Category: democracy 4

My politics game, Democracy 4, has a user-experience issue. Its not critical, and its less prevalent in Democracy 4 than any of the other games in the series, and I am working on fixing it, but it remains an issue nonetheless. the fact that this issue is based on real world phenomena does not make it any less frustrating for a player who experiences it.

Some background: The game is a ‘run the country simulation’, that models an entire country, and the electorate are represented by 2,000 simulated AI voters. Those voters go through a very complex web of calculations to decide what they think in the years that you are in power, and at election day, they cast their vote either for the player, or for one of the opposition parties. In a two party system, the decision is quite simple, as players who have a . 50% ‘approval’# of the government vote for the player, and the rest vote for the opposition.

During the game, the player can regularly check on how they are doing regarding polling, by checking out a screen that shows two graph lines. One of these is voting intention, and the other is average approval. This is a NEW thing for Democracy 4, because previously we would only show voter intention. The difference between the two is incredibly, incredibly important. Let me explain why.

An individual voter has a fluid measurement of their approval, in that it ranges from 0% (Hardline hatred of the government) to 100% (fanatical dedication and loyalty to the government). However, for most people they likely hover closer to the center. Not many people would sacrifice their lives for, or against a government in a western democracy. So people’s opinions float around. For many, that float is always above 50%, or always under 50%, but depending on how the government is doing, a lot of people might float below and above that line, and pollsters call these people ‘floating voters’.

Here is the problem: Although approval is fluid, voting is binary. On election day, assuming you vote, you vote for the government, or you do not. You don’t vote ‘more’ or ‘less’ for the government. You either ticked the box or did not. Thus, its important to realize the distinction between the two systems of measurement. The graph above illustrates the difference. Voting intentions in blue are basically a binary poll of yes/no on the government. The green line is the aggregate of asking everyone to rate the government on a 1 to 100 scale. These are MASSIVELY different.

Take an extreme example, where everyone in the country thinks alike. The government has been…ok, but a bit corrupt, a bit untrustworthy, but the economy is doing well. On balance, the consensus opinion is that the government rates as 49 out of 100. When everyone is asked in a survey, they give the same answer, so average approval is 49%. Not bad. However, if forced to vote now, everyone would naturally vote for the opposition, so the election result would be 0% votes. Absolute defeat.

Then the next turn (3 months in game time), a surprise event happens. Scientists make a breakthrough in quantum computing, thanks to a science budget you raised 2 years previously. This good news bolsters the stock market, some voters feel a bit more wealthy, and this boosts GDP just a tiny fraction, pushing wages up just a tiny fraction. Everyone in the country is just a tiny bit more satisfied with the government, hardly noticeable. Maybe 2% difference?

Now its election day, and again, if answering a survey people give very similar results to before, but just slightly better. Average approval has nudge upwards just a fraction to 51%. Barely noticeable on the green line. However, the election is again binary, so given their new scores, every single voter in the country votes for the government. An overwhelming landslide of historic proportions. Total electoral dominance. The blue line shoots to the top of the graph!

This is why the two-lines are introduced into Democracy 4, because in theory, this could really happen. A slight change in voter happiness can yield a massive electoral win…or unexpected crushing defeat. And when I have paragraphs of text to explain how it works…this makes sense. The trouble is, I’m not there to give post-election analysis to every player who thinks the game is broken because of a sudden reversal of fortunes…

In practice, this problem is not overwhelming, because the voters in the game are very sophisticated AI models of widely varying people. Voters might be in 10 different voting groups, each at different levels of membership, and are being pulled in many directions at once, as you can see in the focus group screen:

…so in most cases, for most players, at most elections, this does not occur that wildly. But nevertheless, its possible, and a good game would never surprise the player with what seems like an unfair loss, or an unfair win. They have to be able to see immediately what is going on.

Funnily enough, all the information the player needs to understand this is already in the game, its just not necessarily in the right place. For example, I present to you the parties screen:

This represents approval as the Y axis (chart on the right btw…) and plots every single AI voter. Depending how distributed they are, the effect we describe may vary. In my nightmare 49/51% example, every voter is on the same horizontal line just moving up and down. In this case, the voters ARE very clumped, and just below the point of approval where they would vote for me, as this is a 3 party country, and they are tempted by the middle party. Right now, this is dreadful news for me, and awesome news for the middle (blue) party, but given the voters are so similar, if I can shift my policy positions closer to those the voter’s prefer, I should get an overwhelming avalanche of voters tilt my way!

