Research efficiency

[Jorgen_CAB (OP)]

Research
I really would like research to be much less linear, this is not very realistic. I also feel that the number of labs per team (scientist) in general is too large. By the time you have enough labs to till your scientist to the maximum you will have enough scientists so it will never matter much in the end anyway. There is also a problem with research that give empire wide bonuses are too powerful in the hands of widespread large empires.

1. Make the number of labs available to scientists a new research path that start at one lab per scientist team.

2. Make every lab added to a team have a diminished effect, lets say 5-10% less RP per lab over the first. So if your RP per lab is 100 you get 100 for lab one, 90 for lab two and 81 for lab three and so on. This will represent that increased funding will never yield a linear result in efficiency which is realistic. This will also add the effect that you want to have as many scientists as possible but the skill of them is still very important.

3. Research with empire wide effect should in some way have a cost reduction/increase depending on the spread and size of said empire. For example if you want to research better industrial production it would be influenced by the number of worlds that have industrial complexes and in how many different systems and the number of sectors you have. It should also be important how many industrial complexes you have as a total. The same should be true for all the rest. Sure you could go with a system where each individual complex is upgraded (or a percentage system per planet would be enough) to make it more gradual, but that is more complex.

Specialization and increased overall efficiency due to improved technology that you are talking about here is handled in Aurora by the techs that improve research efficiency.

If you have the same level of technology, same level of equipment, then having 10 labs working on the same tech instead of 1 should not really be 10 times more effective.

CERN today has around 15000 employed (also including visiting scientists & engineers), compared to 1 million for a single research lab in Aurora. It is hard to imagine any situation where 1 million is not enough for even the largest research projects and experiments needed. Also remember that we already are very specialized due to the nature of Aurora ship building divided into components. Instead of researching an entire Carrier with everything on it you already divide it into dozens of projects. One research project can for example be researching a new missile engine component that will be used inside a missile, inside a launcher, inside a fighter, inside a carrier. It's hard to get more specialized then that.


It also seems you are mixing up alot of concepts here. Economy of Scale and Factory production then sure of course you are right that having one big assembly line instead of 10 small workshops employing the same amount is more effective.

But it doesn't work that way for Research, RnD and Product development. Every day you can see big companies buy up small upstarts with innovative technology and ideas, simply because they can't come up with those ideas themself, no matter if their RnD department is 1000 times larger. In a big company waste and as you put it C3 prevents you from seeing the big picture and making the big science breakthroughs.

If you ever see a big company buy a smaller rural workshop, it is because of their ideas, patents and innovative approaches, not because of their production assets.

I would be leery of assigning numbers too closely to how work is done. The in-game numbers might include significant portions of the rest of the supply/research chain, such as a copper refiner which gets copper to a superconductor manufacturer which gets superconducting elements to the engineers who actually build and set up everything. I don't know the source of your number but I doubt it accounts for such things, while the in-game workforce very well could (and reasonably so, considering you can pick them up and move it around it has to be mostly self-contained).

In any case, I would be especially hesitant to try and predict the future sizes of large research collaborations. I doubt anyone can predict future research project sizes now (although we are rapidly approaching Aurora's default start date), much less once we throw in trans-newtonian materials, jump technology or alien artifacts. Can anyone guess how many people it would to support a long-term manned expedition to study jump points, including planet-bound support staff, ship manufacturers, probe constructors, launch facilities, upkeep and researcher training? That's just to collect the data, much less analyze and utilize it. For comparison, the Apollo program employed about 400,000 people at its peak according to NASA, and that was in the 60s/70s without new TN materials involved.

If research is so much better with small firms, why do DuPont and 3M exist? Why is NIST around, or Argonne, or CERN? I don't think these monolithic companies/labs grow large and then completely ossify, feeding themselves solely through acquisitions? You should check out Bell Labs especially, a longtime research area owned by various companies, responsible for such modern items as transistors and CCDs plus fundamental physics like the fractional quantum Hall effect. It's a shame Alcatel-Lucent shut down their fundamental research division considering how much came out of there. There are plenty of other examples if you would like more.

