Home World Forum
Stars! AutoHost forums

Jump to Stars! AutoHost


 
Home » Stars! 2.6/7 » The Academy » Colony models
Re: Colony models Fri, 28 July 2006 03:10 Go to previous messageGo to next message
multilis is currently offline multilis

 
Lt. Commander

Messages: 789
Registered: October 2003
Location: Edmonton, Canada
"terra on the breeders at the moment they crowd, is big bang for the buck"

Small note that mine settings, HP rather than HG, how much tech increases will improve terraforming or how much time you have to work with (close war or long term nub era before war) all factor in.

For example the mineral starved are sometimes best to focus more on mines in many places so germ piles up rather than waiting for things to max out then being limited by germ supplies. This adjustment may help in begining with real game HW at 30% concentration when you testbedded with 60%.

[Updated on: Fri, 28 July 2006 03:14]

Report message to a moderator

Re: Colony models Fri, 28 July 2006 04:56 Go to previous messageGo to next message
Tomasoid is currently offline Tomasoid

 
Chief Warrant Officer 3

Messages: 182
Registered: December 2005
Location: Ukraine

Hi!

I agree with all of above. Of course, you are right that optimal growth should account empire-wide things. However, that kind of AI is quite hard to write without writing also a part with ship building management and transportation management, scouting, colonizing etc. Such kind of AI is equal or even more complex than re-writing the Stars! game itself, IMO.

Well, one planet economy optimizer is first step to wider AI Wink I plan to allow define the "strategic points" for planet throughout the years, where you would state that you want that terra is done quicker on some planet, set germanium and pop shipping for planet, set a point when you have better teraforming, reserve resources for starbase and ship building (with top or low priority) etc. I hope I would manage to make planet optimizer work with accounting all of these. This also would allow to produce the line of "demands" and "proposals" for a single planet (for example "need XXXX of iron to build starbase" or "I have XXX extra germanium") which later could be used for empire-wide AI (if ever get to it).



WBR, Vlad

Report message to a moderator

Re: Colony models Sat, 29 July 2006 04:48 Go to previous messageGo to next message
m.a@stars is currently offline m.a@stars

 
Commander

Messages: 2768
Registered: October 2004
Location: Third star to the left
JasonC wrote on Fri, 28 July 2006 06:47

The main point, though, is not to think about the problem on a planet by planet basis, but on an empire wide pop managing total econ basis.


I had something like that in mind when I spoke of "good enough" instead of "perfect". Whip Glad to see it so very extensively & well explained, too. Very Happy



So many Stars, so few Missiles!

In space no one can hear you scheme! Deal

Report message to a moderator

Re: Colony models Sat, 29 July 2006 18:53 Go to previous messageGo to next message
iztok is currently offline iztok

 
Commander

Messages: 1219
Registered: April 2003
Location: Slovenia, Europe
Hi!
Micha wrote on Fri, 28 July 2006 08:24

JasonC wrote on Fri, 28 July 2006 06:47

Jason Cawley
(Former serious stars! player...)


Wow ... Shocked If you are really *the* Jason Cawley welcome to the forum and welcome back to Stars! Nod

Hummm...
From his posts I'd say he knows perfectly what he's writing about.
What he's writing about is also very reasonable, and very JC style.
He also likes to write a lot. Wink
There's still no PBEM multiplayer game like Stars!, and some vets already returned to community. Thumbs Up Thumbs Up

Micha, IMO you're right. The JC has returned.

Welcome back! Smile

BR, Iztok

Report message to a moderator

Re: Colony models Sat, 29 July 2006 20:35 Go to previous messageGo to next message
AlexTheGreat is currently offline AlexTheGreat

 
Lieutenant

Messages: 661
Registered: May 2006
Location: Sydney, Australia
When I was completely new to Stars! I became quickly familiar with 5 people - the Jeffs, Ron Miller, Art Lathrop and Jason Cawley.

Art & Jason explained strategy to me. They were prodigious writers and, clearly, prodigious thinkers. These 2 guys had an unusual ability to state things very clearly & to separate the important from the crap.

I feel that I am in the presence of greatness!

AlexTheGreat

BTW Anyone know what happened to Art?

Report message to a moderator

Re: Colony models Sat, 29 July 2006 21:01 Go to previous messageGo to next message
Ashlyn is currently offline Ashlyn

 
Lt. Commander

Messages: 834
Registered: November 2002
Location: Pueblo CO USA
I haven't talked to Art for ages...

http://aaedesigns.com/stars/index.php

the Stars Directory looks to be updated last in 2002...
but i remember it well Cool

Report message to a moderator

Re: Colony models Mon, 31 July 2006 00:36 Go to previous messageGo to next message
JasonC is currently offline JasonC

 
Petty Officer 2nd Class
Stars! V.I.P

Messages: 58
Registered: July 2006
Location: Arlington, MA

Once more into the breech, my friends...

