Resdesigning RCI

[GC13]

The level of abstraction in Victoria and EU isn't suitable for tLF because of the scale. In Victoria there are... hundreds if not thousands of pops, in EU hundreds if not thousands of countries (not to mention provinces). In tLF we have just a handful of races.  We also don't have access to manipulate many of the components that would provide those interactions. I believe the most promising area for complexity will be Racial Governments and Federation Governments.

Right now there are basically eight races and eight planets (making a maximum of 64 "POPs" if every race lived on every planet, but a more likely maximum of under twenty). However, I think the local populations need more attention. A race needs to try to thread the needle between current military strength, future ability to wage a long war, and the population's demands to have good living standards right now. There can be racial differences: Burlusts don't care so much about living standards, and care much more about current strength, etc...

Once you've started thinking about how the population affects their government's decisions in one way, you can then start to think about how they'd feel about the Federation; rather than being something that comes out of the woodwork, the AFA could become something you could influence, since there'd be something actually driving them to exist.

Everything is just too simple right now. I don't think this next expansion is going to add what needs to be added, but maybe the next one will deepen the game a bit.

[ptarth]

• NichG - If you would help out, it would be vastly easier to create a model we can manipulate. It can be based on the population model you previously wrote. What I'd want is the ability to calculate RCI for a given list of events for a given period of time (100 years by solar months) for X different models. The majority of interactivity would happen with the manipulation of the code, not with easily input parameters. I could do it all in R, but it wouldn't be easily interactive for others.
• Component List
• Event Driven: Define an event to be passed to RCI calculations. Each event would have: Event Name, Start month, one-time effect, recurring effect, end month, one-time end effect, recurring end effect
• Multiple Implemented Models. Given all events, it calculates RCI per month and then returns the result.
• Plots: Plotting of all models simultaneously of RCI by Month.
• Individual POP level mechanisms are probably unnecessary because there is only 1 dominant POP per planet. Any others are really, really trivial in size.
• This isn't to say that Racial level AI isn't vital. There also needs to be an integration of Planetary AI with Racial AI.
• After my campaign for an additive + multiplicative mechanism replacing the current multiplicative only mechanisms, Racial AI is my next largest concern.
• RCI - I'll grant you that Residential/Commercial/Industry is suspicious. However, the implementation is so different. In SimCity, they are just different current zoning demands, whereas in tLF it is a planetary status value.
• Once more complex RCI manipulations are available, the Racial AIs can be improved to use them. So development of RCI can take place independently of Racial AI and then integrated later.
• Drastic Events are useful to give a sort of rare disaster feel to the game. But they should be very rare. Perhaps 0-1 per game rare, not every 5 years as is current.
• tLF future - Well that's a loaded question. If we say, there is no chance, then... Looking at past history of Arcen development is suggestive. (Our hopes lie in Xenonauts, read the forum to understand that one.) Given all of this, then I would say that if I system could be fully designed, with all working parts characterized, then it is possible that Arcen could be convinced that it might be good. In the past they have taken player opinions into account. They haven't implemented massive cohesive changes (to my knowledge) based upon player input. They have taken player input and used it to create a completely new system (e.g., Federation Points). In contrast though, the majority of player comments are not well planned, documented, or setup in a way to be implemented. Arcen posts demonstrate a large commitment to improvement and a very reasonable and logical approach and structure. Thus, it seems very reasonable that a good concrete system with a solid presentation should be sufficient to facilitate useful change (baring any troubles with lawyers and IP). Thus far, this thread is a common example of what players do, they complain about something and then X players (NichG and me) argue about semantics for 3 pages. There isn't a cohesive, coherent plan to be seen. Just pages of minutia.
• Level of Abstraction - I agree. The level of abstract in EU/Vicky, isn't the same as what we need. We need to focus on drives at the Racial, Planetary, and Federation level instead.