This is also something you can see if selecting the ‘happiness distribution’ button on the ‘Everyone’ voter group. This time its represented on an X axis, where the left is 0 (angry) and the right is 100% (happy):

This is the same game, with the same voter distribution, and as you can see, in this case (start of game for the UK), its pretty clumpy. BTW the color of each dot represents happiness. Those green voters are happy, but they need to be super-green to vote for me, as there are 3 party options.

So all the information is kinda there, but people still don’t understand those wild swings in voting outcomes. As a result I have decided to try and represent it another way, in another position. I think that in general, players are expecting to get all their polling information from that earlier polling screen, so if I am going to represent voter approval distribution I need to do it there. PLUS, ideally I need to show it changing over time, not just a snapshot in time which is how its shown on these two graphs. Here is my current solution:

Same chart as before, but the background is now filled out with some extra information in the form of green blobs. These are presenting the voter approval distribution at each point in the game. The larger the blob, the more voters it represents. Over time, the blobs both move and change size, as the opinions of the electorate change.

The ideal situation for the player is to have substantial big green blobs way above the line, so they represent committed voters who are unlikely to suddenly drop below the line. If you have the green and blue lines above the threshold…BUT… there are large green blobs right on the line… then this represents fickle voters whose opinions might easily shift against you.

In code terms, each turn I’m asking each voter what their approval is, and sticking them in to one of 20 vertical ‘buckets’ of popularity. Each of these is a blob. Then in the chart I’m showing the number of people in each of those buckets by the size of the circle. Its more complex than it sounds because you need to work with circle areas, not radius, and also do some pre-processing to detect the largest blob, so you can then scale everything so the blobs are as large as possible (for legibility) while also being in proportion to each other and thus accurate.

The HUGE challenge is, as ever, not technical but UI/UX related. Does this help? does this make sense? and given that players often do not read tutorials, guides or walls of text, is it conveyed clearly enough just by the phrase ‘approval distribution’? Does the red approval line need its own key to explain what it is? (I think maybe so).

As always I’m interested to hear what people think. There is a complexity cost in adding this (it complicates this graph, potentially distressing newer and more casual players), and I need to be confident that the increased visual complexity is offset by the greater insight for the player into what is happening out there in opinion poll land…

I’ve blogged about this a few times in the past, but basically I use a lot of staring at data and charts to evaluate how well balanced my game Democracy 4 is. Its a ludicrously complex game, and balancing it perfectly, so it plays like chess, is actually pretty much impossible (maybe something for deep mind to tackle next), but its always worthwhile chasing that goal as a more balanced game will mean more replayability, and more player satisfaction, so here we go again :D

The last time I collected stats on the games late game balance was in January and some of the headline numbers are here:

Its pretty clear that there is a problem in that everything gets solved as things go on, and things are too easy in the late game (3 terms on…) so lets look at the data for the current shipped version:

A pretty trivial improvement to be honest. Sure, popularity in term 3 is a BIT harder, but health and education are still pretty much solved problems, and crime is still relatively under control. Unemployment is *not* solved, but this doesn’t seem to be holding people back in either economic or popularity terms :(. Stronger, or more creative measures are called for… But first lets compare the other stats. Old:

and new:

Again, not much change to talk about here. In fact equality has got better! Lets look at the voting stats. Old:

New:

‘Winning’ is an internal hidden measure I use that scales some difficulty-related measures, so the increase there is a *good sign* and shows that some internal hidden difficulty-adjusting systems are working more strongly now./ This seems to have resulted in a bit of a drop in votes in later terms, which is a good thing. Winning with 72% of the votes is still not ‘fun’ but its more fun than winning with 75%. Still… I am being too meek about this.