I understand you want to cop out and just model one world, but what the heck are you modeling when you model the return on terraforming one just one world?

What do you think terraforming is about? What are you buying for their 100 (70 for TT) resources?

Do you think you are buying a 2% increase in the factor applied to the pop growth rate, maybe 30 basis points a year more people? (Assume the delta hab is 2 - we can relax that below).

Nooo. Not the point.

You are buying *room*. The capacity of the world rises 20000 people (22000 for OBRM, JOATs extra, etc). Since races are all soon constrained by room, and typically can get 2-4 resources per 1000 colonists once the factories have caught the pop, that means you are raising *economic capacity* by 40-80 resources.

You don't get those extra capacity resources for free, you still have to wait for and manage the pop, and pay for the factories. But you can't get that capacity by putting up another factory, either.

And room constrains not on one world, but empire wide. Extra space is always useful, but the use it is put to varies. On a small, low-hab world, the change in pop growth from a few percent of terraforming is trivial. Does this mean the little ones aren't worth terraforming?

Noooo. The pop just comes in from elsewhere. Pop is not that little guy's strong suit. His strong suit is the rock underneath and ability to get at it with planetary mining. Plus the ability to set up factories that can't be added to a breeder without crowding it beyond the pop growth point.

On a small world, the terra trade is almost a straight capacity deal. You pay your 1000 points and you get 10 points of terra and the world jumps from 25 to 45. You can fly in 200,000 more people, and they will soon enough turn out ~600 resources.

That discounted present value calculation looks like this - spend 1000 resources of local currency over 2-10 years or whatever, then add 200 a year from externals (the pop), simmer at a low boil for 10-15 years (factories catch new pop), then collect your 600 a year return.

Meanwhile it sets up the following trade on the breeder. It could keep its 200k people (and perhaps ~1000 G as well) and let itself "go up". The decline in PGR could be finessed by "lifting". It still brings in 600 a year when all the factories are built, and it probably builds them faster. There is just one difference - the PGR is now gone. Crowding killed it.

So indirectly, buying capacity on the low hab world is keeping the pop growth in the "on" position on the breeders.

And that also has a return calc. The terra cost on the small world, the delay in putting up the factories on a half developed small world instead of a big ready breeder, and the travel time - those are the inputs. The output is +45k pop per year (by preventing breeder crowding while still using all breeder pop), turning soon enough into 135 resources per year *per year*. Plus any incremental PGR on the small, one-off. But that is the least of the impacts and not the reason to pay for the trade.

Now, the above I estimated the delta hab at 2% per point. It varies with the hab scheme, and this is the single largest reason for differences in terra procedures for different races. A two narrow one immune race sees a green go from 30s to 90s for 10-15 terras. It can easily get 5% jumps from one tick. A two wide one narrow race gets 2-3% jumps in the narrow field - the first few points typically - and then ~1% in the remainder. All wide no immune gets poor returns from terra, except in cases where one attribute is near an edge and the others are both good, when it acts similar to the usual one-narrow case.

They also have different "histories" in the room getting and pop getting division of labor. Two narrow immunes know that every green, even a 12 or 20 one initially, is a breeder in disguise. Eventually it will be perfect. They set their widths precisely to have their bre
...

Report message to a moderator

Re: Colony models Mon, 31 July 2006 05:32 Go to previous messageGo to next message
Dogthinkers is currently offline Dogthinkers

 
Commander

Messages: 1316
Registered: August 2003
Location: Hiding from Meklar
^^ that pretty well sums up why I rarely terraform non-breeders until they are at or close to 100% capacity, or I have no other use for the resources.

[Updated on: Mon, 31 July 2006 05:49]

Report message to a moderator

Re: Colony models Mon, 31 July 2006 06:58 Go to previous messageGo to next message
Tomasoid is currently offline Tomasoid

 
Chief Warrant Officer 3

Messages: 182
Registered: December 2005
Location: Ukraine

Hi!

It is really good point that decision whether we should terra first or build factories first is dictated by colony model we choose for it (breeder vs industry/poor hab).

Well, I was going to create a calculator that fills in production queue of the planet in an optimal way to reach some goal (usually the best final resources). Whether to import/export people or not, whether to build starbases or transport etc. is up to the user of the porgram - user can input all of that manually. There are still a lot of things to calculate past the empire-wide things, so I think such calculator might be still useful.

