Research Cost Efficiency

Discuss Research and Social Aspects of the game here.

Moderators: Balthagor, Legend, BattleGoat, Moderators

Post Reply
Eric Larsen
Colonel
Posts: 350
Joined: Oct 25 2005
Location: Salinas, CA

Research Cost Efficiency

Post by Eric Larsen »

I was playing the US in the World scenario when it dawned on me that I wasn't getting a very good research cost efficiency. I was only researching two items, augmented reality (more like depleted reality in boost) and advanced construction materials, when I realized that considering the $1.4 trillion I was spending on research at 33% for techs should have gotten the projects researched really quick rather than 60 and 105 days respectively. At $1.4 trillion times 33% that's $462 billion per year in money devoted solely to techs. At a combined cost of $12.5 that should mean the techs should have been researched in about 37 days total for both.

I then ran Southern California in the US California scenario and only researched augmented reality at a cost of $4,000M. On the second day I had $1,224,856M devoted to research and 33% for techs at 94% research efficiency. That's a whopping $404,202M per year or $1,107M per day. The augmented reality would take 44 more days to finish researching, or $90.91M per day of it's cost would be researched each day. That's a mere 8% of the $1,107 devoted to researching techs each day. Extremely cost inefficient.

So I thought that it must be that it's very cost inefficient to research just one item so I decided to further test that theory. I went back to my US California scenario and added 7 research projects to fill the queue expecting to see a better cost efficiency. So much for theory.

On day two I was spending $13,911 for research and 33% of that was devoted to techs for $4,590.63M per year or $12.58M per day. I researched water conservation at a cost of $450M. It showed it would take 114 more days to finish or $3.95M per day of it's cost. That yielded a cost efficiency of 31%.

I reran the same scenario with the same parameters except I added 7 projects to the list. Once again I had $13,911 at 33%, or $4,590.63 per year and $12.58M per day of investment. Here's a breakdown on the projects:
item...................cost......time.....days........progress/day
water cons........$450...3.1 yrs...1,132............. .4
dig soc serv........450...2.8 yrs...1,022............. .44
hydro plant.........500....568.........568............. .88
jet thrust vec....1,000...3.7 yrs...1,351............ .74
ERFB BB Ammo...450....3.1 yrs...1,132............ .4
Mil Bod Armor.....500....3.7 yrs...1,351............ .37
Nav Cat System..800....3.1 yrs...1,132............ .71
Research per day........................................$3.94M per day
Cost efficiency percentage is 31%

I had expected that there would be a better cost efficiency with more projects but was rather dismayed to see it stay statistically the same. What's even more interesting is that digital social services will get researched faster than water conservation or ERFB BB Ammo with the exact same costs. The Hydro Plant and Naval Catapult System had double or more better cost efficiency than the three others I mentioned and jet thrust vectoring was also up there. Military Body Armor fared poorer than any other project.

Why the differences in cost efficiency between items? Should they not all progress at the same rate? What's even more distressing is the poor cost efficiency in the World scenario, almost as if the more we get to spend the more the game somehow screws us. There's quite a difference between 8% and 31% cost efficiency between the two scenarios. I certainly hope BG fixes that so we can enjoy the World scenario's higher monetary scale and get the same results from our research regardless of scenario size or scale.
Thanks,

Eric Larsen
User avatar
tkobo
Supreme Ruler
Posts: 12397
Joined: Jun 04 2002
Location: In a vast zionist plot ...RIGHT BEHIND YOU ! Oh Noes !

Post by tkobo »

Speed at which a tech research is completed works on total cost ,cost per day, and total days(all combined and than capped).

Unless ive missed it,you dont figure out the cost per day of each tech,and use that to help determine your effeciency.

The reason this is important is that even if the total cost is the same, the total days might not be, hence the cost per day can also be different.

Also dont forget that the cost per day is also capped.You cant spend more than the cost per day each day on each tech.So at some point the extra money put into research techs is diverted automatically to research tech level and research effecieny
This post approved by Tkobo:Official Rabble Rouser of the United Yahoos
Chuckle TM
User avatar
Balthagor
Supreme Ruler
Posts: 22083
Joined: Jun 04 2002
Human: Yes
Location: BattleGoat Studios

Post by Balthagor »

It should also be mentioned that there is notable disagreement between the dev team on how it should all work...
Chris Latour
BattleGoat Studios
chris@battlegoat.com
Il Duce
Brigadier Gen.
Posts: 577
Joined: Aug 10 2005
Location: Venice - the Doge's palace on the Pacific.