[NichG]

• NichG - If you would help out, it would be vastly easier to create a model we can manipulate. It can be based on the population model you previously wrote. What I'd want is the ability to calculate RCI for a given list of events for a given period of time (100 years by solar months) for X different models. The majority of interactivity would happen with the manipulation of the code, not with easily input parameters. I could do it all in R, but it wouldn't be easily interactive for others.
• Component List
• Event Driven: Define an event to be passed to RCI calculations. Each event would have: Event Name, Start month, one-time effect, recurring effect, end month, one-time end effect, recurring end effect
• Multiple Implemented Models. Given all events, it calculates RCI per month and then returns the result.
• Plots: Plotting of all models simultaneously of RCI by Month.

I'll poke at this a bit, and maybe add a few elaborations. One thing I think would be useful would be that when we start talking about decision-making depending on RCI, it would be nice to actually visualize the decision tree and have the various branches 'light up' as they're chosen due to changes in the RCI. So for example, you'd have a flow-chart like the attached one (oversimplified for sake of doing it in 5 minutes for this post), and that would explicitly show you what the engine is thinking in order to determine the planet's self-generated events as you play with the model toy.

[Teal_Blue]

Quick question -

How many variables are brought into the decision tree vs. what is available in TLF right now?
Do these new decisions allow for more numerous race interactions?
Or do they change how the base race makes it's own decisions?
Does it provide bonuses to those races?
A means of fighting off bad events?
Does it buff the race in other areas? (provide more research? Provide more interactions with the race & Hydral?)
Does it affect other areas of the game other than RCI?
Why should the system that is in place be changed?
What benefits or advances does the new model have that we don't have now?
What problems do we have now that are addressed by this new system?


please excuse me guys, but even with all the interesting new systems designed to do x, i am having a hard time seeing what benefit this brings, what pluses are added and how this makes for a more varied or interesting game.

Can we talk about what the end-goals are and how they are better, before changing everything?

[GC13]

Individual POP level mechanisms are probably unnecessary because there is only 1 dominant POP per planet. Any others are really, really trivial in size.

How do you figure that? Take a single state in the USA, for instance, and you can easily come up with myriad ways to break them up. Even if we were to be strait-jacketed by Victoria 2's modeling of the nineteenth century, you could still break the POPs in TLF by ethnicity, job class (scientist, factory worker, white collar, technical, etc...), and literacy (if not true literacy then make it technical literacy, which would be crucial for the kinds of work the POP could be asked to perform—more advanced means of production, more advanced warships or weapons, better research all would require a more technically apt individual to do).

You shouldn't be able to just say "he's a Skylaxian and therefore he's the same as the other four billion Skylaxians". This is a game about influencing things, so we have to go deeper.

[ptarth]

Quick question -

Lots - How many variables are brought into the decision tree vs. what is available in TLF right now?
Yes - Do these new decisions allow for more numerous race interactions?
Yes - Or do they change how the base race makes it's own decisions?
Yes - Does it provide bonuses to those races?
Yes - A means of fighting off bad events?
Yes - Does it buff the race in other areas? (provide more research? Provide more interactions with the race & Hydral?)
Yes - Does it affect other areas of the game other than RCI?
It's too simple - Why should the system that is in place be changed?
It makes it more interesting - What benefits or advances does the new model have that we don't have now?
It's too simple - What problems do we have now that are addressed by this new system?


This is natural. The type, nature, and degree of changes necessary have not been specified. Specification of the changes is the immediate goal. - please excuse me guys, but even with all the interesting new systems designed to do x, i am having a hard time seeing what benefit this brings, what pluses are added and how this makes for a more varied or interesting game.

The end-goal is to make tLF more engaging. - Can we talk about what the end-goals are and how they are better, before changing everything?

To answer your unspoken question: I don't have a full system developed. I can't characterize it in great detail because I've not spent sufficient time on developing, programing, and then testing it. I've spent most of my time on arguing with NichG about math and advertising that there are other people interesting in working on this problem in the hopes of spreading the workload out.

GC13
We are having a semantic disagreement and a scale disagreement, not a difference in goals. In regards to semantic, perhaps unnecessary should have been replaced with not currently feasible to implement with the resources available. As you know, I'm a big proponent of increasing the game complexity, I just don't think that POP is the immediate best choice. In regards to scale, there is currently no framework to allow for POP level interactions, nor the framework to break down the populations into smaller POPs to have those interactions. There is a ton of work that must be done in planning and in implementation to make such a system feasible. Developing smaller mechanisms that would readily fit into a larger POP based system would be more immediately useful, and demonstrate that further such changes has merit.

