Announcement

**LDiCesare** · February 20, 2002, 05:28

What you propose is not LIKE the 51%, it is EXACTLY the median system. Mathematically speaking, you are just describing another algorithm to find the median (which would be very slow btw). The idea of scaling the power shown to the player is good.

**Mark_Everson** · February 20, 2002, 13:50

Hi Rodrigo:

A well-reasoned argument! I have problems with the things like rescaling power. The compromises part is really good, and could well be an important component of any voting-type system. What percentage the "moderates" are could be determined using govt model parameters, so I think that is potentially interesting. I don't have a lot of time because I'm at work. I want to think about your proposal a bit more. But in the meantime a modification of your system occured to me that would fix my problems with it. (Although it might introduce others)

My modified proposal:

There would be no rescaling of ruler power. The single difference I propose is that the rate of change of the policy or power be
1)handled potentially differently for each individual policy, and
2) be proportional to the power voting in the direction of the change.

Specifically if a vote were won 51-49 it would indicate a very delictate situation where movement of the policy would have to be done Very slowly. If All votes wanted to increase a policy, it could change very rapidly, because there would be unanimous support for the change by the governmental structure.

So even if the ruler has 51% of the power, if they are substantially going against what the other 49% wants, changes would be very slow. With 100% of the power, the dictator can change anything the want in almost no time. If the 51% ruler is going in the direction much of society supports change can be very rapid. I think this dynamic is similar to what happens in the real world.

If the current value of a policy was "low" by most votes, the change could be rapid until much of the govt thought it was about right, and then its rate of change could slow drastically if the final voted value was held only by a slim margin.

I think this modification is less artificial than altering powers, easier to understand in real-world terms, and it would give an interesting dynamic to govt changes. But it might be confusing to people. There might also be overshoot issues, but they could be handled relatively easily.

What do you think?

**roquijad** · February 20, 2002, 14:29

well, after initiating this thread I realized there's another problem with the proposed system. If polblock1 and polblock2 have each 25% power (and the ruler has the rest, 50%) then if polblocks want one same value for a policy, they can defeat the ruler because in the transformation to adjusted polpowers polblock's powers aren't changed while ruler's is adjusted to 39%. So the voting is:
value polblocks want: 25%+25%=50%
value ruler wants: 39%

And that doesn't feel right because for the player's eyes, he's equally powerful than his adversaries...

It's a problem, but I'm not sure how big it is. I mean, the problem appears in rather small range... In the same example, if polblocks had 20% each, while the ruler has 60%, the voting is:
value polblocks want: 20%+20%=40%
value ruler wants: 41%

here there's no problem.

I don't know. I have doubts now....

------------------
LDiCesare: "exactly" like the median system????

I'm not sure. Can you enlighten us?

------------------
Mark: It looks like the perfect solution!!! Let me think about it.

**Plutarck** · February 20, 2002, 14:39

Building on what Mark said, consider this as an application of the idea to slavery, but in more rounded numbers so I don't have to use a calculator (hehe):

Slavery is currently 10 on a scale of 100. The Meanies (Political Block) hold 45% power, the Weenies hold 30% power, and the Ruler holds 25% power.

The Meanies, being the meanies that they are, want Slavery to be 20. The Weanies want it to be 0, and the Ruler wants it to be 10.

Rather than being 3 votes, there are only two - Yay and Nay (For and Against).

Now, Meanies want to Increase Slavery with their 45% power, but 55% power (the combined of Weanies and Ruler, who both don't want slavery increased) says Nay. Because they do not have a majority (51%), they loose, and the rate of Slavery does not increase.

Now comes the question of whether to Lower Slavery. The proposal succeeds 55/45, and Slavery is lowered.

Now, as stated by Mark, the rate Slavery is lowered is determined upon the disparity of powers of the involved partys.

To do this, first we see what each of the "successul" partys want, and pick the most "conservative" one (this for simplicity sake. there could be a much more complicated way to doing it to figure compromise, but let's just see how this works first). In this case that's the Rulers desire, which is to have a Slavery of 10, which is change of -10.