Research is not a zero-sum game. The fact that companies acquire others does not mean the big company failed, only that they saw something they needed in the smaller lab. I have never stated that a small lab could not produce useful research. I merely stated that large labs are at least as good as small labs on average. I think there is a wealth of evidence and specific examples to support that assertion, including many projects that would be outright impossible without large teams of researchers.

*edit* I should add that it might be best to split a research discussion off so as not to clog the suggestion thread, especially if we keep up the debate.

Ok...

You still are mixing up the concepts here, technology being one of them as an enhancement in efficiency.

There is always a break even number of people/resources needed to actually perform any project. I'm more or less talking about anything we humans do in life. If you and your husband/wife make dinner will it go twice as fast if you do it together (even if you had twice the space and tools to do it)?

Any resources you throw at a project after that will have a diminishing effect.

Example

Koenigsegg is a Swedish luxury and hand crafted sports scar with a team of builders of about 50 people. They can hatch a new model with R&D and craftsmanship in about two years.

Now lets be extreme, but not impossible so...
Now lets say we grow that number of people and resources to x2000 which is about 100.000 people. Do you really think they could build the same handcrafted luxury car in about 8 hours?!?

Even if you only added another 150 people, could they make the same car (from the drawing board, R&D to full concept) in half a year?

This is just impossible... and lets remember we are not trying to make more cars, it is suppose to be rare.

In real life nothing can be increased in a linear fashion once you reached critical mass for performing something. Yes you can finish something faster but there is always a limit on how fast something can be done thus throwing more resources at is is wasteful.

In Aurora I would deem a single lab as being the critical mass, it is after all about 1.000.000 people involved in one way or another. Obviously only a fraction is actually top scientists.

You can never compare output from a centralized industrial complex in regard to producing "stuff" (as you previously did), this has nothing to do with human organization and intellectual innovation or human efficiency.


*** added ***

In my current campaign I use the following house rule for all my current nine factions...

1. Every research lab that is built will be assigned to a certain field and project as I wish. But once the lab is assigned to field it is dedicated to that specific field (Energy weapons, Propulsion & Power, etc..)

2. If I want to move labs from one field to another I must move two labs and destroy one.

3. Once a research project is finished all the lab devoted to it must be as equally divided as possible into two new projects. If there are no available scientists I must destroy 1/3 (rounded up) of one of the halves and add it to the new project I start up or distribute them to other projects in the same field.

4. At the start of each year I may take one lab from a project and transfer it to any other project within the same field.

This has for me added much more realistic advancement in technology over all and also institute some diminishing returns, at least if I want to change focus. I find it fun.

I do something very similar with industrial output.

[CheaterEater]

Ok...

You still are mixing up the concepts here, technology being one of them as an enhancement in efficiency.

There is always a break even number of people/resources needed to actually perform any project. I'm more or less talking about anything we humans do in life. If you and your husband/wife make dinner will it go twice as fast if you do it together (even if you had twice the space and tools to do it)?

Any resources you throw at a project after that will have a diminishing effect.

In real life nothing can be increased in a linear fashion once you reached critical mass for performing something. Yes you can finish something faster but there is always a limit on how fast something can be done thus throwing more resources at is is wasteful.

In Aurora I would deem a single lab as being the critical mass, it is after all about 1.000.000 people involved in one way or another. Obviously only a fraction is actually top scientists.

You can never compare output from a centralized industrial complex in regard to producing "stuff" (as you previously did), this has nothing to do with human organization and intellectual innovation or human efficiency.

I don't think I'm mixing up anything. I have shown several very specific examples where large organizations have succeeded in producing valuable research above and beyond what any smaller organization did. Isn't that the point here? What exactly am I mixing up? I am trying to compare large and small organizations of similar technology levels, is there a case where I haven't?

Be careful of using small examples, they are very easy to bend and twist to suit your needs. Here's one to illustrate: you can your friend need to move two different 200 lb objects. Each of you can only lift and move 100 lbs (discount dragging or machines, stick to the example). How much faster can the two of you move the objects if you work together, one at a time, versus splitting up and tackling each object individually?