This is Arcen's game. We can make suggestions, which they may consider. However, if all we do is provide nebulous one-liners or walls of text without useful instantiated forms, then what we do is just a pipe dream. We'd need a solid & explainable system, that could be broken up into straightforward implementable parts. In other words, we need a design and implementation document suitable for presentation to a management team.

[Teal_Blue]

Quick question -

Lots - How many variables are brought into the decision tree vs. what is available in TLF right now?
Yes - Do these new decisions allow for more numerous race interactions?
Yes - Or do they change how the base race makes it's own decisions?
Yes - Does it provide bonuses to those races?
Yes - A means of fighting off bad events?
Yes - Does it buff the race in other areas? (provide more research? Provide more interactions with the race & Hydral?)
Yes - Does it affect other areas of the game other than RCI?
It's too simple - Why should the system that is in place be changed?
It makes it more interesting - What benefits or advances does the new model have that we don't have now?
It's too simple - What problems do we have now that are addressed by this new system?


This is natural. The type, nature, and degree of changes necessary have not been specified. Specification of the changes is the immediate goal. - please excuse me guys, but even with all the interesting new systems designed to do x, i am having a hard time seeing what benefit this brings, what pluses are added and how this makes for a more varied or interesting game.

The end-goal is to make tLF more engaging. - Can we talk about what the end-goals are and how they are better, before changing everything?

To answer your unspoken question: I don't have a full system developed. I can't characterize it in great detail because I've not spent sufficient time on developing, programing, and then testing it. I've spent most of my time on arguing with NichG about math and advertising that there are other people interesting in working on this problem in the hopes of spreading the workload out.

GC13
We are having a semantic disagreement and a scale disagreement, not a difference in goals. In regards to semantic, perhaps unnecessary should have been replaced with not currently feasible to implement with the resources available. As you know, I'm a big proponent of increasing the game complexity, I just don't think that POP is the immediate best choice. In regards to scale, there is currently no framework to allow for POP level interactions, nor the framework to break down the populations into smaller POPs to have those interactions. There is a ton of work that must be done in planning and in implementation to make such a system feasible. Developing smaller mechanisms that would readily fit into a larger POP based system would be more immediately useful, and demonstrate that further such changes has merit.

This is Arcen's game. We can make suggestions, which they may consider. However, if all we do is provide nebulous one-liners or walls of text without useful instantiated forms, then what we do is just a pipe dream. We'd need a solid & explainable system, that could be broken up into straightforward implementable parts. In other words, we need a design and implementation document suitable for presentation to a management team.

I agree with the desireability of making changes show their benefit, and thus make for a more deep and interesting game. But so far everything is very nebulous and vague. This is not a criticism, as perhaps we have to look at the generalities before we get to specifics. But having those specifics would make others see the benefit much more quickly than discussing over-views first.

If bad event swings are a 'major' culprit of making the game seem out of control, then Misery's suggestion of reining that in may be the best thing we can do to immediately address that.  Decrease all swings by 10x is what i believe he proposed.

If the level of complexity seems shallow, then perhaps increasing the number of interactions available to the player to interact with the races, based on a sheer number basis for the moment would address that? from say 6 to 10 available in general to increase to say 20 to 50?

If the level of complexity in the interactions seem too straightforward (1 action equals 1 result), then perhaps increasing the 'angles' of interaction, would help? (2,3,4 or more actions per action equal 2,3, or 4 results per action?)

In this way,  Steal  becomes = influence another party, hire a thief, = payoff a race official, = do it ourselves... with several success and fail states, such as = succeed completely, = succeed but seen and therefor now at risk, = succeed but now starts a minor war with the hired party's race, = succeed but now have a general purge of the party in power to eliminate corrupt officials, = succeed and have the race come to the hydral and pay her to find the culprit?    Etc... etc...

What i personally think would help more than anything, it the specifics...
We don't have to have a ton of them.
We don't have to change the game into something completely different.
All we need to do is change some items here and there to make for a deeper and more interesting game for both beginners and for strategy pros that love the deep end of the pool.