One statistic of interest is complacency. Its modeled in game based on the happiness of each voter group. If they are massively happy, then complacency creeps in. They take the policy decisions you made for them for granted, and start to expect more. The trouble is, a canny player can keep all the voter groups happy… but not too happy, resulting in an electorate that keep voting for you, because they prefer you to the opposition, without any complacency having an impact…

In the real world, there is definitely a phenomena where people get just ‘sick of’ politicians and political parties. Eventually, the long list of dumb things a specific politician has said, or crisis that happened on their watch, becomes so long that the electorate becomes tired of them, and starts to want change. Here is some data from UK elections:

What I find interesting is that we don’t have many cases where a party wins one term, then gets kicked out. Its almost like once you win an election, the next win is pretty guaranteed, and then it gets a bit harder from then on. The 92 conservative win was pretty narrow, for example.

Obviously there are ‘events’ which influence each election outcome, but it does look like British politics is basically a sine wave, oscillating between center left and center right governments fairly regularly. The US picture is fairly similar:

Basically a regular swing from red to blue and back again, with actually pretty small swings. It really does feel that there is this big block of ‘swing voters’ who are very susceptible to a ‘I am tired of this team, lets back the other team’ mentality.

I guess this could be coded into the game by a steady increase in complacency of the ‘everyone’ group, but that feels a bit hacky. I also think that doing so would really just be papering over the problem, which is more simulation-model based, rather than political. In other words: The player keeps winning, not because the voters are not sufficiently cynical, but because things are actually going really well for them.

In the real world, this is not the case. The US/Germany/UK/Italy have not all collectively solved healthcare, or solved education, or eliminated crime. What has happened is that either the goalposts have moved, or the causes of the problems have mutated, but they are definitely still there.

For example: Mental Health. This is a much bigger deal in 2020 than in 1940, but are we pretending that there were not mental health issues in 1940? of course! but other problems were so bad, we didn’t pay much attention. Gender dysmorphia is not a new phenomena, nor is anorexia, nor is depression, or related conditions, but there is a much stronger focus on these issues now, probably *because* so many other health issues have been improved upon.

The game frankly does not model this. We model education just as badly. Its a 0->1 scale, that never changes, and if you improve education to 1.0, you are done! but the real world is different. The world has become more complex. The understanding of modern economics, modern electronics, modern physics involves a lot more stuff now than in 1950. Go back to when I was at university, and economics was simply macro/micro. Then along comes behavioral economics and we all have to go back to school.

Medicine is another area where the goalposts are in constant motion. As our lifespan increases, and technology advances, the demands on healthcare race upwards too. People who in 1950 would already have died, are now living relatively healthy independent lives….but at a great cost in terms of pharmaceuticals and technology.

I really need to improve my modeling here. Right now, the game *does* have some of this, because for example, technology does boost healthcare demand (by up to 22%), but maybe this needs to be on a curve, and more pronounced. Lifespan also boosts demand, but again, not by much. Maybe this needs looking at, and a review of other similar effects. The only design dilemma is how to represent this. Currently the entire game is coded around a system where voters look at the current state of affairs, and judge it against an arbitrary value like this:

“Healthcare is at 60% so I am 60% happy. When it was 62%, I was 62% happy”

Whereas in reality people probably think:

“Healthcare has fallen 2%. Things are getting worse. I’m voting for the opposition, they might be better?”

I REALLY do not want to change any core mechanics of the game, especially now we are no longer in early access, but I am continuing to think of ways in which to accommodate this phenomena into the way the game already works…

side note: I use WordPress (An absolute trainwreck piece of software btw), and as of the last update, the font it shows in post editing mode is totally different to the one the actual blog uses, with seemingly no way to change or fix this. presumably this is just some stupid new bug introduced by a pointless update I never asked for. Is it too much to ask for software developers at big megacorps to give a damn about quality? The almost weekly complete redesigns of their UI apparently do not trash their users productivity enough, so introducing dumb-ass bugs like this has apparently now become a priority. Absolutely useless…

My current game: Democracy 4, currently has very good steam current reviews and all reviews scores. As I check right now it sits at 90% positive recent, 85% positive all reviews. However, 90% is not 95%, and any improvements to the game are bound to translate into higher scores, longer play times and better sales, and as balancing the game requires a huge amount of data analysis and thought, and is therefore what I consider fun…lets do some more!

