Universal metric for progress?

[Jtownsend]

So, there've been informative threads recently about the questions of how to time the early rex - after growing to work all expanded tiles - and what to send with settlers - a worker and an escort.

But one thing that's generally considered unattainable (at least as far as I've seen) is a universal "How many X do I need by Y" yardstick. How many cities of which types, etc.

This put me in mind of an interesting point I think Dave McW made - I couldn't find it after furious searching so perhaps it was actually a very old one - that what mattered wasn't cities or total pop, but the number of citizens working improved tiles.

Could this be a universal, one-sized fits all yardstick for progress, at least if one set certain other criteria like difficult level and turn speed?

IE, couldn't one say, Emperor level, normal map, normal turns, X citizens working improved tiles by 1500 BC, 1000BC, 500BC, 0, etc? Or, if that were considered too generic, even total goals for hammers, commerce and food?

And then, if the powers that are could agree on such benchmarks, they could then be played around with for people trying to climb the ladder, with either new scales or estimates made for lower difficulties. (And of course higher. I suggested emperor and not immortal or deity because I assumed that some of the difficulty tweaks might be less translatable up and down the difficulty scale, as opposed to Emperor.)

It seems like it'd be useful to have an entirely immutable and unambiguous yardstick with which to measure progress in 'learning games' in which you only play to the early hundreds AD to try and improve efficiency. Part of evaluating one's play could be, well, looks like I didn't have enough citizens (or better, hammers, commerce, etc.) Where did growth fall down? Did it need to happen in another city? Etc.

I mean the why's are so varied - start location, high or low happy cap, fundamental civics choices - that it sometimes makes the advice for playing civ sound like the advice for playing chess. Open your files, develop your pieces, play for the centre unless you're an expert.... There's opening sequences, but in a lowbie game you're just 5 moves when the sequence stops and the best advice becomes "be good at chess."

 

[r_rolo1]

That isn't a universal metric, especially if you think on OCC. And not only that... what is better: 10 worked tiles and the AI having 20 or 5 worked tiles and the AI having one or none?

 

[Joshua368]

And then specialists throw the whole thing off.

 

[DaveMcW]

But my point is, pop and settlers are a red herring. The real way to measure early progress is by number of citizens harvesting improved resources.

After you've claimed every resource possible, you start producing GPP which really messes up the "economy score" calculations.

 

[Jtownsend]

So in other words, the originally best metric - citizens working improved tiles - stops working at a certain point - could that point be standardized? - and then the only meaningful universal yardstick - again given a standard map at a set difficulty - would be net hammers, beakers, gold, GPP and food surplus, but these would be too variable in their ways of arriving to constitute a very -useful- yardstick?

Or could there still be a useful hammer/beaker/gold/gpp/suprlus food by date yardstick for a standard game at difficulty, even if the means of attaining it aren't standardizable, and are indeed problematically varied?

For a given "testbed," if you like, Emperor/balanced/Normal/6 AIs, the AIs' extremely varied performances would still create an average performance in Hammer/beaker/gold/gpp/food surplus, if you average out the 6 AIs over an adequate number of games, whether that's 10, 20 or more; even with the crazy number of variables in civ the AI has a chance to approximate its average 6 times per game; after 20 games you'd have 120 performances to average.

This is of course ludicrous to actually do. It's just to suggest that a number must actually exist for the 'average' gross output of good things to output in civ, even if the weight by which each of the outputs is valued varies enormously by human players. (On which note, I've forgotten entirely about culture and espionage, both of which I not coincidentally disparage.) So if that number exists, maybe rather than arrive at it systemically we could just eyeball it by having good players strive to play for 'generically good gameplay with generically good output' at emperor/standard/normal/6AI, and then tally up their hammer/beaker/gold/gpp/surplus food at, 10 set dates dates from the start of the game?

Or Monarch, if such 'impressive generic output' gameplay is just inappropriate at emperor.

 

[Soirana]

Feel the difference *imrpoved tiles* and *imroved resources*.

There is quite a difference between plains farm and irrigated corn.

 

[r_rolo1]

Any metric that is based in your own empire resources only ( like if you were playing in a sandbox game ) is faded to not be representative. You're in the game to win vs the other civs and as long as you can keep competitive or even with a upper hand ( no matter how ), you'll be ok. It is better to bankrupt and having units striking, but with all the AI cripled to oblivion, than having a very solid empire in economical terms but with a gigantic AI or 2 with the double of everything you have.....

 

[Refar]

I think the assumption: "best available improved tiles" is obvious. Edit: Crossposted, comment went to:

Feel the difference *imrpoved tiles* and *imroved resources*.
There is quite a difference between plains farm and irrigated corn.

----

The questions on my mind are however - is such a metric really needed ? would the effort to collect and evaluate the data really be worth it, in terms of improving the gameplay ?