Post by Il Duce »

Intuitive observations -
My experience at low numbers of infrastructure,
one or two facilities, is that at max investment I get better
performance researching 1 of 4 items, over 2 or 3.
This is based on some replay testing, and on the 'days to complete' stats, as well as actual achievement.

I suppose that this is pretty subjective, and also very heuristic -
items being developed were military units, not techs, and most of them
had relatively short [90day to 400day] dev times stated.

BTW - it seems like the 'top number' of days/cost seems to be based on
max funding and allocation of an entire facility to a specific item -the effective days to deliver [progress bar] seems to be a valid estimate of current funding and facility capacity.

Again, observances relate to unit development and not to tech development.

Tactically, I find that introducing new units one-at-a-time is more effective in terms of getting them built, too. I rarely have enough capacity to bring up a new brigade of [for instance] a new infantry and a new armor type all at once. Easier to develop the armor, get a regiment of them started to build, and then develop the infantry.

Overall, I suppose that my methods would be to have no more than one unit and one tech active per facility at a time [ or overall, for that matter]. In some regional maps, having just one faility and funding it fully is enough to keep you in the top three for research score. In these maps, every time I get anxious and add a research facility, I ususally end up deactivating it or destroying it, or just regretting it. I rarely play on the western hemisphere.

I am also very careful about the techs I build. In some cases I am willing to forgo all of the techs and units at a given tech level and just put all my funds into getting to the next level. Also, as a rule, I start out each game with research funding at max, and sustain it there forever. These are subjective techniques based on experience - I gave up trying to apply any numeric formulae to research items. Fuzzy math.
Colorless green ideas sleep furiously [but otherwise, they do not worry and are happy].
User avatar
tkobo
Supreme Ruler
Posts: 12397
Joined: Jun 04 2002
Location: In a vast zionist plot ...RIGHT BEHIND YOU ! Oh Noes !

Post by tkobo »

Balthagor wrote:It should also be mentioned that there is notable disagreement between the dev team on how it should all work...
Hhahahahhahahahahahahahahahhahahahahahahahahhahahaahh.
This post approved by Tkobo:Official Rabble Rouser of the United Yahoos
Chuckle TM
Eric Larsen
Colonel
Posts: 350
Joined: Oct 25 2005
Location: Salinas, CA

More Opinions

Post by Eric Larsen »

Balthagor wrote:It should also be mentioned that there is notable disagreement between the dev team on how it should all work...
Chris,
What would be really nice is if you could give us some clue how it all works in the first place. I'm really interested is seeing how research works down and dirty with the details. I'm curious about how to maximize efficiency as to how many projects to research at a time. In this lastest test I was trying to find the "sweet spot" as far as cost efficiency goes and how many projects to research to get this. I tried from 1 to the max and even in the middle and found little statistical difference. One thing ofr sure is that researching one thing at a time gets that one thing done quicker which gives you the benefit, or detriment, faster. I always wait until I've got one day left before I add another item to not lose any research time or investment.

I would have thought that somewhere in the middle should be the sweet spot. Pouring all your research into just one project at a time should be wasteful because you'd end up with lots of duplicated effort. Researching hte max number should also be just as wasteful as each project would be spread too thin. I keep thinking that researching somehwere in the middle should yield the most efficiency. Something like for every research center plus the freebie one should be the ideal number of projects to research. Or perhaps 1.5 items per research center plus the one freebie. That way you'd spread out your research nicely and give each project good investment. If I have 20 research centers then either 21 or 31 research projects should yield the best results as far as cost efficiency goes.

It really looks like someone stuck a speed governor on the World scenario for research cost efficiency. In my test I noticed that cost efficiency was about a quarter of that of the US California scenario. I went to the World scenario just because of the huge money scale and wanted to see research projects get finished really quick. After jacking up research investment to $2.5 trillion I'm seeing some quick completions when I just research one item at a time. I sure hope that the World scenario's research hasn't been dumbed down. but it sure looks like it.

So please have someone at BG give us some ideas about researching one item at a time or multiple items at a time and what is best overall. A little knowledge goes a long way to allay player's fears.
Thanks,

Eric Larsen
Eric Larsen
Colonel
Posts: 350
Joined: Oct 25 2005
Location: Salinas, CA

Still Missing the Point

Post by Eric Larsen »

tkobo wrote:Speed at which a tech research is completed works on total cost ,cost per day, and total days(all combined and than capped).

Unless ive missed it,you dont figure out the cost per day of each tech,and use that to help determine your effeciency.

The reason this is important is that even if the total cost is the same, the total days might not be, hence the cost per day can also be different.