There are many times where having more resources helps move things faster as you surpass the "minimum". If you can cover overhead easier with more people or a project is modular there are many areas where having experts work on each module or getting your work rate significantly faster than the overhead/upkeep rate can bring benefits. If it costs $1 million a day just to maintain a project, throwing $2 million at it will bring much faster results than just $1 million.

We can debate back and forth on what exactly the critical mass is. We really don't know, and for all intents and purposes anything we say is completely arbitrary. We have no experience with TN materials, alien life, large-scale spaceborne travel or system-traversing jumps. These projects could take millions of people, they could not. How about we focus on gameplay rather than realism, since we left realism behind some time ago? Sound reasonable? At worst Steve can decide the scale he wants.

We can most certainly combine large-scale production with research. We are talking about organizing large groups of people (or aliens) here, that's the core problem. Communication, workload sharing, time management, knowledge sharing, all of these reduce or increase efficiency. That's what we're trying to tease out here. Although they may have different end goals they share similar structural problems in reaching those goals, namely organizing their workforces effectively. They also share similar benefits to organizing into larger conglomerates.

Here seems to be the crux of the issue for you (correct me if I'm wrong): massive numbers of labs produces results far quicker than you feel they should. In many cases I agree with that; throwing 50 labs at a project should not go 50 times faster for a 1000 RP project; there's just no way to get everything together and working that quickly before the project is finished. My issue with your proposal is that it makes a very broad statement about research efficiency which I believe has not been shown historically; in many cases gathering more resources for a project has made it faster or possible. Is there some alternative(s) to resolve your issues while still encouraging more labs on projects that deserve them? Here are some:

Lab spool-up/spool-down time. A project could start get 1 lab a month working and, after the project is done, free up 1 lab a week (or whatever). If a project can only get so many labs "in-action" in a certain amount of time it discourages throwing large numbers of labs at a small project as many will not end up working. The spool-down time ensures that your labs are committed to the project. This still strongly encourages large numbers of labs for big projects as they will have time to spool up and operate at maximum efficiency.

Treat labs more like shipyards. There can be a cost associated with moving labs around or changing projects that encourages an appropriate number of labs scaled to the size of the project. Throwing large numbers of labs at small projects becomes cost-prohibitive.

Scale lab costs with the number of labs on a project. While they may not have reduced efficiency it may cost more to integrate and keep labs working together. This encourages smaller groups of labs, but utilizes wealth (which is very liquid) rather than research output (which is very difficult to modify quickly).

Reduce the number of labs available to researchers. As the first option listed, it would create a cap for the number of labs per project. Alternatively, make it a soft cap where more labs reduces efficiency if you go over the cap.

Those are some quick ideas. How do you feel about those? I think most will more strongly encourage scaling lab allocation to project size.

[Jorgen_CAB (OP)]

I did say that my example was a little silly...

But I think that we did reach a consensus regarding critical mass and anything you add after that will decline in efficiency, perhaps even rapidly in some cases. Critical mass is the resources you need to get the most bang for your buck sort to speak.

In my view one lab should be considered the critical mass, unless there were a specified critical mass needed that perhaps differ from project to project, but that I think is overly complicated and unnecessary.

Lab spool-up/spool-down time. A project could start get 1 lab a month working and, after the project is done, free up 1 lab a week (or whatever). If a project can only get so many labs "in-action" in a certain amount of time it discourages throwing large numbers of labs at a small project as many will not end up working. The spool-down time ensures that your labs are committed to the project. This still strongly encourages large numbers of labs for big projects as they will have time to spool up and operate at maximum efficiency.

Treat labs more like shipyards. There can be a cost associated with moving labs around or changing projects that encourages an appropriate number of labs scaled to the size of the project. Throwing large numbers of labs at small projects becomes cost-prohibitive.

Scale lab costs with the number of labs on a project. While they may not have reduced efficiency it may cost more to integrate and keep labs working together. This encourages smaller groups of labs, but utilizes wealth (which is very liquid) rather than research output (which is very difficult to modify quickly).

Reduce the number of labs available to researchers. As the first option listed, it would create a cap for the number of labs per project. Alternatively, make it a soft cap where more labs reduces efficiency if you go over the cap.