In a particular game it is usally fairly easy to see "how we are doing", by comparsion to the competitions in terms of achieving immediate objectives (landgrab/wonders/conquest - depends on stratey) and the final goal (victory).
And in the end being better than the competition is the only thing that matters anyway...

 

[Jtownsend]

I meant "improved bonus tiles," my apologies.

Based on what I've learned I think up to a certain point, which unfortunately would vary somewhat by luck in terms of bonus tiles and hap cap, that up to a certain point "number of citizens on improved bonus tiles" is the all-consuming focus, particularly in early turns which have 'compound interest' value in the game as a whole.

would the effort to collect and evaluate the data really be worth it, in terms of improving the gameplay

I think so. I said, in a hypothetical sense, that, with infinite analytical resources and a large number of games, there would be a way to find out what the testbed performance of "an AI" is on (Emperor/Standard/Normal/6 opponents) - that is, such an average exists. Actually - finding- that result is an absurdly involved project, as we'd like say, 10 answers for hammers/beakers/gold/GPP/excess food even if we left aside culture and espionage, which may be inappropriate.

But, much less involved would be asking good players - for example, people who still play on emperor most of the time, but kick ass at that level - to record a screenshot of their city screen at 10 set dates or so. Then if they feel the game was a case of generic economic success, put the screens someplace, I'd tabulate them, and we could see what generic economic success looks like at dates 1 through 10. If Emperor is not conducive to 'generic games' happening, perhaps Monarch would be.

It may not be useful! I mean, a list of how much material you should have at 10 discrete points in a chess game would be an absurd 'learning aid.' But I guess that's what I'd ask the experts.

A lot of the good advice out there says, eg, "get 2-3 good production cities, a good GPP farm, 2 commerce cities with 10+ cottages or equivalent pre-bio, then evaluate where you need to go from there." That in itself constituting the "early game" in which a tutorial is most helpful, since it's very varied from there. But what I find, when using that helpful guideline, is, hm, I really don't have a good second commerce site, I have to go put it way the hell over there, or my production cities are kind of mediocre, or I have ton of plains tiles. And in those cases I can be following the 'city level' instructions but in fact be missing the level of production or commerce or gpp generation that those city guidelines are implying. Wheras an average performance target for the empire is either met, or not.

Any metric that is based in your own empire resources only ( like if you were playing in a sandbox game ) is faded to not be representative.

Any single game, yes. Any sandbox game, certainly. But it isn't particularly unwieldy to get 10 city screen screenshots from multiple emperor/monarch players each playing a game of Emperor (or monarch) / balanced / normal / 6 AIs. It isn't requiring people to go much out of their way, and you wouldn't need all that many samples, I wouldn't think, to arrive at a very tentative average performance. I guess until one got the figures it wouldn't be clear how variable they'd be. If 10 samples were all over the map in results than an average would be unreliable.

 

[lilnev]

(total hammer cost of existing units, buildings, and Wonders, plus hammers invested in queue items) + 100*(number of cities minus one) + 1.2*(food equivalent of all cities) + 0.5*(beaker value of techs known + beakers towards other techs + cash on hand) + 500*(total great people)

By "food equivalent" of a city I mean all the food required to grow it to its present size plus the current contents of the food bar. The 100*(cities - 1) represents the hammer cost of the settlers. "Total great people" is fractional e.g. two plus 150/300 gpps towards the third would be 2.5.

Feel free to argue for different constants of proportionality. Mainly I'm proposing a functional form here.

It does leave some things out -- forests saved for post-Math vs. spent already, worker-turns spent roading or pre-chopping, turns of anarchy switching to slavery that will have to be done eventually anyway, map information, unit promotions, etc.

Also, you'd want to know both the above measure and it's time derivative -- what's the accumulated output of your empire to date, and what's its current output per turn.

I would suggest that if you tracked a measure like this and fitted it with an exponential function, the resulting time constant would tell you something interesting about your fundamental growth rate in that game.

 

[Jtownsend]

Lilnev: In keeping the figured relatively unpacked and unweighted, as Hammers/Beakers/Gold/GPP/Surplus food what I'm seeking to do is provide a realtime yardstick. An all-in-one weighted performance guide certainly has value, but is maybe closer to what the game scoring does at the end: with an idea of what an 'average winner's' H/B/G/GPP/SF is at 1500bc and 500bc and 0, one can say "Ok, well, I have a little less production than this target I'd like to meet or exceed, but my GPP and beakers are both higher," and decide, accordingly, whether or not to take particular steps out of the usual routine; perhaps decide which way to specialize an ambiguous city site to make up a production deficit, or if hammers are overperforming and everything else is slack, go to war?

I mean there are a lot of good ways to get better at Civ. It just might be interesting to have a set of figures to nail to the front door and say: "Here. On Emperor/Balanced/Normal/6 randoms, hit these figures. The ones you don't hit will tell you something."

Or, that idea might fall down somewhere, or just lack unique appeal.