General question: is it worth at all to create such calculator for just a single planet???


About empire-wide things - I was going to leave this up to the user of the program - by allowing to specify pop export and import, transport building and starbase building etc. manually. So, for example, you would say that:
"when next year planet reaches 25% cap, set up teraforming and starbase building with most priority"
"when next year planet reaches 25% cap (accounting previous), set up transport building"
"export pop to leave 25% cap"
"at the year XXX import YYY pop"
ect.

For example, if we use such calculator for poor planet, user can say that "at the year XXX I will bring YYY pop". The program would try to optimize the production queue in such way that there is a room for additional YYY pop, and do this in an optimal way (so we have no too much room - we can spend resources elsewhere instead of extra terra).

Another example:
When going to build a strbase, and it could not be done in 1 year (lack of something), calculate what is the optimal distribution of building factories/mines/terra through the years for starbase building to get the optimal result, and still build starbase in time, or even quicker (quicker can be if build mines)?

This claculator was going to simplify and speed up things which otherwise you need to spend a lot of time to calculate manually (or using simpler calculators, like hab% caluclator etc.)

Well, after your words, it looks like it would be worth to automate things a bit.

So, for the best results, it looks like my program should allow in addition:

Allow set up the planet mode (breeder vs industry vs poor hab). Dependent on that, program would propose different economy model.

For breeder planet, it would try to schedule planet building queue to get pop export up and running as soon as possible (balance between teraforming, plus factories building for getting teraformed as soon as possible, and also teraformed up until it is profitable to terfaorm vs just export pop, or both if have remaining resources). Set up auto-export of pop (include transport building). Allow specify returning transport (manually).

For industry planet mode, try to get economy as best as possible. For this, we would import a lot of pop initially, then allow it build the industry. Program should allow clearly see how much pop we need to bring initially and when/how to teraform to have minimum loses in pop growth on that planet by overcrowding, and have the best economy at the same time.

For poor hab planet, try to get teraforming as quick as possible for having a room. Account how much of pop we can bring to it, so we can build factories/mines using remaining resources (and also get more resources for teraforming;) ).

When try all 3 of above for some planet (say 30% hab), it should be more or less clear what model is the best for a given planet.

(I'll try it out, though cannot guarantee that it would be in the best way. This is because need to account tech research, transport/starbase costs, miniaturization etc. I'll still leave this (building costs) up to the user entry.)

Note that I'll still build simple calculator first, then extend to have economy model switch Wink (It's hard to have everything right from start...)

Also, the progress: this weekend I got the idea of algorithm with complexity good enough to work in few secon
...




WBR, Vlad

Report message to a moderator

Re: Colony models Mon, 31 July 2006 20:48 Go to previous messageGo to next message
JasonC is currently offline JasonC

 
Petty Officer 2nd Class
Stars! V.I.P

Messages: 58
Registered: July 2006
Location: Arlington, MA

I think you want to let the optimization part depend on what the player is trying to do. What your code needs to do for him is show him the input output pairs that he doesn't know, because they involve too much intricate internal stuff.

Distinguish problem *parameters*, from inputs on the one hand and outputs on the other.

Parameters are - race settings for PGR and the factory screen stuff, habitat value, delta hab per click of terra, max hab (CAs just enter this as hab value), TT or no, g concentration. These need to vary in your code but they should be regarded as fixed for the input to output stuff.

Another possible parameter is a global "discount rate" to prefer present everything over future everything. But I think this is better dealt with in a different manner - see below.

Inputs are - initial pop, initial G, average pop per year imported during growth, average G per year imported during growth, and *time*.

Outputs are - the resource integral, ending resources per year, exported pop above a single fixed 50% hold for 70% and higher hab world, exported G, and spendable resources (resources not used by the local world).

You don't need to include base building or ships to move pop. They might not be made locally, the race might have ISB - too many variables. Just lump it all in the category "discretionary spending" or "spendable income". You don't need to treat these as an input, just assume spendable can always be found for it one way or another.

Now, when weighting things like pop sent, or resources used before the end, or the resource integral, the time value of stuff matters. And there may be trade offs among things like sending colonists or waiting longer, or sending G vs. getting spendable resources along the way.

I think the way to deal with all this is as another level of "valuation" after you have the input to output relations. The basic model just needs to say - this much stuff in, this much stuff out. How valuable that stuff is, is a different step.
And not one that one can really address intelligently until the inputs for outputs trades are known.

The conventional way you optimize among multiple goals once you have the whole input-output model is, you put prices on things. Assuming that G is worth 0.5 resources and the time value of resources is a 10% rate of discount and exported pop is worth (direct resources per pop)/rate of discount plus max resources per pop, and you have to spend 25% of ongoing resources on spendable stuff, the net value in and the net value out for this mix of inputs is ...