Because it is a ludicrously complicated game, the only possible way to ‘balance’; the game is statistical analysis. There is no way I could hire a team of QA people to give me feedback on where the game is unbalanced, because the number of play styles and strategies, and political opinions, and then the different countries, difficulty settings and other options would mean even if I had 100 people analyzing 20 games a day, its still ludicrously trivial compared to the actual ‘problem space’ that the game represents.

Thankfully, the game sells well, and if I pick yesterday as an example, we clocked up a total of 3,754 games played by players in one day. These are proper paying customers whose stats don’t lie, so analyzing the data from these playthroughs is invaluable. I do not collect ALL possible data (that would strain even my dedicated server, whose current database size is 14 GB), but I do collect some interesting snippets (anonymously).

I rely on my own intuition and feedback in reviews (good and bad) combined with forum posts to give me a feel for areas where the game needs improvement. A topic that seems to come up a lot (with justification) is that the first term of a game is pretty well balanced, but if you ‘survive’ to a second term, things get easier and easier and the game loses its challenge. To check if this was really true, and look at some possible causes, I collected stats by each term and got this:

T1 Wins = 54.19 %    T2 Wins = 79.23 %    T3 Wins = 83.97 %   

Thats term 1,2 and 3+ win rate for game version 1.45. Its pretty clear that the intuition is correct. The game gets easier the further you go. The big question though is exactly WHY does this happen. In order to collect relevant stats I needed to think about what factors could make governing a country too easy. Maybe GDP gets high and stays there? Crime gets solved for good? Technology gets ahead and stays there? Unemployment solved for good? I collected player stats at the time of each election and plotted them over 1,2 and 3+ terms to get these stats:

and also…

So it seems that the stats do not lie. Multiple things seem to get good and… stay good. Education seems a bit easy to raise after 2nd term, but maybe thats because a booming economy allows for infinite spending on state provision? Sadly I do not have data for that, but its clear that GDP has an inexorable rise. Its also interesting to see that crime steadily declines, especially violent crime.

Stats like these are invaluable. Democracy 4 has a bunch of self-balancing mechanisms that are not artificial game constructs but based on what appear to be real-world auto-balance phenomena. It does seem to be the case that skyrocketing GDP in the real world can drive massive inequality (relative poverty…. even if actual absolute poverty declines). This rise in inequality can give rise to higher rates of crime, and also to political dissatisfaction.

Higher GDP and Technology can be a sign of increased industrial automation, which should really cause big spikes in unemployment, and thus also ineuqlait6y, but these do not seem apparent. Luckily, I have control of every equation in the game, so after reviewing all these stats I could make a bunch of adjustments and hope to see a budge in the direction of a more balanced (ie: more challenging) late game. Lets look at the new stats for build 1.46:

T1 Wins = 54.54 %    T2 Wins = 76.46 %    T3 Wins = 83.06 %   

Ok… so slightly better T2 (a bit harder…. 76% win vs 79%) and very marginal improvement from term3 and beyond. How did my actual core stats change:

If anything WORSE for GDP, although I did make unemployment worse, and a slight term 2 boost to crime, but fairly trivial. second row:

Ok… so equality IS now harder to raise in later terms, which is a slight improvement, but basically I have been way too meek in my adjustments and need to ramp those changes up. I also started tracking some additional stats:

This is new data for build 1.46 without prior data to compare to, but I felt it important to start measuring. Winning is the value of a special hidden variable I use in the game to adjust difficulty. That percentage DOES go up over time, which makes sense, although its clearly not hitting the upper 90s, so I could simply take into account consecutive election wins as some input into calculating its value?

Turnout is more of a curiosity, but vote percentage is much more interesting. In the *ideal* game of Democracy 4, the player scrapes narrowly through on each election, excited to view the results, and relieved to know they won another turn. 52 and 58% feel about right, but averaging 74$% is pretty worrying. Perhaps cynicism is not a strong enough effect here, especially given that it clearly drops between terms 2 and 3! (in the real world show me a president who people did actually become less cynical about over time…).

So the main thing is that I have useful stats and a system to compare them and check I am heading in the right direction. The game clearly got a bit more balanced between games 1.45 and 1.46, so here is hoping that being able to spend more time post-release on balancing versus features will allow me to improve that even more…