 

[vale]

Any metric that is based in your own empire resources only ( like if you were playing in a sandbox game ) is faded to not be representative. You're in the game to win vs the other civs and as long as you can keep competitive or even with a upper hand ( no matter how ), you'll be ok. It is better to bankrupt and having units striking, but with all the AI cripled to oblivion, than having a very solid empire in economical terms but with a gigantic AI or 2 with the double of everything you have.....

This is exactly correct. Context is everything. Any metric must somehow consider the state of the player's civilization relative to his AI rivals. As such, any metric will be purely academic for most of the game as full information about a rival's state could only be obtained by entering world builder.

 

[Jtownsend]

Context is everything. Any metric must somehow consider the state of the player's civilization relative to his AI rivals. As such, any metric will be purely academic for most of the game as full information about a rival's state could only be obtained by entering world builder.

If you have a sampling of games, the empire-wide gross outputs would reflect those games in which interactions between players slowed development down.

Surely the size of the sample needed for a practical average is the question, not whether an average is possible at all? Also, I'm thinking of a benchmark which is made up of players very comfortable with either monarch or emperor, depending on which level would work better for such a test. As such the numbers would probably be reflecting games that are won - that is, they'd be figures high enough to win, without needing relative intelligence from worldbuilder to see if they have any bearing on your game.

I can see what you're getting at - you don't build another library to meet the 500BC beaker benchmark if Monty is next door - but I'm not aiming for perfection.

 

[CivCorpse]

Due to the practically unlimited AI/Map variations, there is no real "yard stick" for Civ. What was good enough in one game may be woefully behind in another. I have found that beakers (however you come about them) and hammers (either through slavery or production) are the two key benchmarks to determine how your civ stacks up vs the AI. Not just what you can currently produce, but also what you can plan on producing in the forseeable future. So it comes down to "Land is Power". Usable land. Great big worthless tundra/desert empires are not that powerful. How you use that land is up to play style and the situation you are facing.
If you are unsure where you stand take a look at research times displayed for the AI. Can they research a tech faster than you can? Then your beaker production is probably lower. If you are a bulb and trade sort of player then this may not concern you as much as a self tech player. Or you might be running an EE.
Production can be measured fairly easily by running about in the enemies lands and counting units and watching their city populations to see if they are whipping. Looking at the demographics will show you where you rank but not who is in front of you. Having a general idea of how many units vs buildings each particular leader favors can give you a better idea. Happy little builders like Ghandi won't have as many troops but you can get an idea of how many hammers are being invested in buildings by the number of troops. If he has a pile of them then you can reasonably expect him to have the majority of the importnat buildings. Nutjobs like Monty are easy. How many hammers worth of troops does he have? That number plus the cost of a barracks is pretty much the total sum of his hammer production.

 

[r_rolo1]

You might have a average, but I'm pretty skeptical of that number being useful. Just consider this situation:

You have a empire of 6 coastal cities ( in a penisula , for example sake ). You made the Great lighthouse. Now consider the diference that you would have between a diplo situation where no one would OB with you ( bad luck with religion spread or somethingl ike that ) and one where you can OB with everyone, allowing all your trade routes to be foreign. That can be a diference of 24 raw coins/turn easily .... but you can't make diference of the two situations based on tile output only.

Other situations: simply directing the teching to a tech that can easily be brokered vs a tech everyone has, shrine of high spread religion vs shrine of low spread religion, presence/absence of AP religion, the AI you have in game, the diplo stances ......

There are simply too many variables that have nothing or little to do with your worked tiles in terms of the "health" of your empire for any agregate based solely on that to be representative.

 

[Jtownsend]

Ahh. Fair enough.

 

[TheMeInTeam]

Just use score . It's as good as anything else you're going to come up with. Score is mostly a function of land, pop, and tech. Mostly. Point is, those things usually do translate to a winning position.

Yes, on high levels it is common to overcome huge AI score leads, but that isn't the point, because frequently the player gets top score or close before winning. Don't forget, however, that score is a metric of a static situation. If the player were to swap positions with the top-scoring AI around 1000 AD, the game would likely be much EASIER, not harder.

Granted, it's possible to win with lower scores, but that is ALSO true of any metric usable. Idiot AIs will get the wrong techs first in space races, and really puts no effort into AI diplo manipulation (the game would be kind of impossible if it did this).

In conclusion, aside from answering "how likely am I to be able to win from this position?", the best indicator of your empire's potential at a given time is score. Whether that potential is met or not depends on the player or AI .

 

[Ormur]

I can only imagine something like the hammers, science, espionage, gold, culture, food (perhaps redundant) and GP's in some acceptable ratio of importance in comparison to the same for other civs dependent on how many they are, or perhaps the average of them combined.

Then you'd have to determine how one beaker per turn compares to one hammer/gold/culture/EP/GP.

It could still be bonkers though. Maybe you are dead last but everybody loves you so you win a diplomatic victory or a culture victory.