[NichG]

Quick question -

How many variables are brought into the decision tree vs. what is available in TLF right now?
Do these new decisions allow for more numerous race interactions?
Or do they change how the base race makes it's own decisions?
Does it provide bonuses to those races?
A means of fighting off bad events?
Does it buff the race in other areas? (provide more research? Provide more interactions with the race & Hydral?)

For me, the most important thing out of the things you've mentioned is to change how the base race makes its own decisions, because that's the most likely place for there to be useful strategic feedback to the player. Think about how the Boarines play right now - ostensibly, if you muck about with their RCI, you can push them to be warlike or defensive or seek Solar Unity or whatever. I think that particular mechanism is a good preview for how it could be with the kinks worked out.

As far as bonuses, buffs, etc - those are numerical tweaks and there are already things like that in the game right now. Certainly it could be adjusted to be improved, but I don't think those are really at the center of something pitched as a 'redesign' of RCI from the ground up.

There's at least one thing I want to add to the list:

- RCI as a gating mechanism that determines what interactions are available (between the player and the race). This makes adjusting the RCI be a necessary strategic intermediate objective for achieving particular longer-term objectives, which enriches the strategic depth of the game.

Why should the system that is in place be changed?

I would say the main reasons are:

- RCI as it stands tends to run off to the extreme values, making it hard for the player to actually shift RCI once its been hit by major events (or just ignored for long enough). That means that if you tie things to RCI as it stands right now, it will feel more random and arbitrary than something that the player can actually seek to control. Part of the reason for the population discussion is that that kind of swinginess also happens with planetary populations (two diseases in a row completely wiping out a planet, etc).

- RCI right now does have effects on what the planets can do, but the effects tend to be somewhat understated compared to how much player effort is required to push RCI around, so its unclear that the player should worry about it at all from a strategic perspective rather than, say, focusing most of their efforts on influence or armada size or other things. This seems disappointing since RCI is one of the big long-term modifications a player can make to a planet (many of the buildings and techs purely influence RCI, dispatches, etc).

- Similarly, there isn't very clear causality between the player manipulating RCI and then later on having success or failure in their goals, because most of the effects that RCI has are perturbative - increase research speed a little bit, increase production, etc. When combined with other perturbative factors the influence of RCI can end up getting lost in the interactions of the behind-the-scenes model. Something where there's a more clear, quantized impact of RCI would be easier for players to understand and also construct strategic plans around, so this is something that I think can be improved in a way that makes the game deeper (in the sense of there being more distinct feasible strategies that players can come up with that are not all tweaks of the same basic strategy).

What benefits or advances does the new model have that we don't have now?
What problems do we have now that are addressed by this new system?

I kind of don't want to get too much into this right now because it led to the 'wall of text' style of post. Suffice to say, it seems like the current plan is for ptarth and myself to create some sort of mini-TLF simulation with the new models so that people can tweak them and actually test out the various gameplay ideas directly rather than argue back and forth about them on the basis of mathematical models alone, which can be somewhat unintuitive.

[Teal_Blue]

i don't mind your wall of text posts, and thank you for the reply. I hope i haven't discouraged ptarth or yourself, i find your discussions interesting, though again, over my head. I guess i just wanted to chime in and see if i could add something worthwhile, but to be honest, i think i'm going in another direction.

Sorry for all the questions, hope they didn't seem to attacking.

I don't mean to dis-rail you so i will leave it at that, but thank you for the reply.

[NichG]

i don't mind your wall of text posts, and thank you for the reply. I hope i haven't discouraged ptarth or yourself, i find your discussions interesting, though again, over my head. I guess i just wanted to chime in and see if i could add something worthwhile, but to be honest, i think i'm going in another direction.

Sorry for all the questions, hope they didn't seem to attacking.

I don't mean to dis-rail you so i will leave it at that, but thank you for the reply.

No problem at all, it didn't seem to be like attacking. I think if we can get more people involved in this discussion (and get it to a place where its more comprehensible) then that'll be good.

[ptarth]

If anyone wants to do something, but don't know what, I'd like some more measurements on RCI changes and a comprehensive list of effects that change RCI.