Now the successful partys both agree on a decrease of -10, but they only have 55% of the available power to do it. 55% of 10 is 5.5, so Slavery is planned to be reduced by that amount.

This could happen over a variable amount of turns depending on the disparity of powers, but for the sake of simplicity we'll just say that change happens over 10 turns.

So policy is set every 10 turns.

Now each turn that would mean that the rate of Slavery decreases 0.55, but we're going to round Up the nearest tenth, giving us a 0.6 change each turn, leaving us with total of 6 reduction (there's multiple reasons for rounding up, and it's not merely for aesthetic or computational reasons).

After 10 turns Slavery now sits at 14, and the proposal of slavery comes up again. The successful party still wants a reduction in slavery, and all the powers are the same.

Now Slavery get's a planned reduction of 4, leaving us with 0.04 per turn, which get's rounded Up to 0.1. At the end of 10 turns the change is -1, leaving Slavery at 13.

This last scenario will happen 3 more times until the rate of Slavery now sits at 10 (assuming no powers or opinions change).

Now with a rate of Slavery of 10, the issue comes up again.

For Raising Slavery the vote will be 45/55 (proposal fails), and for Lowering Slavery the vote will be 30/70 (proposal also fails).

What this means is that there is simply no concencus, and no change will occur. This is because in no situation could anyone succeed, because no group can act with a majority of the power (in other words, anyone trying to act will fail because their adversary is superior to them in power).

So Slavery will now be set at 10, and will not occillate, until opinions or powers change.

In the next post I'll look at some other examples of what would happen under this system.

EDIT: Changed "70/30" to "30/70". Just an error fix.

**Plutarck** · February 20, 2002, 14:51

Ruler Power in a tie

Now, consider the following political makeup:

Liberatarians 25%, Anarchists 25%, Ruler 50%.

Issue: Ruler Power.

Proposal: Increase Ruler Power.

Result: 50/50, tie. For now we'll say that a tie is a failure, and in systems without some tie-breaker that could be how it works.

What could happen in case of tie is it goes to the vice-president (if there is one), who will vote his belief. If he's a Libertarian or Anarchist he'd vote Nay, if considering himself part of the Ruling class he'd vote Yay.

All kinds of possibilitys here, such as flipping a coin (random tie resolution), vice-presidential equivalent, vote of the people, goes to International Tribunal, etc.

But since we're in a system where there is no tie-breaker, the motion fails.

Proposal: Decrease Ruler Power:

Result: 50/50, tie. Same as above, which means the motion Fails.

Issue Result: No change to Ruler Power.

**axi** · February 20, 2002, 15:05

The median in a distribution of size N is the (N+1)/2 value (sorted in ascending order) if N is odd and the avg(N/2 , N/2+1) value (sorted in ascending order) if N is even. And it doesn't matter if we're counting from the low side or the high side either, since it's the same number.

If 25% want 0,2 / 45% want 0,6 / 30% want 0,0 then the median will be the average value of the 50th and 51st vote which are both ruler's votes meaning the policy will be set to 0,2.

We could always have stochastic values for each political block's desired policy. A discrete gaussian distribution could do the job, with a standard deviation that could be related to the climate of compromise and moderation existing in one political scene. It would simulate some political figures of the same alignment "reaching" towards the left while otheers are doing the same to the right. Also artificially polarising the political scene is a fine way for some parties to win the elections.

There would be no rescaling of ruler power. The single difference I propose is that the rate of change of the policy or power be
1)handled potentially differently for each individual policy, and
2) be proportional to the power voting in the direction of the change.

Specifically if a vote were won 51-49 it would indicate a very delictate situation where movement of the policy would have to be done Very slowly. If All votes wanted to increase a policy, it could change very rapidly, because there would be unanimous support for the change by the governmental structure.

So even if the ruler has 51% of the power, if they are substantially going against what the other 49% wants, changes would be very slow. With 100% of the power, the dictator can change anything the want in almost no time. If the 51% ruler is going in the direction much of society supports change can be very rapid. I think this dynamic is similar to what happens in the real world.