Also dont forget that the cost per day is also capped.You cant spend more than the cost per day each day on each tech.So at some point the extra money put into research techs is diverted automatically to research tech level and research effecieny
tkobo,
You certainly should reread my first post as you've missed a very important point. That cap you mentioned is really something to be concerned about. First off I found no real statistical diiference between researching one project or multiple projects. The total cost of completed item research doesn't differ between one or many projects. All it means is that with more projects your research times increase quite a bit for each item with no real increase or decrease in cost efficiency.

When you look at the amount of money poured into researching techs compared to the progress you're actually making you can see that the cost really is far more than listed.

Here's a simple example for you. In the World scenario I'm currently playing as the US I've got 28 research centers (up 7 from the start) with a research budget of $2.5 trillion (rounded down) and a research efficiency of 123%. I'm getting ready to research facility defense which costs $4 billion. I can do it in 30 days if I research it alone which I will do. I have my tech research set to 33% so of that $2.5 trillion I have $833 billion devoted to researching techs. So in 30 days I will spend $27.78 billion in researching the tech while it only is supposed to cost $4 billion. The actual cost is almost 8 times as high as the stated cost.

That cap you mentioned is really a problem when you look at cost efficiency. I would not bet on the wasted capped tech amount getting reshuffled into tech level or efficiency. I bet it goes down the black programming hole where the savings from water conservation wrongly go.

I wasn't looking at research efficiency but at cost efficiency. I wanted to see how close the real cost of researching an item compared to the stated cost and in the world scenario there's a huge discrepancy. In the smaller ones the cap is smaller and you get a better cost efficiency, but not by a whole lot.

What you don't understand is there is no total cost difference between researching one item or multiple ones. The total cost of research being researched remains the same, or maybe is even a bit worse with multiple projects. The rest of that capped money just goes down the drain and you should be more concerned about that than you think.
Thanks,

Eric Larsen
User avatar
Balthagor
Supreme Ruler
Posts: 22083
Joined: Jun 04 2002
Human: Yes
Location: BattleGoat Studios

Re: More Opinions

Post by Balthagor »

Eric Larsen wrote:Chris,
What would be really nice is if you could give us some clue how it all works in the first place. ...So please have someone at BG give us some ideas about researching one item at a time or multiple items at a time and what is best overall. ...
Sadley, you're asking the wrong person. The implementation of the research model was done by George, only he could explain it.
Chris Latour
BattleGoat Studios
chris@battlegoat.com
User avatar
tkobo
Supreme Ruler
Posts: 12397
Joined: Jun 04 2002
Location: In a vast zionist plot ...RIGHT BEHIND YOU ! Oh Noes !

Re: Still Missing the Point

Post by tkobo »

Eric Larsen wrote:

tkobo,
You certainly should reread my first post as you've missed a very important point. That cap you mentioned is really something to be concerned about. First off I found no real statistical diiference between researching one project or multiple projects. The total cost of completed item research doesn't differ between one or many projects. All it means is that with more projects your research times increase quite a bit for each item with no real increase or decrease in cost efficiency.
The research of multiple items makes each tech take longer to research BUT gets more techs done in a given amount of time becuase there are more techs progressing at the same time.

I see no problem with this.

As for the " real increase or decrease in cost efficiency",there is a change in the allocation of funds.
The total over all may not change BUT how that money is spent among the three possible allocations does indeed change.
If the change complies with what you believe to be a "real" change or not is something thats based soley on your opinion. As you define the extent reguired to meet your definition of "real".


Eric Larsen wrote:
Here's a simple example for you. In the World scenario I'm currently playing as the US I've got 28 research centers (up 7 from the start) with a research budget of $2.5 trillion (rounded down) and a research efficiency of 123%. I'm getting ready to research facility defense which costs $4 billion. I can do it in 30 days if I research it alone which I will do. I have my tech research set to 33% so of that $2.5 trillion I have $833 billion devoted to researching techs. So in 30 days I will spend $27.78 billion in researching the tech while it only is supposed to cost $4 billion. The actual cost is almost 8 times as high as the stated cost.
NO. this is already explained why your wrong on this.All the money you allocate to tech research does NOT go to tech research IF the amount allocated is OVER the amount per day that the system allows you use on a particular tech.
Currently any money above the capped daily amount you are allowed to spend on a given tech is automatically redirected to research tech level and to research tech efficiency.

So the moment you allocate more money than the cap allows you are than also paying funds for tech effeciency AND tech level that is taken from the funds you allocated to research the given tech.


Eric Larsen wrote:
That cap you mentioned is really a problem when you look at cost efficiency. I would not bet on the wasted capped tech amount getting reshuffled into tech level or efficiency. I bet it goes down the black programming hole where the savings from water conservation wrongly go.