I did a check today with the Boarines. They had 2 Sport Complexes and 3 Universities, but the change in Economic RCI was only 0.32, not the .08 predicted. If someone wants to track that down, that would be great. I've already added it to the Mantis.

Here is the list of effects that alter RCI that I've gotten so far.

							Gain					Lost
Name	Type	Months	MinYear	Building	Cost	MaxNumber	Economic	Medical	Environmental	Public Order	Economic	Medical	Environmental	Public Order
Automated Personal Vehicles	Building	40	3036	Automated Transit System	5000 Terbium	5	0.008				-10			-10
Ecological Engineering	Building	40	3036	Ecology Station	5000 Molybdenum	5			0.008		-10	-20
Industrial Automation	Building	40	3036	Automated Factory	5000 Perovskite	5	0.008				-10			-10
Quantum Power Plants	Building	40	3015	Quantum Power Plants	8000 Uranium	1	0.264		0.264		-20		-20
Digital Sentience	Planet Improvement	40	3000				0.024			0.024
Gene Splicing	Planet Improvement	20	3018	Half negative medical trend
Mindreading	Planet Improvement	180	3009	Doubles poisitive public order
Nonlinear Mathematics	Planet Improvement	90	3018				0.016	0.016	0.016	0.016
Pacification Gas	Planet Improvement	300	3045	Halves rebellion events
Synthetic Fossil Fuels	Planet Improvement	40	3000	Halves negative environmental trends
Wireless Communications	Planet Improvement	20	3000	Halves negative public order trends
Anti Grav Suit/Personal Jet Packs	Safety	20	3012	Flooding Immunity			0.016
	Building			Exo-Zoological Park	8000 Molybdenum	1			0.016	0.008			-5	-5
	Building			Hotel and Casino Complex	5000 Xenotime	2	0.016				-2			-5
	Building			Sports Complex	5000 Perovskite	3	0.016				-1			-5
	Building			University	8000 Cesium	3	0.016	0.016	0.016					-10
	Building			Wireless Communications Infrastructure	8000 Perovskite	1	0.008				-5			-15
	Building			Teaching Hospital	8000 Cesium	4		0.016				-5	-5
	Building			Planetary History Museum	8000 Xenotime	1	0.016	0.016	0.016	0.016	-1			-5
Andor: Distribute Environmental Boosters/Entertainment Packages/Financial Aid/Medical Technicians							20	20	20	20
Evuck: Huge Tax Breaks							50	-40	-40
Evuck: Use Spy Probes to…							(probe +1)*-10	(probe +1)*-10	(probe +1)*-10	(probe +1)*-10
Impose Tariff Against Planet							X
Acutians: Dump Toxic Waste On Planet									-150
Acutians: Undermine Rival Economy							-150
Trend							0.032	0.032	0.032	0.032

   

                                                                                       


Alternatively, if "someone" wants to provide the data files with all of the tech, building, & other stats, that would make this easier too.

[Teal_Blue]

• NichG - If you would help out, it would be vastly easier to create a model we can manipulate. It can be based on the population model you previously wrote. What I'd want is the ability to calculate RCI for a given list of events for a given period of time (100 years by solar months) for X different models. The majority of interactivity would happen with the manipulation of the code, not with easily input parameters. I could do it all in R, but it wouldn't be easily interactive for others.
• Component List
• Event Driven: Define an event to be passed to RCI calculations. Each event would have: Event Name, Start month, one-time effect, recurring effect, end month, one-time end effect, recurring end effect
• Multiple Implemented Models. Given all events, it calculates RCI per month and then returns the result.
• Plots: Plotting of all models simultaneously of RCI by Month.

I'll poke at this a bit, and maybe add a few elaborations. One thing I think would be useful would be that when we start talking about decision-making depending on RCI, it would be nice to actually visualize the decision tree and have the various branches 'light up' as they're chosen due to changes in the RCI. So for example, you'd have a flow-chart like the attached one (oversimplified for sake of doing it in 5 minutes for this post), and that would explicitly show you what the engine is thinking in order to determine the planet's self-generated events as you play with the model toy.

