Hmm, so you are saying that in your experience, the party that gets most votes, when you lose, is already the centrist party?
This is interesting and annoyingly I don’t have stats on WHICH opposition party win, but do on coalitions and wins/losses:
I guess it could be argued that proportional representation will funnel votes towards whichever of the 2 opposition parties is the smallest (as they have the most to lose with wasted votes under FPTP).
So currently I am assuming the PR effect is:
- Move from 40-60 towards 33 66
whereas what it should perhaps be is…
- Move top from 60 to 66
- Move the lower threshold to whichever value seems to provide a mroe even split between the other two parties.
Which is doable. After all, I have the approval value for every voter, every turn. Right now the way the opportunism works is:
- For every voter, calculate their approval, and thus find the average ‘gravitational center’ of all opinion.
- Move both thresholds towards that center of gravity, by a fixed margin (opportunism, defaulting to 20%).
I could simply apply that to the lower threshold separately, polling the center of gravity of everyone whose approval makes them vote for either opposition. So then the algorithm is…
-
For every voter, calculate their approval, and thus find the average ‘gravitational center’ of all opinion.
-
Move upper thresholds towards that center of gravity, by a fixed margin (opportunism, defaulting to 20%).
-
Re-poll all the voters whose approval is below that upper threshold to get a new ‘opposition center of gravity’
-
Move the lower threshold towards that new midpoint
1 Like
Then, if I understood correctly,
- Upper Threshold Opportunism: Works the same way as it does
- Lower Threshold Opportunism: Moves toward opposition (or general) center of gravity - (but not sure which center to go since opportunism might work in a way that squeezes smaller ones. who’s going to be more opportunistic? the moderates or the radicals?)
- Upper Threshold w/ PR: Goes up by a bit (+6%p maybe)
- Lower Threshold w/ PR: Moves toward ‘opposition center of gravity’ by a bit (can’t be specified the current LT might be very close to the opposition CoG)
I guess this would work nicely. Though it wouldn’t be a magic cure for late game landslides, I guess this will definitely work better and might bring interesting situations. Players might be able to make political landscapes more favorable by playing divisive and letting the oppositions fight each other over opposition voters.
1 Like
I have coded this, but its actually not working quite as I intended and needs more thought. These lines show the thresholds (in debug mode), this is a 3 party PR system using the algorithm I described.
I think this is a classic case of confusing approval with voting intention! (which players do a lot :D). In the scenario shown, the red party needs to move that bottom threshold way higher. I don’t need to count the average approval of the opposition voters (nor should it work this way in the non-DLC game btw)… it should be a case of counting binary yes/no votes at each level of approval…
1 Like
Maybe PR-induced pseudo opportunism doesn’t really need to be that strong? The third party just performing marginally better (need to be visible though) would be enough. Or you can just make that happen over several turns and say “there’s an inertia to it” or something.
This was one of the issues discussed when fan Korean translation was ongoing. There was no way ‘approval’ in Korean can be understood as ‘average value of approval intensity of all voters.’ So, the words that imply level of approval became ‘satisfaction’ (never used with any words meaning ‘ratings’) and others referring to ratio of voters became ‘(approval) ratings’ in Korean. It would have been nice if official localization did something similar and prevented players from misunderstanding.
1 Like
Yup, this is what I am currently balancing. There is a limit to the opportunism change per turn to ensure parties cannot flip on a dime. Right now, the 3rd party is still an irrelevance, so I’m still tweaking 
Just a bit to add regarding the dev blog:
Glad that party membership unbalance under 3-party-system is going to be addressed soon. But I personally think that the moderate party should have much less members in the first place for some reasons.
- I don’t think moderate voters are that active at political participation. Why would they join a party and pay a fee when they don’t even hold a strong view or feeling?
- With a large enough party membership, they can also have a lot of party activists, effectively nullifying the voting probability penalty from lower Strength of Feeling with high Party Motivation. It might be okay if the moderate party made efforts to build that membership but it’s not (at least for players’ eyes).
- Also, party membership itself is more valuable for the moderates than the radicals since they are basically less likely to vote without it thanks to low strength of feeling. So, all 3 parties having some membership is more favorable for the moderates than all of them having less membership.
- I personally find it unrealistic that more than half of the voters have a party membership. I just want it to be much smaller.
- The moderate party already has an advantage of being in the center which is likely to include more voters than the other regions. The combination most populous region with high loyalty from membership makes coalitions in a close election even less likely.
Also Read: Voting Probability Simulator and Impacts of Voter Turnout
But there is a DLC with all law in this photo or is only statistics?
First-past-the-post, or a constituency-based system, are voting systems where voters cast votes for local candidates, who then decide in a separate process who to vote for as president/prime minister. This can result in a lot of ‘wasted’ votes in parts of the country where the majority always votes the same way, and minority votes are effectively wasted. This has a negative impact on voter turnout.
Proportional representation is a voting system where the final result is always directly correlated to the overall number of votes casts for each party, regardless of where those votes are counted, so the president/prime minister always received more votes that their opponent(s). PR is more likely to result in a coalition, and also means voters are more likely to actually turn out and vote.