If the current value of a policy was "low" by most votes, the change could be rapid until much of the govt thought it was about right, and then its rate of change could slow drastically if the final voted value was held only by a slim margin.

I think this modification is less artificial than altering powers, easier to understand in real-world terms, and it would give an interesting dynamic to govt changes. But it might be confusing to people. There might also be overshoot issues, but they could be handled relatively easily.

I totally support Mark on this.

**Plutarck** · February 20, 2002, 15:48

Complex Political System

Note: In these systems the total power always adds up to 100%, though I guess it wouldn't Have to...but it probably always should. Otherwise people will probably get Really confused. It's mostly a sepperate issue though, so I won't consider it here.

Political Blocks:

Wimps 10%, Pansys 7%, Meanies 10%, Masochists 3%, Moderates 30%, Ruler 40%.

Issue: Slavery (starts at a rate of 20 of 100, which can be represented as .2 or 20%)

Proposal: Increase Slavery.

Desires: Wimps 0, Pansys 0, Meanies 50, Masochists 100, Moderates 20, Ruler 30.

Result: Yays- Meanies, Masochists, Ruler = 53%. Nays- Wimps, Pansys, Moderates = 47%.

The Yays have it, proposal succeeds.

The smallest desire among the Yays is the Ruler's desire for a +10 to slavery. Here Yays have 53% of the available power, so they can get 53% of what they want done. So that leaves us with 5.3, which becomes 0.6 a turn, which is 6 at the end of the policy session of 10 turns.

This will continue until 30 is reached, then when the vote happens:

Proposal: Increase Slavery (currently 30).

Change in power: Ruler decreases to 35% from 40%, Moderates increase from 30% to 35%.

Reult: Yays- Meanies, Masochists. Nays- Wimps, Pansys, Moderates, Ruler. Proposal Fails.

Proposal: Decrease Slavery.

Result: Yays- Wimps, Pansys, Moderates. Nays- Meanies, Masochists, Ruler. 52/48, Yays have it, proposal succeeds.

The most conservative desire is the Moderates with the desire of -10 Slavery, using their 52% power the result is 6 change during this policy period.

And on and on, etc.

I think this is about all that needs to be used as an example of the use, and I definately agree with it now. Sounds like it would make a great addition to the game.

Howeve, one thing that Is apparent to me is that there should be a more detailed compromise mechanism, where the powers of the opinion holders are taken into consideration to decide what the planned reduction will be. But I'm not talented enough with math for that, so I won't suggest anything.

Another thing is one small addendum/change:

What will be decided in the policy meeting is the planned changes the controlling decision makes. Such as "decrease slavery by 10". What change takes effect per turn is decided by that turn's power ratio.

So if on the first turn there was 100% agreement Slavery would be reduced by 1. On the next turn the power may have eroded so that there's only 90% power, and thus Slavery is reduced by 0.9. Then there's a falling out and those in power suddenly only have 40% power, and can only manage to reduce slavery by 0.4 for each turn thereafter.

So at the end of the policy session the change is 5.1, though 10 was originally planned. And of course now in the new policy session the proposal to decrease slavery will fail, because those in favor now only hold 40% of the power.

All in all, I see no problem with the aforementioned system, unless I'm just being dense here.

**Mark_Everson** · February 20, 2002, 20:36

Glad you guys liked the idea!

Plutarck, I indeed had in mind about something like the approach in your last post. Thanks for fleshing it out! I think the compromise suggested by Rodrigo in the original post is all that's needed, though I might be missing something. Is there something you don't think would work with it?