I am of course not very good at this sort of thing, but wanted to ammend the flowchart you uploaded to include a parameter to try to bring back the parties at war into some semblence of peace, perhaps through some armistice mechanism which of course doesn't exist yet. The devs would have to add it. But it would allow at least from one end to the other then some 'basic' control, very macro and coarse, but perhaps useable? 




p.s.   Ptarth, i am not sure i can help with the chart you posted for the Boarines, not totally sure i understand it, but didn't want to just ignore it. just so you know.

[NichG]

I didn't actually mean for the flowchart to be exhaustive. Mostly I was trying to give an example of something I thought would be neat to be able to see in-game (with the various links lighting up as they're satisfied or not). It clearly needs to be a bit more thorough than what I posted 

Anyhow, there is a question of how I should design the simulator to be most accessible to people. My thought right now is to do something where there's a side panel that lists the various sorts of events or decisions that the planet can experience, and you can turn them on or off with check boxes. A given configuration of 'model + event list + action list' could be created as a tab in the display, and you could click 'Simulate' to gather time data on a random run of that combination (or even do something interactive where you can play the Hydral and do dispatches and such?). Then, you have the data of a full playthrough for each tab, and when you view the various graphs you could select to show them overlapping.

However, this doesn't necessarily make it easy for people to add their own events or models, so I need to figure out if there is a way for me to make that simpler or if its just something we have to live with.

The basic model of an event looks like:

- activateEffect: this is a function that runs when the event starts
- duringEffect: this is a function that runs once per month when the event is in play
- removeEffect: this is a function that runs when the event ends
- canTrigger: this is a function that returns 'true' if the event can happen right now, and 'false' if it cannot
- chancePerYear: this is the probability per year that the event will occur if canTrigger is true
- duration: this is how long the event lasts once triggered (set to -1 for permanent)

The basic model of an action that the planet itself can choose to take looks like:

- resourcesToComplete: this is the cost in some sort of resource track to perform this action. This could tie into a budget model, or be things like the mineral resources.
- minTimetoComplete: this is the minimum time to complete the action given that all the necessary resources are instantly available
- conditionFunc: this is like canTrigger, but it determines whether or not the planet can decide to perform this action given its current status
- weightFunc: this is a function which returns a floating point number that determines how much the planet wants to take this action
- duringFunc: this function is applied each month while the planet is taking this action
- interruptFunc: this function is applied when the action is interrupted
- finishFunc: this function is applied when the action completes

The idea was that the planet would perform an action and when finished, would consult its list of possible actions and choose the next action to perform based on RCI, resources, etc, using a weighted random selection. The planet then completes the next action and so on. Each action can also let itself be cancelled using 'duringFunc' to call 'interruptFunc' if the situation calls for it, but this decision is made on the action basis not on the global decision-making basis.

Goals:

In order to make the decision-making more coherent, there's also a list of hidden states that can be set by actions - 'Goals'. Essentially, you could make an action that says something like 'when Economy<-50, add 'repair Economy' to the Goals list and finish immediately'. You could then have an action 'when Economy>0, remove 'repair Economy' from the goals list'. Then you could have other actions that can trigger only if 'repair Economy' is on the goals list. That way, you don't just hover around -50, fixing the Economy RCI by 1 point whenever you dip below - instead, the planet can repair all the way up to zero before it stops pursuing that plan of action.

[ptarth]

re: Boarine & RCI.
It's pretty straightforward. The chart is a list of all the RCI modifying actions I could find (not including events, which should be added, but are troublesome).  When I attempted to predict how RCI will change per month, the numbers I calculated and the numbers the game produced were different. So I'd like someone to take a few more measurements to check to see if RCI growth is consistent with the listed growth. You'll have to add up all the modifiers from buildings, techs, events, etc, and then record the current RCI. Wait a month, record RCI. Repeat a few more times. Then calculate the difference in RCI per month and check that against what we expect to happen.

re:NichG
It seems that you are expanding from an RCI simulator to something closer to a Planetary AI simulator. That's a large scope change.

[NichG]

Yes, I am. I actually posted an RCI simulator JSFiddle a couple pages ago but I guess it got missed? Here's the link again: http://jsfiddle.net/QtQ37/

So I figured the next thing to do is to expand it so that the population model, RCI model, event model, etc are all interconnected so we can see how they influence eachother.