But that is the second stage. First just the inputs above for the outputs above, turned into a function. Then you can put prices on the inputs and outputs, and look for bargains in that function.

See the idea?

Report message to a moderator

Re: Colony models Mon, 31 July 2006 20:57 Go to previous messageGo to next message
JasonC is currently offline JasonC

 
Petty Officer 2nd Class
Stars! V.I.P

Messages: 58
Registered: July 2006
Location: Arlington, MA


Incidentally, on the side issue of calculating what to build along with a base or battleship that takes 2 or more years, I'll tell you my trick for it. It doesn't get a perfect answer but it gets darn close and it involves zero time with a calculator.

I just let the queue do all the work.

First, put in the base or BB. Line black, reads n years.

Now, put in factories directly *below* the base or BB, until the number of years increments by 1. Take out the last until the number of years again reads "n". (Note - if you were G limited this may mean 0 factories stay in, but that will be rare).

Now, bump the factories up to above the base or BB. They should be green.

Next, put in a mines line below the base or BB again, and repeat the "until the years increments" procedure, above. When the years are still n, that is you number of mines to build.

Now, bump the mines line up above the base or BB line.

Check that the base or BB line still reads "n years". (If for some reason it doesn't, back out factories - may be a G timing issue).

You are done.

When you are used to doing this, the whole thing takes only a couple of minutes and goes by rote, zero thought involved. And it will build a nearly optimal number of factories and mines in the period you work on the base or BB, without slowing the arrival of the main object.


Sincerely,


Jason Cawley

Report message to a moderator

Re: Colony models Tue, 01 August 2006 05:27 Go to previous messageGo to next message
Tomasoid is currently offline Tomasoid

 
Chief Warrant Officer 3

Messages: 182
Registered: December 2005
Location: Ukraine

Hi!

JasonC wrote on Tue, 01 August 2006 03:57


Incidentally, on the side issue of calculating what to build along with a base or battleship that takes 2 or more years, I'll tell you my trick for it. It doesn't get a perfect answer but it gets darn close and it involves zero time with a calculator.

I just let the queue do all the work.

First, put in the base or BB. Line black, reads n years.

Now, put in factories directly *below* the base or BB, until the number of years increments by 1. Take out the last until the number of years again reads "n". (Note - if you were G limited this may mean 0 factories stay in, but that will be rare).

Now, bump the factories up to above the base or BB. They should be green.

Next, put in a mines line below the base or BB again, and repeat the "until the years increments" procedure, above. When the years are still n, that is you number of mines to build.

Now, bump the mines line up above the base or BB line.

Check that the base or BB line still reads "n years". (If for some reason it doesn't, back out factories - may be a G timing issue).

You are done.

When you are used to doing this, the whole thing takes only a couple of minutes and goes by rote, zero thought involved. And it will build a nearly optimal number of factories and mines in the period you work on the base or BB, without slowing the arrival of the main object.



Well, that's for 2 years. If something is built in 3-4 years, you most likely need to ship minerals. What if no enough minerals? You probably can build more mines and get item built quicker. Now, question is, if we build mines for first and second year, would it improve? Answer to this question is by calculations only, because you cannot "split" a single item in queue to partial building actions. Add here factories, and calculations become repetitive up until you balance things between germanium spent for starbase abd factories and mines built. Manually it is long to do.

Second issue: assume now in your example that you have all factories/mines built on the planet up to what pop can handle, and pop is growing, as well as you still have some teraforming left to do. You need to build starbase, and see it is in 3 years. You schedule factories/mines. After some calculations, you see that you end with 60 leftover resources because no ironium for starbase. You can either:
- remove factories up to have 100 resources and build one teraforming click
- pile up extra mines/factories so next year you will build teraforming while new grown pop would reach the point when previously built extra mines could be used.

Then calculate what is better. I know the difference would be small, but for empire with 100 planets every little bit counts Wink Now, imagine calculation of above for 100 planets Wink

There are many other problems like that. For example, there are cases when you cannot ship germanium, and germanium is short. If mine cost is expensive, sometimes, instead of building factories till the last bit of germanium, it is worth to build few tens of mines right before the year when you get no remaining germanium. In the end, it may give better final output Wink Question: how many mines you need to build that way instead of factories to get maximum effectivity?

I know all above might be not very worth to do because it gives only small improvements. So, is it needed?



WBR, Vlad

Report message to a moderator

Re: Colony models Tue, 01 August 2006 05:35 Go to previous messageGo to next message
Tomasoid is currently offline Tomasoid

 
Chief Warrant Officer 3

Messages: 182
Registered: December 2005
Location: Ukraine

Hi!