**LDiCesare** · February 21, 2002, 04:37

When I say it is exactly the median system, look at the figures, what the median gives, and behold: The final result after infinite iterations is the median system. So the only thing which changes, but is not clearly apparent, is how many iterations you do.
Once you know what the target value is, then instead of switching to that value, the suggestion is to switch to an intermediary value, based on the polpower of those who promote that change, like going (polpower of those who want the change)% in that direction.
If the amount depends on the amount of change wanted, Machiavelism will prevail in order to increase the correct amount. With Plutarck's example, in order to get 30, you would manage it by asking for more (around 39).
We can have the amount of change fixed, but it requires knowing a scale. If that scale is open-ended, like an amount of cash to be spent, then there is no way to avoid the player cheating with numbers in order to get the result they want.

**Mark_Everson** · February 21, 2002, 13:04

First, a comment on Plutarck's implementation that I endorsed last night. I wasn't paying enough attention!

The proportionality constant that gives the rate of change of the policy should not be based on the absolute power of those on the yea side. That gives at most a factor 2 difference between the 51% and 100% power cases. What I had in mind was that the rate would be proportional to the Difference between powers on the winning and losing side. The amount of plurality if you will. With 100% power, the change should be almost instantaneous. For a 51-49 decision the change would then be 50x Slower! ((100-0)/(51-49)) These factors can of course be modified, but I think there shoud be at least a 10x difference in rate of change between a close vote and a unanimous one.

If the ruler pursues a 51-49 victory, it may be a Phyrric one anyway. A lot of things can change in something like 50 turns, and there is absolutely no guarantee the value will Ever get to the state the player wants. So this system give a great deal of difference between the effectiveness of a 51% power Ruler and a 100% one, which was the big goal to remove from the median case.

Having said that... now I can address Laurent's points.

I agree that the steady-state results are the same as the median. Also I agree that there can be a small benefit, a factor of approximately two if the ruler votes 39 instead of 30. (at least with Plutarck's system as outlined with only my one modification) That is indeed a defect in the proposed system, thanks for pointing it out

.

I'm not sure it is a large enough defect to need to really worry about though. By overestimating his desire for slavery, the ruler will disappoint the Moderates more than would have been the case at 30, and that disappointment might have ramifications in terms of moderate support for the ruler that cancel out any advantage. In short, I think there are tradeoffs and perhaps we can just leave this approach simple for the time being. If there are clear exploits of it, I think we'll only find out about them for sure when the game gets in playtester's hands. And we can plug any Big holes then.

**Plutarck** · February 21, 2002, 15:10

Mark:

Actually, I was thinking the same thing about Disparity. I actually had it in my last post, but deleted it before posting it

The thing I was thinking is, would the changes go too slowly? But perhaps it won't be as bad as I thought at first...

Now that I think about it, I've swung back to thinking it should be the Disparity in power that determines it, but I don't know exactly how it should be computed.

For instance if you only have a 1% majority, you'd only get the minimum amount of change per turn of 0.1...but actually, that would equal 1 point of change every policy period. So to get a 10 change would take 10 policy periods, which is 100 turns. But at least it wouldn't be 0.01 point of change per turn...but then again, it could be, if you Really wanted to drive home the difficulty.

So yeah, I think this would work just fine as outlined.

What could be done is use a variable (or constant) to determine what the minimum amount of change per turn will be. Will it be 1, 0.1, 0.01, 0.001, etc (not counting No Change, which is possible). I like 0.1 as it would speed up things a bit, but for longer games I could see 0.01 being fine too. There could also be a max change per turn, but none is needed for now (good to go ahead and include that though, since it probably wouldn't add but a few lines of code).

And as for how many turns between policy meetings, that should be handled as a variable also, but I think I already mentioned that.

Now, here are a few problems/scenarios, and how they could be fixed/implemented:

1) Machiavelism: In the scenario, why shouldn't the Ruler just put down 50 for Slavery? The result of the proposal would be the same if the desired variable was in the range of 21-50 (any more and it wouldn't have been the most "conservative" proposal).

FixA: The exact desires of polBlocks wouldn't be known. So as Ruler you might be able to see that the Masochists want Total Slavery, Meanies want Considerably More Slavery, while Pansys and Wimps want No Slavery. But you won't be able to know the exact numbers (except for Total and None, which are 100 and zero respectively).

