Politics game design question

[cliffski]

I'm struggling with part of my politics game here...
Basically the game has lots of voters, and you need these guys to vote for you. They will vote for you if their opinion of you is over 50% (to keep it simple). There is a whole shedload of code that determines what they think of you...
Anyway, I can generate a ton of numbers, but there is one number I need to show the player to show him how well he is doing (he can always drill down the screens to see more specific data).
So the possibilities are these:

1) I average everyones opinion and present that to the player (so dude 1 gives you 45% dude 2 gives you 11% so your popularity is...28%)

2) I work out what percentage of people will vote for you and give that percentage (so dude 1 gives you 51%, dude 2 gives you 10% so you get.... 50%)

On election day (every 24 turns or so) I have to use method 2) because people either vote for you or they dont. The problem is, that is possible for the two numbers above to be VERY different. Maybe your populairty averages 51% but on election day you get 91% of the votes (as the popularity is equally spread) or maybe a lot lower...

So the problem is how to I convey the right information to the player, whilst maintaining the accuracy of the simulation? Dare I show him both numbers somehow?
BTW this is a fairly serious simulation game. I'm not worried about pleasing the pophouse games crowd.
Thanks in advance for any advice.

[DavidRM]

I think #1 sounds like "poll" numbers, so it's perfect for the pre-election period.

Or what you could do is have it be an actual poll: a sample of the voters (instead of the whole group), calculating using #2. Different polls (and you can have several) would show different results, which would be a nice feature, I think. Adds uncertainty.

-David

[Escotia]

Well I'm firmly in favour of showing them 1). It leaves the election result in the air a little so gives a bit of tension and it also accurately reflects how well the player is doing.

Showing them 2) could tell them they're 90% and they might think they're flying but if they're all at 51% then one tiny decision could swing their 'popularity' massively. Of course it'll swing the vote massively but it won't look so huge against 1)

[Bluecat]

An Aussie pollie (I can't remember who) made the comment, "that there is only one poll that counts."

To be accurate you should show polling demographics like age, gender, location, but these should not give a guaranteed result in the actual election. You could do this the same way that real pollsters gather data, take a much smaller random selection of the game population. If you have a pop of 1000, take your polls with 10 randomly selected individuals. Try and vary the poll questions too. Instead of all "who would you vote for" questions, throw in a few "who do you think is better on security/finance/leadership." These are questions on things that influence the final election, but may not determine it.

cheers

John

[Bluecat]

Rereading your question, another thought comes to mind.

After an election, all sides sit down and do a comprehensive analysis to find out why they won or lost. That might be an option for your game. After your election, give the player a screen full of advisors giving their reasons for the results. Maybe something along the lines of the Caesar/Pharoah games might be appropriate.

As a serious sim, I think that you want to give enough info through the polls to give your player a reasonable idea of how they are doing, but not enough to second guess an election result. After an election is probably a good time to give them the rundown on exactly how they did with the different issues.

cheers

John

[Screwball]

You could do a split into to four categories and just have bar style charts. Four categories would be:

Secured Voters (70% +)
Your swing voters (50% - 70%)
Oponents swing voters (30% - 50%)
Oponents voters (0% - 30%)

That way you showing the player you have x amount of voters secured, you have x amount of voters that you could get to swing in your favor and you have x amount of voters who don't really want to vote for you.

[MirekCz]

You use option #2 of course, because:

1.That's how it will be used in the end
2.It ain't the best way to compare those methods with 2 voters.
Once you will get 1000+ voters there will be plenty of ppl with all kinds of %... if you do something good, you will notice it because some will be in the 45-55% zone which will drive the immediate change for your final result.

All in all, politics is about voters that will VOTE FOR YOU and it doesn't always have to mean that you're overally liked. If there are 5 ppl that barelly like you over someone else, but vote for you and 4 ppl which hate you like hell... you still win.

Also, if you take any poll, you will see something like:
1.You like person x
2.You like person y
3.You like person z
4.I won't vote
5.I'm undecided

I think one of the things you're missing is the undecided/won't vote option.

If you use a 0-100% statistic for all politicians per voter (ie if you have got 3 guys to choose, first might get 10%, second 50% and third 55% (yes, it doesn't have to add to 100) - it means this voter likes opinion of politic nr2 and 3, but dislikes politic nr.1. All in all he would probably declare himself as undecided (between 2 and 3, but this is not in statistics usually).

If one voter would have a score like 1-20%, 2-30% , 3-5% then he might be declared as "I won't vote" as neither of the candidates really deserves to be voted for. (One more option might be usefull here, a value which shows how much "I have to use my democratic rights/go and vote" feeling this person has... if he has got this feeling strong (might be boosted by candidates telling that ppl should vote etc.) then even if he has got an overall bad feeling about politicians he will go and vote.

And David's idea regarding pool uncertainity is very good... but to use it you will have to make the number of voters much bigger (more processing power/memory used). If it becomes too much problem you can simply throw some random numbers at it, it's simply a cool feature but doesn't need complicated implementation to look right.

If you need a better explanation of this let me know.
I will also gladly help you with testing/ideas as your game looks very interesting and promising.

[NuriumGames]

Hi,

You could show #1, and allow the player to carry a poll (often in a while or whenever you think) to get #2.

[cliffski]

hmmmm. all very interesting thanks guys. I like the idea of showing secured and swing voters. I already put really unhappy people into a special 'disillusioned' group who will never vote for you again.

[wazoo]

Hey Cliffski,

nice idea for a game....a political sim is sorely needed!

I guess you don't want to "reveal" too much about your title yet, but it'd be cool to be able to hire subversives who'll infiltrate the other side and subtly try to shift the opponents' swing vote in your favor (ie. by spreading disinformation about the other candidate, whipping up emotion, etc..).

Also you could allow "standard" political tactics, like lowering taxes in an election year, government grant incentives, new jobs, etc..

The scope creep in your game is going to be nasty to fight though IMHO..;)