JasonC wrote on Tue, 01 August 2006 03:48


I think you want to let the optimization part depend on what the player is trying to do. What your code needs to do for him is show him the input output pairs that he doesn't know, because they involve too much intricate internal stuff.

Distinguish problem *parameters*, from inputs on the one hand and outputs on the other.

Parameters are - race settings for PGR...

Inputs are - initial pop, initial G,...

See the idea?


Yup. You confirming already what I am doing, so I assume I'm on the right path Very Happy I already have parameters and inputs clearly separated. Not exacly like you proposed, but very close. Rolling Eyes

I have a bit better idea: the Excel spreadsheet way of calculations for planet is quite good. However, Excel gives little of way to program optimizations and is a bit too slow, so I'm writing a program. To simulate Excel-like style of work, I will have an output file that repeats all the inputs in the same format, and also includes spreadsheet for planet with all queue actions. Now you can correct it a bit (for example, shift starbase building 2 years earlier or later) and run program again with corrected output file used as input. I think this way user would be able to not only reserve how much resources could be spent on something, but also tell when it should be built. A separate input file will hold general information about the race and game parameters, so you can have such spreadsheet stored for several planets with no need to repeat the same race info in each.




WBR, Vlad

Report message to a moderator

Re: Colony models Tue, 01 August 2006 10:20 Go to previous messageGo to next message
LEit is currently offline LEit

 
Lt. Commander

Messages: 879
Registered: April 2003
Location: CT
Tomasoid wrote on Tue, 01 August 2006 05:27

Well, that's for 2 years. If something is built in 3-4 years, you most likely need to ship minerals. What if no enough minerals? You probably can build more mines and get item built quicker. Now, question is, if we build mines for first and second year, would it improve?

I often do this for more then 2 years, usually only when building bases fairly early in the game. You can balance the g need for factories and mines with just the queue.

Put base in top of queue, if resource limited add factories first, if mineral limited add mines. Adding additional factories after the base, and then moving them before will let you easily know how many to put before the base.

Also, building factories before the base gets more resources, so you have to recheck again once you move the factories before the base. Mines are a bit easier, but sometimes by building mines you can then build more factories (and often can build the factories first -- allowing you to build more mines)

If you are mineral limited and will ship in more minerals, and/or are resource limited and will ship in more pop, then the queue won't work.

Tomasoid wrote on Tue, 01 August 2006 05:27

If mine cost is expensive, sometimes, instead of building factories till the last bit of germanium, it is worth to build few tens of mines right before the year when you get no remaining germanium.

I'd noticed this when running the Humanoids in the demo before I had the full version. They often benefit by building some mines first. However, when mine costs drop to 3, that doesn't seem to apply. I think it was because a full year of building mines didn't allow many factories to be built the following year, however, building a few mines and letting them dig for 10 years works better. The lesson I learned from that is that without some odd setup, mines should cost 3.



- LEit

Report message to a moderator

Re: Colony models Tue, 01 August 2006 10:36 Go to previous messageGo to next message
Tomasoid is currently offline Tomasoid

 
Chief Warrant Officer 3

Messages: 182
Registered: December 2005
Location: Ukraine

LEit wrote on Tue, 01 August 2006 17:20


Tomasoid wrote on Tue, 01 August 2006 05:27

If mine cost is expensive, sometimes, instead of building factories till the last bit of germanium, it is worth to build few tens of mines right before the year when you get no remaining germanium.

I'd noticed this when running the Humanoids in the demo before I had the full version. They often benefit by building some mines first. However, when mine costs drop to 3, that doesn't seem to apply. I think it was because a full year of building mines didn't allow many factories to be built the following year, however, building a few mines and letting them dig for 10 years works better. The lesson I learned from that is that without some odd setup, mines should cost 3.


Well, not few mines first. My example is for building LOTs of mines at the year right before you are getting short of germanium after building factories from initial minerals. This may happen not only with costly mines. With mine cost 5 resources and factory cost 4 germanium this may happen too. The trick here is that you get mines work one year more, so you balance your resources in better manner between factories and mines building: resources you lose because not building factories year before you re-get by possibility to build more factories this and few next years.

Anyway, you're right, usual design rarely uses mine cost greater 5 and factory cost 4 germanium. That was just a simple example. There seems might be a lot of other balancing cases when add teraforming here, pop shipping etc.