So you can try to tinker around, but you never really know Exactly what the optimal number would be.

FixB: Discontent. The greater the difference between the Ruler's desire and a polBlock's desires, the less that block will like/support the Ruler.

Ruler desire would probably take "effect", for gameplay purposes, right before each policy session, remaining constant until the next policy session.

So you can't change your desire right before the session, then change it back right after so as to minimize your reputational hit (it would basically be an exploit in the making, and it reflects that people only Really care how you act/vote - but no one told the media that, apparently).

That's on the low ends of discontent. On the higher ends they could do all kinds of bad stuff, like riot or even rebel!

This would allow for an American Civil War-like scenario, so you could end up knowing how Abraham Lincoln must have felt

Or George Bush, or...well, it's probably a pretty common feeling. So you'll be forced to make compromises if you want to get anything done, which is more realistic than any system I've seen. Now just as long as people don't hate it...

Anyway, the point is that while asking for more than you want/need is an oft-repeated scenario in real life, it does have setbacks. And if you overdo it you can get yourself into some serious trouble. ("He wants what? That's it, now he has to Die!")

So over-state your desires at your own peril.

2) Intensity of belief. For instance, consider Abortion. There are people who'd be willing to compromise to get something else they want done, and then there are people who won't even consider giving a millimeter! Over their dead body will that happen, they say!

I can think of 2 ways to represent this.

A: Leeway: In the voting process, on each desire there is a specified "leeway". If the leeway is 10, for instance, and their desire for something was 30, then they'd be fully accepting of any value in the 20-40 range.

Someone who doesn't care would have a max leeway, and a real hardcore type would have a 0 - in other words if you don't agree with them completely, they won't agree with you.

So if the current value was 20-40, they will vote against any change.

B: Lack of willingness to compromise in the compromise round.

**Plutarck** · February 21, 2002, 16:08

Flow of Policy Meeting

Here is how I'd think the "automatic" Policy Meeting would go.

This could actually be made into a really cool kind of actual "meeting" type, where you negotiate on issues and proposals - if you choose not to have it done automatically.

But for now, simplicity.

On the turn of the Policy Meeting (perhaps it happens after clicking End Turn, or maybe at the very begining of your turn before control is given back to you) you'll get a popup, informing you of the Policy Meeting which will occur.

You are given the option of making any changes to your policy desires, and you make sure it's just like you want it. When done you can choose "I'm ready, let the meeting begin!"

You then get the following choices:

* Watch
* Participate [For future use, but just leave it invisible/off]
* Just give me the results.

"Watch" will just treat the user to a play-by-play of what happens, at reduced speed (so they can see what's happening). The user should have the option to say "I'm done watching, just give me the results."

The Policy Meeting proceeds as follows.

Proposal Round:

The computer grabs the topic of the proposal, then puts up the proposal in favor of increasing the value, then in favor of decreasing it.

Example: "Proposal to Increase Slavery"

Voting Round:

Each party votes on the current proposal.

The options are "Yay" (In Favor), "Nay" (Opposed), and "Present"/"No Vote" (for those who just don't care either way).

Vote Tally Round:

The No Votes are discarded, then all the Yays/Nays are added up (the amount of "Votes" each party gets depends on their Power, which is currently 1% of power = 1 vote). If a 51% majority of those that voted Yay/Nay are in favor of the proposal, the proposal succeeds. If not, it fails.

If it fails then the "Negative" proposal of that topic comes up. So it would go back to the Proposal Round and be a "Proposal to Decrease Slavery".

If that also fails the proposal topic goes to the next one.

But when a proposal is successful flow continues to...

Negotiate Policy Round:

The Yays decide amongst themselves how much the topic's value should change. The simplest way is just adopting the proposal which desires the least amount of change (the most "conservative"), but this can be fleshed out a lot more in the future.

Perhaps it all depends if the majority of power is considered Conservative or Liberal, even. So a Conservative government would pick the smallest change, but a Liberal one would prefer the most drastic change. Which is what the words mean in every country but America

(both are used as alternating slurs here, hehe)

Set Policy Round:

For simplicity the policy will just be set to the agreed upon amount.