Those are some quick ideas. How do you feel about those? I think most will more strongly encourage scaling lab allocation to project size.

I like all of these suggestions, if not even better than what I said in my original proposal.

[CheaterEater]

OK, good, I'm glad we agree on some parts. I think having one lab as the "critical mass" leads to a less interesting game, however. Especially as tech levels climb up you must have tens of labs per project to finish it in time to design, research and build ships before your game succumbs to heat death/boredom. While a single lab might be reasonable at low levels, at higher levels it will take a decade for a research project. Managing a long-term research project sounds interesting, but just having it be a flat line down doesn't sound very exciting. At least to me, that is. I find it similar to the size 1 missile bonanza.

[NihilRex]

Perhaps the critical mass should be 1lab per 1000 RP needed for the research?

[Jorgen_CAB (OP)]

I think a better way would perhaps be that you need at least 1 lab per technology level. I think that each technology belong to a certain tech level of some sort. One lab for each 1000RP will get some trouble at higher levels when you need hundreds of thousands of research done for a single project.

So, that could roughly translate into something like.

500-3000RP you need 1 Lab at minimum to do any research
4000-9.000RP you need 2 Labs...
10.000-19.000 you need 3 labs...
20.000-39.000 you need 4 labs...
40.000-79.000 you need 5 labs... and so on...

Any effect on efficiency would scale with the same number of labs for each level.

Don't get one started on size 1 missiles and other smaller missiles, that is for another thread..

[wilddog5]

you could use the scientist's political reliability bonus to give a positive bonus(to represent them successfully lobbing for more funds/other positive reasons) or negatively (bureaucratic infighting ect) when the project starts to make the research non liner and give a risk / reward to spice things up.

[CheaterEater]

you could use the scientist's political reliability bonus to give a positive bonus(to represent them successfully lobbing for more funds/other positive reasons) or negatively (bureaucratic infighting ect) when the project starts to make the research non liner and give a risk / reward to spice things up.

I always assumed that the scientist's research bonus implied their organization and management skills rather than their personal research. The schmooze-skill could certainly be included though for reducing/increasing inefficiencies rather than total bonuses/penalties, I think it would be most appropriate there.

Overall I think treating labs like shipyards would be best even if it is more complicated. Especially when I look at Jorgen_CAB's house rules, it makes much more sense to make it difficult to move massive lab complexes around, break them up or combine them. Adding costs to reorganize labs would encourage smaller sets of labs that can offer some flexibility along with a larger lab complex or two for those big projects.

[Jorgen_CAB (OP)]

I liked your idea of having the labs take some time to start working. This would waste some resources if you add many labs to one project at any one time instead of moving them more often but in smaller quantities.

Although waiting for getting the labs back on each 5 day cycle is perhaps not practical but waiting for them to start working you will not have to do anything.

Lets say you have to wait 30 days for each lab to become available if it is applied within the same field and 60 days if applied to a different field. This should not be a major code change I believe but it would significantly alter the way you view technology progression overall. Then remove 5 days for each administration level of the scientist leading the project.

[alex_brunius]

Nice, this discussion moved ahead at lighting speed and with high intellectual level  

Many good ideas around like a set up time for research projects and other sort of diminishing returns or limits to how many can be assigned to different fields and moved around.

I like, more inertia to change all projects ( industry and research alike ) is great for making the model better and more realistic.

Lets say you have to wait 30 days for each lab to become available if it is applied within the same field and 60 days if applied to a different field. This should not be a major code change I believe but it would significantly alter the way you view technology progression overall. Then remove 5 days for each administration level of the scientist leading the project.

The code complexity comes when players start trying to find loopholes, for example start a new project with 1 lab, wait 60 days, then assign the remaining 19 labs...

I always assumed that the scientist's research bonus implied their organization and management skills rather than their personal research. The schmooze-skill could certainly be included though for reducing/increasing inefficiencies rather than total bonuses/penalties, I think it would be most appropriate there.

I agree that the scientists research bonus does not make much sense. And I currently think they hit the nation far to hard if the only leader with say skill 60% dies.