WBR, Vlad

Report message to a moderator

Re: Colony models Tue, 01 August 2006 22:10 Go to previous messageGo to next message
JasonC is currently offline JasonC

 
Petty Officer 2nd Class
Stars! V.I.P

Messages: 58
Registered: July 2006
Location: Arlington, MA

My SOP for early mines was to put 90 in the Q before the G ran out, or before the year I needed to ship pop to keep the HW from crowding. Giving 100 mines working from the date shipping demand arrives. Yes this was meant to help with G. It is frequently just as important for iron, however, to get enough shipping to expand rapidly and move high pop growth off the HW without crowding.

This could be year 5 or so in Acc BBS start (depends on growth rate etc), and the factories were generally high enough that it took less than the full year's income. The temporary interruption of resources hitting the factory line piles up G, and so do the mines themselves of course. By year 20-25, all the factories the HW can operate at its 50% hold are typically done, and it becomes a G exporter. Every drop - shipping can live on the mine income.

Is 100 always a perfect optimum? No, but it is close enough. It handles the G crunch for shipping, the iron crunch (with help from occasionally looping freighters to a nearby rather than a farther world, bringing back iron from it, etc). It undoubtedly also eases the G to year 20 stuff. Occasionally with HPs that aren't as high in resources yet, I'll split the order over 2 years.

On the programming side of things, I don't use Excel and I don't have to write C code for this sort of thing. I just use Mathematica. Smile

Report message to a moderator

Re: Colony models Tue, 01 August 2006 23:40 Go to previous messageGo to next message
Dogthinkers is currently offline Dogthinkers

 
Commander

Messages: 1316
Registered: August 2003
Location: Hiding from Meklar
Hehe, I take the opposite approach to early micromanagent. I figure extreme micromanagement in the first 20 years is fairly 'cheap' in the context of the whole game - to spend the same hours on mm in later years affects only a small proportion of your empire, but in early years it affects 100% of your empire... So I usually micro my HW build queue to an extreme degree, in the hope of pulling that extra 1% of performance...

Typically I micro my HW mine building so I produce exactly the minerals I need, when I need them, and no more. The tricky bit is predicting the minimum mines needed to generate the minerals to cover all scouting/colonisation/booster/freighter needs - given some are one time only costs (scouts), some recycle quickly (boosters generally return quite soon after departure), some recyle (short/med range freighters), some are repeating costs (colonisers). It's complex but usually I rough it out by figuring I'll need to cover 2 freighters, 1 coloniser, 1 booster every year. Then I calculate when my mineral stocks will expire given this and early scout and extra booster builds. Then I figure out the last moment I can wait until to build my mines...

When playing a -f race, I use the same method, but instead of building the mines at the last minute, I build somewhat less, but right at the start (well, after flock of scouts (and I micro even that... I'll build as many scouts as I can in year one, then fill up queue with mines, then repeat the following year(s))


EDIT: BTW, I totally disagree with the observations that with increased mine costs it becomes better to build some mines earlier than later... I've got a spreadsheet I use for a colony simulation, and whenever I force it to build mines early the overall economic growth is hurt (as measured by total accumulated disposible income at year 2425.) The loss became greater the more I increased mine cost, not less.

Example:
17% growth, 15/8/16 4g facs, 10/15/10 mines
120,000 pop, start with 10 facs, 10 mines
Germ concentration 55, 300 surface germ.
Pop magically disappears at 33% cap

Build queue "Factories > Mines" for duration. At 2425 2820 res had been invested in research.

Build queue mines only in 2400, 2401, 2402, then "Factories > Mines". At 2425 2089 res had been invested in research.

Build queue mine only in 2405 (the year before we'd run out of germ) "Factories > Mines for every other year. At 2425 2733 res had been invested in research.

Change mine cost to "3". Results were: 10302, 9654, 10253.


So building mines early hurts, in all cases. The erlier you build them, the more it hurts. Building them just before you run out, still hurts, but not by much. Obviously this was purely from a econ perspective - no ships had to be built, but the same concept applies - the longer you leave it, the more factories you build. The more factories you build, the more resources you have to spend on mines...


Extra test - QS econ...
pop eff 1000, mines 3: 17704, 17355, ? (last test not performed - germ ran out at 2403, lol)
pop eff 1000, mines 15: 10877, 10331, ? (ditto)
Results much tighter, demonstrating that 1/1000 eff doesn't need to be so worried about optimising factory building.

The real decision is whether you are willing to sacrifice resources in exchange for making more shipbuilding minerals available earlier and in total. From a single planet econ perspective there's no reason to build mines early.