I dont have a problem with it.I have been told that the money is indeed rediricted. If YOU choose to disbeleive what the Devs have stated ,thats your issue.
While they may or may not be correct due to some bug or oversight,til i see tangible evidence to the contrary i will take them at their word.
Eric Larsen wrote: I wasn't looking at research efficiency but at cost efficiency. I wanted to see how close the real cost of researching an item compared to the stated cost and in the world scenario there's a huge discrepancy. In the smaller ones the cap is smaller and you get a better cost efficiency, but not by a whole lot..
Not sure what your saying here.If your saying you get less effect as you put more money in,maybe its just "law of diminishing returns".
But either way you not wieghing cost and effect correctly.
The reason your not, is that you keep only looking at tech researchs achived and dont for some reason include the other 2 places your money is going and the advances in them (tech level AND tech research effeciency)
Eric Larsen wrote: What you don't understand is there is no total cost difference between researching one item or multiple ones. The total cost of research being researched remains the same, or maybe is even a bit worse with multiple projects. The rest of that capped money just goes down the drain and you should be more concerned about that than you think.
First off there are differences, They are in the amount of money allowed to be used by each research tech/s,the amount of money redirected from the research tech/s to the 2 other funds using allocations,etc..

Youve been told where the extra money goes. Again, you simply choosing to disregard that explanation does NOT mean the system has an issue.

The problem is, you cant make an accurate accounting for how the money is used.You cant do this becuase you dont have the formulas to work with, nor is it shown in itemized account .

So Instead your trying to use "theory craft" to explain an issue you have with the system.
Which is unlikely to work.
This post approved by Tkobo:Official Rabble Rouser of the United Yahoos
Chuckle TM
szabfer
Captain
Posts: 126
Joined: Aug 24 2005

Post by szabfer »

Tkobo, are you sure that "The research of multiple items makes each tech take longer to research BUT gets more techs done in a given amount of time becuase there are more techs progressing at the same time." ?

As far as I saw, If I research 2 tech (with same cost) simultaneosly then their research time will be doubled. So for example, if one tech required 40 days alone then the research times will be the followings:

-1-
Researching one by one:
First tech finished at day 40, second tech at day 80.

-2-
Researching 2 techs parallel: both will be ready at day 80.

So it seems that case 1 have a little benefit, as you get the first tech sooner.

I remember that couple months ago we already discussed this with the devs. Did the patch changed something in that?
User avatar
tkobo
Supreme Ruler
Posts: 12397
Joined: Jun 04 2002
Location: In a vast zionist plot ...RIGHT BEHIND YOU ! Oh Noes !

Post by tkobo »

szabfer-
One of the devs tested this and put his results here on the forum a while back.
So unless the recent patches messed things up,yes.
This post approved by Tkobo:Official Rabble Rouser of the United Yahoos
Chuckle TM
szabfer
Captain
Posts: 126
Joined: Aug 24 2005

Post by szabfer »

Ok, I will dig in the recent posts...
User avatar
tkobo
Supreme Ruler
Posts: 12397
Joined: Jun 04 2002
Location: In a vast zionist plot ...RIGHT BEHIND YOU ! Oh Noes !

Post by tkobo »

szabfer-

you have to go back to pre 2 patch times i think for the qoutes.

I found and cut and pasted some in the test thread.


Eric Larsen-
you may have uncovered an issue with researching mulitple techs parralel.
This post approved by Tkobo:Official Rabble Rouser of the United Yahoos
Chuckle TM
User avatar
Balthagor
Supreme Ruler
Posts: 22083
Joined: Jun 04 2002
Human: Yes
Location: BattleGoat Studios

Post by Balthagor »

tkobo wrote:szabfer-
One of the devs tested this and put his results here on the forum a while back.
So unless the recent patches messed things up,yes.
I was the one who tested it and found that it was something like 2 techs simultaneously took ~ 900 days but the same two one after the other took about 500 days each. Simultaneous is more efficient, but ppl are usually short on patience. I know I am! ;)
Chris Latour
BattleGoat Studios
chris@battlegoat.com
szabfer
Captain
Posts: 126
Joined: Aug 24 2005

Post by szabfer »

Well... the main problem with it that the estimation system is a little bit unprecise (because too many factors involved), so to actually see the benefits of the parallel research requires 900 days :) It is very much realworld time to see - I want to play in that time :)

And look the other way: you get the first tech (with its effect) 400 days sooner when you research it alone. It is a HUGE difference and that make the whole parallel stuff unusable. The penality is too much...
Post Reply

Return to “Department of Interior - Research and Social Services”