I would prefer to see a set of general values tied to the colony that change depending on your long term lab assignments. If I appoint for example half my labs to research within a single field and keep them there for a decade or two I should build up a large bonus/bias towards research in that area, but other areas with little or no labs assigned may have their bonus deteriorate.

Leaders could still provide a minor bonus/penalty (like the order of 10-20% maximum), more for flavour. The bulk of the know-how should be accumulated long term within the institution imho and not tied to a single leader.

[Jorgen_CAB (OP)]

The code complexity comes when players start trying to find loopholes, for example start a new project with 1 lab, wait 60 days, then assign the remaining 19 labs...

You would of course need to account for any strange and odd behaviours.

In this instance I'm not so sure what you mean, when you add those 19 labs each lab will take 60 days to activate and only ONE will activate on the same project each 60 days. That is the whole purpose I suppose.

What you could do is...

Project X had 10 labs and is now finished.

You now assign 1 lab to project Y and 1 lab to nine other projects. After 60 days you remove one lab from one project and add it to project Y, you repeat this after 60 days until project Y have all ten labs.

You could do this to gain a small benefit while the labs are gearing up to that project you really like.

It is not perfect, but it will be very tedious to min/max the system at least so most will not do it and I think it would be against the spirit of the system as well.

There probably is a better way to handle it, more like how naval yards is used where you build more permanent lab complexes devoted to a specific field. You can change field but it will be very expensive and time consuming if the complex is large. Smaller complexes is faster to change and can be used for smaller research project such as components. There can also be a small gearing cost between project change but much smaller than changing the complex from one field to another.

You could now also lower the bonuses from the team and let each complex add some smaller bonus the longer it remain in the same field... or some such. The administration level of a research team could also have an impact on how efficient they can use different sizes of complexes. A team with a low administration level will get a negative bonus from using a larger complex.

[alex_brunius]

You would of course need to account for any strange and odd behaviours.

In this instance I'm not so sure what you mean, when you add those 19 labs each lab will take 60 days to activate and only ONE will activate on the same project each 60 days. That is the whole purpose I suppose.

Ok then It makes sense, but that is very similar as adding diminishing returns I guess. Lab number 20 would not be effective until after 1200 days = over 3 years. Overall now that I think more about it I think just diminishing returns would be more simple to display.

[CheaterEater]

Ok then It makes sense, but that is very similar as adding diminishing returns I guess. Lab number 20 would not be effective until after 1200 days = over 3 years. Overall now that I think more about it I think just diminishing returns would be more simple to display.

You can change the time to get things up and running faster. If you assume most players will want a research project done within, say, 5-10 years, 30 days per lab wouldn't be bad at all for 20-30 labs for early to moderate tech. You could also add technologies to decrease that time or link it to the RP generation tech or even scientist skill.

[CharonJr]

I like the idea about diminishing returns as well as the project size being a factor for the optimal number of labs, but I think that more micro is not a good idea.

How about using a time target for each project at base research speed to determine the optimal number  of labs. E.g. a project would take 20 years wih 1 lab at 100rp each year with perfect lab utilisation = 2000rp. So a 8k project could use 4 labs at 100% with the 8th lab at 50%, the 12th at 25% and the 16th providing no additional benefit. Higher research techs and research boni would still provide additional research and each research tech level would provide the opportunity to set up a special research project with can use twice the number of labs.

[alex_brunius]

How about using a time target for each project at base research speed to determine the optimal number  of labs. E.g. a project would take 20 years wih 1 lab at 100rp each year with perfect lab utilisation = 2000rp. So a 8k project could use 4 labs at 100% with the 8th lab at 50%, the 12th at 25% and the 16th providing no additional benefit. Higher research techs and research boni would still provide additional research and each research tech level would provide the opportunity to set up a special research project with can use twice the number of labs.

Using time as the baseline for diminishing returns is actually really smart. Because it scales automatically with higher cost techs requiring more labs.

For example you could use 5-6 years as "No penalty" research cutoff and gradually introduce diminishing returns the more below this you want to reduce research time. Down to 2-3 years there would not be a big penalty, but it would grow quickly if you try to reduce research time down to a matter of months.

For racial techs there could be half time = 3 years for "no penalty" ( since they are less complex then base research normally ).