[Updated on: Wed, 02 August 2006 00:15]

Report message to a moderator

Re: Colony models Wed, 02 August 2006 01:45 Go to previous messageGo to next message
multilis is currently offline multilis

 
Lt. Commander

Messages: 789
Registered: October 2003
Location: Edmonton, Canada
Quote:

there's no reason to build mines early

In some (mineral tight) situations if you build mines at last minute, you will max out mines and still not have enough germ. In such cases you may get factories maxed out faster by flipping to mines sooner.

Report message to a moderator

Re: Colony models Wed, 02 August 2006 04:39 Go to previous messageGo to next message
Dogthinkers is currently offline Dogthinkers

 
Commander

Messages: 1316
Registered: August 2003
Location: Hiding from Meklar
multilis wrote on Wed, 02 August 2006 15:45

Quote:

there's no reason to build mines early

In some (mineral tight) situations if you build mines at last minute, you will max out mines and still not have enough germ. In such cases you may get factories maxed out faster by flipping to mines sooner.


ok.
dropped conc down to 20 instead of 55
1/1000, 3 cost mines, everything else as above
13490 (facs first), 13512 (3 years mines first), 13528 (mines the year before run out)
=== wow, you are right - better to build mines FIRST with low concs

1/1000, 15 cost mines
7442, 6992, 7340
=== but still better to build facs first if mines are expensive

1/2500, 3 cost
6923, 6530, 6871
=== hp eff wants to build the facs first still if mines are cheap

1/2500, 15 cost
1201, 810, 1120
=== still facs first.


OK, so if concs are poor, and mines are cheap relative to factories cost/efficiency then it becomes better to get mines up front. Thanks for making me look again Multilis, it's always nice to find exceptions to the rules you set yourself Wink

I think I might have to automate this to make it easy to find the breakpoints for any given race / concentration combo. Of course I'd normally be shipping germ to that germ poor world. I'll have to look at conc 30 for those nasty starts. But not now - gotta run for bus Wink

Report message to a moderator

Re: Colony models Wed, 02 August 2006 09:30 Go to previous messageGo to next message
Tomasoid is currently offline Tomasoid

 
Chief Warrant Officer 3

Messages: 182
Registered: December 2005
Location: Ukraine

Hi!

Dogthinkers wrote on Wed, 02 August 2006 06:40

EDIT: BTW, I totally disagree with the observations that with increased mine costs it becomes better to build some mines earlier than later... I've got a spreadsheet I use for a colony simulation, and whenever I force it to build mines early the overall economic growth is hurt (as measured by total accumulated disposible income at year 2425.) The loss became greater the more I increased mine cost, not less.



Embarassed
I should have said factories cost increase, not mines...
Sorry for that confusion. (What I was drinking before writing that?...)

So, the larger difference between factory and mine cost (factory more expensive and mine cheaper), it is more benficial to build some mines earlier before gemranium shortage. This is also more beneficial if factory costs 4 of gemranium instead of 3.




WBR, Vlad

Report message to a moderator

Re: Colony models Wed, 02 August 2006 09:54 Go to previous messageGo to next message
JasonC is currently offline JasonC

 
Petty Officer 2nd Class
Stars! V.I.P

Messages: 58
Registered: July 2006
Location: Arlington, MA

"Building them just before you run out, still hurts, but not by much"

You notice with mines cost 3 it made only 50 tech resources difference out of 10000, half a percent - and it solves the shipping problem, which is a vastly more important one than half a percent. A single turn of excess crowding on the HW will reduce achieved pop by more than half of of 1%.

I didn't arrive at my SOPs by flying by the seat of my pants. They aren't optimal in the strict mathematical sense of optimal, but they are *darn* close and a heck of a lot easier to remember and for a new player to learn.

Incidentally, I notice you plan a single 33% hold. 25%, 33%, and 50% holds actually vary in best use depending on the alternate hab available and average travel time, as well as representing a trade off in immediate spendable (for ships and tech e.g.) and future capacity (bigger econ at year 50 from higher achieved total pop). Which is better depends on the "discount rate", theoretically. In practice, early resources (thus higher holds) are worth more when you have to fight for space, and less when you can peacefully expand (thus lower to maximize pop by shoveling more out to the best greens, while there still is uncrowded high hab green space available).


Sincerely,


Jason Cawley

Report message to a moderator

Re: Colony models Wed, 02 August 2006 10:01 Go to previous messageGo to next message
JasonC is currently offline JasonC

 
Petty Officer 2nd Class
Stars! V.I.P

Messages: 58
Registered: July 2006
Location: Arlington, MA

The "mine drag" term in the rate of factory compounding is approximately -

PE = (factory cost + Sqrt( factory cost ^2 + 4*G/factory*minecost*fact eff/mine eff*100/Gconcentration)) / 2* fact eff.