I took some time to read those descriptions carefully and… there are some things to add.
US: typical FPTP (“in a separate process who to vote for as president”)
UK: another very typical FPTP
FR: still closer to FPTP…? both presidential & legislative
DE: PR, obviously
CA: FPTP
AU: kinda less wasted votes here but still FPTP I guess
ES: PR
KR: FPTP but no local candidates or electors in presidential election
IT: really curious what 37% FPTP & 61% PR would mean here…
JP: FPTP mixed with PR
- there are obviously some unclear cases. but I’d still say it’s kinda okay since some simplification is necessary.
- FPTP description seems to be assuming the president is elected via indirect election but this is obviously not the case in France or South Korea.
- PR description says it ensures the president to be elected with more votes than the other candidates. this is what happens in nationwide-constituency FPTP election such as ones in France or South Korea.
- I’m not really interested in these categorizations. but I think there’s no need to make their descriptions misleading.
Those are all new policies.
More work in progress stuff:
1 Like
What about adding voter group names or portraits to each selectable simulations as a QoL feature?
im not sure what you mean by that?
I mean something like this? To make it easier to understand who I’m going to satisfy or piss off. I honestly don’t really remember which groups are affected by a given simulation except for GDP and Foreign Investment.
1 Like
Those sim items are clickable-through, so you can evaluate the data in its entirety before making a decision
Did some more work on this today. This is the UI for a new screen just before the election results are calculated. Basically this lets you gamble on running a negative campaign, or play it safe with an optimistic and fact-based campaign.
The more negative the option, the bigger the potential gain, but the higher the risk of a negative backlash, meaning your campaign funds are partly wasted.
2 Likes
Effect of IRV in AUS is that you still see Duverger’s law (results coalesce around 2 parties), but there is no spoiler effect- so long term, the membership and vote of the third party is not suppressed.
1 Like
Have you considered simulating constituencies as well as voters? Even abstractly (i.e. without reference to specific real-world constituencies, because that would be a lot of work), that might help differentiate the systems.
You simulate 2000 voters at the moment, right? At game start, you divide them up in groups of, say 25 voters. Bias the sampling for some constituencies towards various voter groups (along the lines of rural constitutencies, urban ones, etc, patterns recognisable irl). In the election, whichever party carries the plurality of the votes in the constituency carries the constituency (which would look like carrying all of the votes in that constituency, I suppose).
Since the sampling of members would have some amount of noise, you could generate as many unique constituencies as you liked for the election by simply repeating this process until you have the desired number. You could change the sampling biases for each repeat too. Granted, if you wanted to do that, you’d need to duplicate the voters for the purposes of the GUI so you didn’t have a voter in three different constituencies, but that seems surmountable.
Switching between FPTP and proportional should yield different electoral results under this system, as long as the membership of the constitutencies isn’t randomly sampled.
1 Like
The topic of constituencies comes up a lot. I have to say that the only real interesting reason I can think of for modelling them (even as an abstract form like you rightly suggest) would be to reflect the difference in voting choices that are made depending upon the proportionality of the voting system.
TBH, it would not be that hard for me to just double the number of voters that we simulate in the game. The per-voter processing is already pretty quick. The key would be to get enough of them so that the constituency number and voters/constituency seemed realistic.
So in the US, 50 states, evenly (for simplicity) divided, and 4k voters means 80 voters per constituency. Definitely possible. The Uk has 650 constituencies so 6 voters each… which is less so…
Of course the real issue is that constituencies are not the same. Some are wealthy, some skew a certain way. California is rich liberal, Texas is more conservative. Adding democrat votes in California is useless etc…
Ultimately it all comes down to striking the balance between complexity and accuracy and fun. To be accurate, we probably need to model a ton of characteristics for each constituency. Not fun.
…and yet… I can imagine a cool user interface showing each constituency results trickling in along the bottom, showing the vote spread in each one… arggghhh. I just cant decide :D.
3 Likes
Some new screenshots from this DLC. Firstly the slightly snazzier choose election focus screen:
This shows a small accent icon attached to the simulation value you are currently focusing the campaign on, as a reminder from the main UI:
And now if you select that item, you see an extra red icon top left by the name, an icon attached in the very center of the diagram (both have tooltips explaining its the campaign focus) and also any effects from this value to voter groups now also get an icon to indicate they are being boosted.
I’m interested to know if this is ‘enough’ in terms of reminding the player what they chose? Its only for 4 turns, because the focus runs in the year leading up to the election. I have considered finding some space in the new voting tab of the electioneering screen, but there isn’t really much room, and it may feel a bit superfluous?
A reminder:
Its tricky because the focus is definitely a ‘campaign’ thing, which means it really belongs either in manifesto or speeches… but neither of them feel right. I definitely don’t think yet another tab is needed, because 6 tabs is busy enough as it is, especially considering the length of the German language on tab titles…
2 Likes
