	3	6	2	0.35	23
German	13	20	1.54	1.67	32
CivFr	1	1	1	0.04	28
CivPlayers	1	3	3	0.15	23
CFC	3	6	2	0.27	28
Univers	4	4	1	0.14	33

(With the caveat that my documentation has been regular and correct, and that I managed to get the functions correct.)

Some things stand out:

We are the team that has managed to grow pop the fastest among all the teams, only Univers and the Germans has managed to reach over 30.

The Germans have whipped away far more of their pop than any other team, their ratio of 1.67 between whipped away pop to present pop is more than twice ours (0.67), and we have whipped away twice as much pop as almost any other team ourselves.

Most whips are small, with 1-pop whips being the most common for most teams. Among the teams with multiple whips, only we, Apolyton, and CFC have a pop per whip value of 2 or higher.