What could also happen is that all the powers then vote on how much of their power they will throw behind implementing the policy, and how much power they'll use to oppose the policy (Power could come to be a set amount, so you'd have to ration your usage of support/opposition, but that's way too complicated for the present time - especially for the AI).

This could even decide how much the trait will increase/decrease per turn (refigured every turn, or every few turns, to show errosion of support of a policy).

The result of this is the final For/Against Power total, from which the disparity of powers is figured, and thus how effectively that policy will be implemented.

And then the flow goes back to the Proposal Round, going threw all the values until all have been addressed.

Then you display the final results of the meeting (they should be stored so they can be looked over at any time, at Least until the next Policy Meeting is over).

And thus the Policy Meeting comes to an end, and the game continues.

The game Pacific Theatre of Operations (1 and 2, I think)(http://www.theunderdogs.org/game.php?id=1875) did something similar to this with a card game, and it was actually one of the things I enjoyed most in the game. It just happened too rarely, and was entirely too random.

That, and the AI was incredibly stupid when it came to this (or maybe it was just a really accurate model of politicians). My plans would be vastly and unequivically better than anyone elses, but it didn't seem to matter much, because it was almost all luck when it came to who won (leaving me to have to load a save many times to get the final outcome I wanted).

The worst was the "Objectives" round, because everyone would often reccommend rediculous goals. Yeah, at the begining of the war it was be Nice to capture Tokyo or Kyoto, but that seems just a Weeee bit unrealistic, if you ask me.

Anyway, that's my stance on the issue as far.

Thoughts, opinions, comments?

**Mark_Everson** · February 22, 2002, 10:02

Hi Plutarck:

This is getting Waaay too complicated for me to discuss. We might get to the more refined points you're talking a year from now. I used to participate in discussions of possible far-future things, but I never got any coding done!

So its not that I'm not interested, I read your stuff, but I just wanted to warn you that you won't get feedback on anything but Big long-term issues or immediate concerns of any size from me.

One thing I do see running through your analysis that I don't think is right is a fixed Policy Meeting schedule. The PMs should take place whenever the player feels like, and then if the player does nothing further then the automated ones could be done at regular intervals in case something has changed. I'm going to wait to see what Rodrigo says before I comment further, lest we go off on a complete tangent.

Cya,

Mark

**Plutarck** · February 22, 2002, 11:03

Mark:

LOL, not till I reread the post did I see what you mean.

I just threw in the detailed points because they came to mind, and I wanted to make sure that the system wouldn't reach such a quick dead-end that, once it came to the point of fleshing out and detailing the system, that the whole thing would just need to be almost entirely rewritten.

So I thought it out into the future just to see if it could work. And I think it could pretty easily, where you could detail one part of the system without having to really increase the complexety of the whole system.

For instance, the Proposal Round in the begining might just be "whoever wants least get's what they want", but some time in the not so distant future, in a land Far Far Away, that single round could be replaced with something much more intricate, and it wouldn't require that any other "round" be changed, or the whole system altered, or anything else rewritten.

So just ignore all the parts about "eventually", "in the future", and "What could happen", and just note the abstract of the system.

With the exception of the mentioning of "rounds" and how it would all "flow", it's basically the same system I posted on earlier

That and the introduction of the "No Vote", which I think is the only subsantive change, though it's still a "future feature".

Now I see how easy it is to make something way more detailed and complicated than you'll need it to be any time soon.

It raises the question of whether some of the models should have a kind of version of what is expected to be done real soon, then a "Future" version, which can be as detailed and complicated as anyone wants to bother with.

But most of the models seem pretty good about this so far, so it's just an idle thought I include just because I don't want to delete it and have writting it be a Complete waste of time

Announcement

Alternative "Negotiation System" for Govt Model - opinions needed!!

Alternative "Negotiation System" for Govt Model - opinions needed!!

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment