How much should you buy rush?

[Shaka II]

Yes, ROI is a deep subject in Civ, and one that often doesn't have a simple answer. ROI must be evaluated by how a building, or a rush buy in this case, furthers your progress toward your ultimate goal or victory condition. And there are many paths and victory conditions. Some might even play just to have fun, not playing for any particular win condition.

But assuming that one is playing for victory and also trying to maximize their score, then time is a major focus, the sooner victory is achieved, the higher score you get, regardless the victory condition. Also, there are common elements for all victory conditions to occur, expecially in the early portions of the game, including growth, commerce, production, and culture, and from commerce, income and science. But as the game progresses, deviations develop. The expansionist may view science as a way to get a military tech early, using growth and commerce as means to boost the science spending to achieve the ultimate goal of military expansion.

The space science player sees science as the end objective by itself, though there may be empire expansion phases along the way. So he has a much longer term view of science, and the financial and science institutions required to fuel research all the way to space science.

So, the divergence from common development is sooner for the expansionist. Lately, I've not been using banks for that reason, and not too many universities either. But it depends on the game speed and map size, if you're playing for domination. if you can see the end of the game coming in 50 or 100 turns, then stop building infrastructure and focus on the end game. For me, it means building all units at that point, except maybe finishing a military building, a forge for production, or theaters and courthouses in captured cities.

If I was playing a huge map on marathon, I might have a very different view of universities and banks, since the game is likely to go to the end of the industrial era or early modern era, with a full ROI for universities and good ROI of banks in the key cities. I'll probably even do some rush buying under Universal Suffrage, until I switch to Police State for the end game. Especially now that Vel and others have taught us how powerful rush buying is.

[Hermann the Lombard]

Yikes! ROI can be tough enough to calculate for a capital investment like a Xerox machine for an office. Here you have wild variation from game to game and even from turn to turn. One of the milder ones would be calculating the point at which new units have a diminishing rate of return, as per the hint to not build 50 units when 15 will do.

[Shaka II]

I suppose one could calculate the ROI of military units by taking the final game score and dividing by the number of units to get points per unit, which is a nonlinear function of end date, especially when you win by domination. More units actually gives you a higher ROI, because you finish sooner, giving you a much higher score. Much/More = Big ROI

[Yosho]

Originally posted by couerdelion
I should make some comments in defense of ROI and “term to repayment” since I believe these are the very factors that will determine the optimal decisions for rush buying, pop-rushing or choice of build options. The important thing to recognise is that the ROI is a civilisation-wide calculation and not based at the micro-level of the city. Each city serves one sole purpose and this is to increase the research, wealth and military strength of the whole civilisation.

Having said that measuring the ROI or term to repayment of even something as simple as a granary is not easy and so we often rely on either gut-feel or trial and error. If we do make calculations, we still have to make some assumptions with the biggest one being the gold/hammer conversion rate. At least rush-buying gives us a benchmark 2:1 ratio which we can use as a benchmark so that by rushing we are implicitly stating that, for this build, that we value those particular hammers at more than 2 gold each.

Consider the humble worker. At 60 hammers, he can chop a forests in 4 turns for 30 hammers (or 7.5 hammers per turn). Conveniently we have a hammer cost -> hammer yield so we can easily work out.

ROI = 7.5/60 = 12.5%
Term to repayment = 100/12.5 = 8 turns

One of the best investments in the game

Of course, that's assuming the forest he chopped had no other value. And you also have to consider the unit maintance cost of the worker, and the way that building the worker slowed the population growth of your city, and...

[Velociryx]

I agree with what Yosho says. ROI can give you *some* information, but in many (most?) cases, I do not consider the information it can provide to be the most pertinant to making the decision of what to build. A useful rule of thumb in some cases, but its use begs the question "what IS ROI, precisely?"

Unfortunately, there are very few instances where a simple "if x, then y, else z" calculations apply to civ, and IMO, using a strict ROI-type measure for build orders can lead you to making what "the book" says are optimal decisions, but what "the game" says are not.

Again, IMO, the BIGGEST consideration of what to rush (and when) is expressed in answering the question "how many turns of turn advantage will this net me" not (necessarily) "what is the payback time on this investment"

If you could get to gunpowder 50 turns before the neighbor you plan to attack, does it matter what it costs, or does it matter that you will have a large enough window so that when the dust settles, there will BE no neighboring rival?

-=Vel=-

[Velociryx]

Just realized that there was more to say on the subject.

You see, the reason I feel as I do about ROI is this:

In order to make "apples to apples" comparisons about different TYPES of things you can build, your definition of ROI must be consistent across the board. That is to say, the same ROI formula you use to value the return on the bank you're contemplating has got to be the same formula you use when contemplating that Infantry unit.

Trouble is, when you start talking about ROI in general terms, you get one answer. When you start talking in terms of a particular building, you get another (and quite often, the formula changes again, depending on what SPECIFIC building you mean to build), and then, yet another answer when valuing units (and still another when valuing settlers/workers).

With so many different formulae for each different unit/building type, the comparison quickly becomes something other than "apples to apples" and thus, loses any hope of being keenly valuable when attempting to decide between two or more viable alternatives.

Then there is the matter of the game itself. In-game situations can skew the factors (however you define them) that MAKE UP ROI.

Example: If you are being pressed hard by four opponents, does it really matter that the "book valuation" tells you that you will receive the best return on investment by building an Observatory. This is an extreme case, to be sure, but I chose this example in particular to prove the point. Game situations that have nothing to do with the raw, mathematical valuation of hammers invested vs. time to repayment (or any variation of such a formula), no matter how complex, do not take into account the facts on the ground, which can tremendously skew the weights of the variables.

Just to scratch the surface here, another factor never given consideration is that of diminishing marginal utility, and the enormous impact it has on value calculations. I've never yet seen anyone's ROI calc that takes this into consideration, but consider the worker example above.

Now, no one will argue that workers are a POOR investment, but what if you have four cities and two hundred workers? Is the 201st worker as valuable as the first? Nope....but the forumla won't tell you that.

Same with ANY unit, really.

This is but one reason that I tend to shy away from such calculations to make decisions. For the ROI formula to accurately take all the variables into consideration, I'd need to have a specialized program running on another machine (probably at least as robust as the one running civ), devoted exclusively to running my numbers, and at that point (assuming it was possible at all), it'd start feeling suspiciously like work, which I like to avoid at all costs...

-=Vel=-

[Velociryx]

Dang it...I'm gonna blame the drugs for why I can't formulate complete thoughts today.

So you might be wondering...well, if not ROI or some derivation of it, then what the heck do you use????

Can't answer that question for everybody, but for me, it's all about Management by Objective.

Figure out what you want to accomplish, and figure out what you need to accomplish it. Damn the cost, and damn the obstacles. Build what you need to do what's in your head!

Hand in hand with this (and nicely topical) is the question of what/when/where to rush. In my book, ROI don't matter here either. The only thing that matters is that I rush stuff to meet my stated objective faster. If rushing a thing will get me to the goal a few turns faster, then it's worth it, because the goal I've outlined for myself is worth it. The goal I've outlined will either a) win the game outright, or b) is something I consider pivotal to winning outright.

Speed is life.

Turn advantage is speed.

That's all I consider.

Speed and Objective, and what it takes to get there.

-=Vel=-

[Hermann the Lombard]

Originally posted by Velociryx I'd need to have a specialized program running on another machine ... devoted exclusively to running my numbers

Horrors! Now you're running numbers!?!? I hope they don't catch you.

One of my questions above was whether it was really worth it to rush a bank, other than to clear a build queue. There are a couple of factors there: can I make better use of the build queue for the city, and can I make better use of the gold used to rush the bank.


BTW, welcome back, and I hope you're feeling a bit better!

[fed1943]

From what you said,ROI is just an analisys'tool.So it's not garanted useful everywhere.It seems nice to see hammers,commerce,gold,beakers(i.e.buildings).But not military(defend,attack,conquer,raze)too complicated to translate it in hammers or gold or anything.
Best regards,

[curtsibling]

Once I am powerful enough, I rush buy all the markets, grocers and banks in my new cities...

Gives a very nice cash boost quickly...

[couerdelion]

Originally posted by Velociryx
Hand in hand with this (and nicely topical) is the question of what/when/where to rush. In my book, ROI don't matter here either. The only thing that matters is that I rush stuff to meet my stated objective faster. If rushing a thing will get me to the goal a few turns faster, then it's worth it, because the goal I've outlined for myself is worth it. The goal I've outlined will either a) win the game outright, or b) is something I consider pivotal to winning outright.

-=Vel=-

Am I right then in assuming that, when you capture a shrine city, you do not rush banks, markets and grocers first but build forges, barracks and military units because you are still at war? Except in the case where you need units to defend the shrine, it would seem to be obvious that the bank is the best option despite the fact that the unit could help bring forward that next city capture.

I certainly do not deny that ROI is situational and that there are certainly limitations on the worker example.

MY view is that you can compared apples with apples if you can convert the pears you are making back into apples. Think of all the different conversion routes open to you when playing CIV

a) Between food, hammers and commerce through tile selection
b) Food to Hammers via Slavery
c) Gold to Hammers via Rush-buy
d) Hammers/Science/Culture/Gold via selection of specialists
e) Hammers -> Science/Gold/Commerce/Hammers via build order

Despite the changing circumstances of any given game, each of these will have a Civ dependent exchange rate which will vary with time. The real problem in knowing how to compare apples with pears is actually knowing what your exchange rate is.

I also think that ROI is resilient enough to the charges that it might tell you to build an observatory (this example) when that enemy stack is crying out for a rifleman. The reason why ROI is telling you to build an observatory (might as well be a conservatory) is that the calculation itself has not taken into account that little information that you have an enemy stack nearby. In the same way that stock prices move to reflect prices, so do the "value" of specific builds. Once you have the information that your city is under threat, the ROI of the extra defensive unit increases dramatically.

I am not saying that we have anything more than a basic idea of the mathematics behind all these calculations but this should not lead us to say that they are almost meaningless. I also suspect that the full mathematics will be too complex to apply fully within a game but it will at least give us the basic rules by which we can make decisions closer to the optimal.

As a final word, ROI is actually not the correct answer. The real comparison has to be in Net Present Value of the stream of benefits each build gives you. But sadly, my thinking along the NPV route is non-existent so all I have here is a general "hunch" that it will work but know idea how the calculations would be performed.

[Velociryx]

Howdy Lionheart! And please do not feel as though I'm in any way "taking you to task" for your beliefs. At the end of the day, if an ROI calculation of some kind or another provides you with a game-winning decision, then your decision making methodology has merit by default (if we assume that winning, by whatever path you choose, is the ultimate goal of the game...and even this may not always be a valid assumption).

My posts above were simply to point out that there is another way of arriving at those game-winning decisions, and that the alternative I propose has nothing whatsoever to do with math, and in fact, does not consider ROI at all, except for in a passing (incidental) sense.

The chief advantage to the alternative I proposed is that it is extremely fluid. During the course of the game, goals are set in mid-stream (sometimes by design, and sometimes reactive), and as such, the MBO approach responds in perfect sync WITH those changing, highly fluid plans, without the need to consult a spreadsheet, or do calculations, or otherwise break the flow of the game (which again, is starting to sound suspiciously like work, and that is a thing I seek to avoid! ). Of course, the disadvantage is that if you set a bad goal, then your build decisions will reflect that, and by extension, be bad themselves. Then the trick becomes to make sure that your goals are all good ones, but that is another subject.

To answer your questions about my game in particular, I will make one assumption about the scenario you propose, namely:

* This is, or at least has become, an offensive war (I have taken one or more enemy cities).

In this case, my answer would be that the newly conquered (shrine) city would get a forge first, CH to cut maintenance (if needed), border bump, and then financial improvements (this, depending on precisely how much gold we were talking per turn), with other infrastructure added on an as-needed basis (granary, LightHouse, etc)

No troops would be rushed from the newly conquered city.

This is because, when I am engaged in an offensive war, only my developed core is building troops (and then, not always all of them, if I have other matters to attend to).

Before the start of the war, I take steps (MBO, I know I'll be attacking) to have a sufficient body of troops that the war is effectively over before it is even begun, and the new captures can begin focusing immediately on the infrastructure they will need to put them on a paying basis (MBO, make newly acquired cities as productive and profitable as possible, ASAP).

Only in exceedingly rare cases am I surprised by, say, an unexpected stack of attackers which causes me to have to deviate from the frontier development, by rushing a defender. It happens now and again tho...perhaps one time per two fully played out games. A minor blip on the radar, in the greater scheme of things.

-=Vel=-

[couerdelion]

I see no reason for me to take any offense at what is essentially a different approach to reaching similar goals. MBO type approaches probably have more appeal because they avoid mathematical analyses but obviously rely on the gamer to make the right strategic decisions.

The one big problem with ROI or other analytical approaches is that these can either be made too complicated or do not reflect fully the information available. In reality what I am doing while playing is to occasionally check against a general strategy to see that the in-game decisions are all generally in tune with maximising my civ-wide output of science and gold (at least until the middle game). Even units can theoretically be seen as contributing to the basic output.

In shirt, ROI is a general tool that I use to compare investments out of game and applying this theoretical line to games played. Any in game calculations will be a simple hammer cost to per turn benefit comparison when the build queue is empty.

[Shaka II]

I'm going to upgrade my opinion of Universal Suffrage and rush buying based on my recent game. My previous complaint about it not being compatible with war seems to have been formulated from experiences in the modern era, with very large cities. Since I captured the Pyramids in France, I switched to Police State at first, but with 100 years to go, I switched to US to try it out.

The date is 1232AD with a small Roman empire and somewhat larger Indian empire to conquer to gain me domination victory. I already had ~43% of the land area and that proportion of world resources, so my people were really happy. So war weariness was never a problem, and I used rush buying, stopping science research completely, not even discovering banking. I spent the money instead on culture and rush buying military units for the remaining 50 turns. I rush bought the FP in Paris after it was half built, I couldn't force myself to use 7 of Paris's citizens for FP as it was the best city in the world, better than my own capital, Beijing. I rush bought theaters and courthouses, partly because I'm organized and have a personal liking to courthouses.

I built Grenadiers and cannons in about a 3:1 ratio, half produced, half rush bought, from that point forward, with a few Galleons to get several island cities. I was getting ready to take one of good friend Ghengis' cities, but my culture popped just enough to get me over 64%. He was a game long friend, so I hated to ruin a good friendship.

So, I retract my previous concerns, restating it as being dependent on WW sensitivity. In the modern era, war weariness seems more relevant, perhaps it is also dependent on other civic choices? But in the mid game, with lots of happiness resources it doesn't seem to be a problem. Of course Democracy is a bit off the tech path for the conqueror, but building the Pyramids, or better, capturing the Pyramids allows access to US.

It depends again on game speed, map size, when the game ends for the domination player, the last two games being 1332AD and 1336AD for me as Mao. Game time is only 10 hours or slightly less in this game. Curiously, marathon games on a standard pangea are much shorter than normal speed games which seem to take me 15 to 18 hours. The reason is that if the game goes into the modern era on normal speed, you end up building several hundred units instead of 50 to 80 units. It's counter intuitive, but that's the way domination games go. More turns is a shorter game.

Now, playing a huge marathon may put one into the modern era for the end game, so once again, war weariness is apt to be a bear, forcing one to switch to Police State.

As far as ROI for my rush bought units, well, there wasn't much else to build, and each one played some role in promoting victory. The ROI was only measurable in personal satisfaction.

[smbakeresq]

Well thought out discussion. The problem with ROI in the game is that the game is finite, therefore each game turn value is increased equivalent to the time compression as the game turns go by while at the same time the value of each improvement increases the earlier you build, when each turn is less valuable. Since you have no control over the value of each turn, turn value is the biasing element upon which decisions are built.

These discussions are key to understanding, playing, and enjoying the game.