With the rate of compounding 1/PE. Fact eff and mine eff are in units of 1.1, 1.2, 1.3, etc - rather than 11 12.

It is the solution of a quadratic, with the second term under the radical coming from G availability issues.

"Approximately", because it uses the same rate of discount for resources and G.

If mines cost 0 or you had an infinite Gcon, the second term under the Sqrt goes to zero, so that becomes the Sqrt of factory cost squared, thus just factory cost again. Add to the previous gives 2 factory cost, divided by 2 factory efficiency, gives the familiar eff/cost rate of return. (e.g. 12 fact eff / 8cost means 15% return).

You can see directly what the impact on the ROR is for things like changes in Gcon, the benefit of the G box etc. And compare them to things like factories cost 8 etc.


Sincerely,


Jason Cawley

Report message to a moderator

Re: Colony models Wed, 02 August 2006 10:05 Go to previous messageGo to next message
Tomasoid is currently offline Tomasoid

 
Chief Warrant Officer 3

Messages: 182
Registered: December 2005
Location: Ukraine

Hi!

JasonC wrote on Wed, 02 August 2006 16:54


Incidentally, I notice you plan a single 33% hold. 25%, 33%, and 50% holds actually vary in best use depending on the alternate hab available and average travel time, as well as representing a trade off in immediate spendable (for ships and tech e.g.) and future capacity (bigger econ at year 50 from higher achieved total pop). Which is better depends on the "discount rate", theoretically. In practice, early resources (thus higher holds) are worth more when you have to fight for space, and less when you can peacefully expand (thus lower to maximize pop by shoveling more out to the best greens, while there still is uncrowded high hab green space available).



That's right. I try to hold at 25% as much as possible until have space to spread out, or travelling is too far away.

However, just to add here:
For +f race, there is such factor as tech research. With 25% you get less free resources for early tech research than with 33% - 33% gives possibility to use more factories. That might be a reason why 33% cap is more oftenly used than 25% (as far as I have seen).



WBR, Vlad

Report message to a moderator

Re: Colony models Wed, 02 August 2006 21:08 Go to previous message
Dogthinkers is currently offline Dogthinkers

 
Commander

Messages: 1316
Registered: August 2003
Location: Hiding from Meklar
JasonC wrote on Wed, 02 August 2006 23:54


"Building them just before you run out, still hurts, but not by much"

You notice with mines cost 3 it made only 50 tech resources difference out of 10000, half a percent - and it solves the shipping problem, which is a vastly more important one than half a percent. A single turn of excess crowding on the HW will reduce achieved pop by more than half of of 1%.


I don't fovour this way of looking at it - it is arbitary which year I chose to consider. If you looked at resources at 2600 the % would be tiny... It is better to think of it at 50 resources lost, not 0.5%. In the particular example you are comparing with, that was 50 resources in exchange for getting about 75 mines a single year early, so about 40 iron and boranium extra (the model actually came out with about 10 LESS germ overall.) So the question becomes "is 40i and 40b worth 50 resources?" at that point in that particular game.

I made it quite clear in my post that I was arbitrarily ignoring costs of shipping pop out.

Quote:

I didn't arrive at my SOPs by flying by the seat of my pants. They aren't optimal in the strict mathematical sense of optimal, but they are *darn* close and a heck of a lot easier to remember and for a new player to learn.


Yes, they are certainly a good quick simple start. Bear in mind I join maybe 1 game every 6 months, so a little extra work at the start pays off well for me. It's really not much work for me - maybe half an hour or so added to the total time invested for the game.

Quote:

Incidentally, I notice you plan a single 33% hold. 25%, 33%, and 50% holds .....snip


No. I arbitrarily chose 33% for the purposes of the test. All I wanted to look at was the effect of building mines on econ on a single world, nothing more.

I've also spent time applying optimisation theory in various models of increasing complexing, to observe different strategies across multiple worlds with regards to maximising either population growth or the population integral over time. It became very quickly clear that even the 25/33/50 hold point strategy is inefficient. It's not bad and is usefull as a rule of thumb when managing large empires, but with relatively small empires (maybe through to 10 planets) you can pick up an extra couple of % with a little optimisation. Since I already have some hacked together models I can get a rough guide on what I should be doing in a matter of seconds, so it doesn't create a MM issue. But I think the OP was just talking about individual worlds, so I'll stop talking about this before I get further off topic Wink

[Updated on: Wed, 02 August 2006 21:14]

Report message to a moderator

Previous Topic: Armageddon Damage
Next Topic: Mid-game gateable missle defence reserve
Goto Forum:
  


Current Time: Mon Dec 02 06:28:33 GMT-5 2024