Happy New Year! I hope you all enjoyed your holiday.
Today’s dev log will talk about what I’m thinking about doing with the congressional system in the game, explain the thought process behind it, as well as give a quick update on primaries.
I first want to start by addressing the fact that The Electorate’s scope has evolved significantly from when I first started this endeavor. At first, I planned on making a basic presidential election simulator with just the 50 states and DC. That turned into all of the counties in the country. Then, after several other increases in scope, I realized that this wouldn't be complete without congressional and gubernatorial elections. I added it as a post-launch feature to be developed, but after discussing with some of you and gathering some feedback, I feel like it’s important to deliver a better experience at launch, with features to add post-launch. Here’s what I’m thinking.
Data-Driven Campaigns
While I fully recognize that there are games on the market that allow you to assume the role of a congressional candidate, and even as low level as a school district official, I cannot truthfully say that I find that type of gameplay fun. The Electorate is a sprawling campaign with plenty of options for how you want to run for office, but it’s built to operate at the highest level. If I gave players the ability to play as a congressional candidate competing for a single district, you would be locked into only campaigning in that district, which means strategy goes out the window, and you’d be constantly spamming the rally event since it would be the only reasonable thing to do.
I just don’t think that’s fun.
However, I have a solution.
Instead of playing as a congressional candidate, you can play as a party chairperson. Want to run the Democratic National Committee, manage candidates, send funds to their campaigns, and build a sprawling ground game infrastructure? You’ll be able to.
Want to run a small independent party that’s hyper focused on electing deeply conservative candidates? You can do that too.
I think expanding the scope and allowing you to assume the role of the party is far more engaging and fun than playing as a single candidate in one district.
Meanwhile, these campaigns can overlap with the presidential campaign, senate races, and gubernatorial races, and as a party leader, you can help those candidates win as well. Then of course, you can get the same type of overlap when playing as a presidential candidate as well, and even campaign with those candidates in their districts. But that brings me to the next issue.
The Congressional Map
If you’ve been following American politics as of late, you may have heard about the redistricting battles happening across the country right now. Maps are redrawn at least once or twice every decade. That means that a map I make for the game is guaranteed to be outdated at some point. The issue, of course, is that the maps I have for the game are not the easiest to generate, and require significant work to get right as well as map its data with the game’s systems.
I’m not saying I can’t do it, but I am saying that it’s a real pain to do.
The biggest issue is that voter population data is tied to counties. Some counties have several districts inside them, and some are so badly gerrymandered that counties may have a small sliver of a district.
I think a potential, admittedly slapstick solution to this, is to simply multiply the county’s data by the percentage of the district that overlaps it to get the right proportions, and then just divide the total US population by 435 (the number of House seats) and apply the proportions to the number that comes out.
So, if District 1 is 100% in Palm Beach County, then Palm Beach’s voter profile would be applied to ~800,000 voters. If District 1 is 80% in Palm Beach and 20% in Broward County, then we take 100% of Palm Beach’s profile and apply it to 80% of the district’s population, and then 100% of Broward and apply it to 20% of the district.
This would obviously not be a fully accurate depiction of districts. Districts that run through the urban core of a city are likely to be significantly more democratic than a large county that holds the city and its more conservative suburbs. In this case, I may apply boosts to voter shares based on the dominant county’s ideology. Ideology for counties are set based on the highest share of voters that live in that county. If a district is large and covers multiple counties, then we’d search for the highest percentage in the geographical boundary, get the main ideology, and apply a significant boost to the voters who share the ideology in the district, while removing an equal amount from the other groups to keep it balanced.
Once again, not a perfect solution, but it's at least somewhat close to being accurate. Importantly, though, it does allow it to be dynamic and make it possible to change district barriers without having to make completely new maps from scratch. At least in theory.
The alternative to this is to simply not have a congressional map and list districts by state and the counties they are a part of. It’s not ideal, but it’s much easier to do this in the meantime as I get the game closer to launch, and revisit the map later as a post-launch content update. There can be a mix of auto-generated candidates along with their real-life counterparts, and you’ll be able to see a heatmap of how these battles are going by looking at a map mode of the states. This would only be for House elections only, though. You would still get a full experience with Senate elections and gubernatorial races, but that’s because those maps are based on counties.
Update on Primaries
I’m happy to say that primaries are really fun in The Electorate. Primaries take full advantage of the voting system, but with a couple additional factors. Candidates who share similar positions with an ideology group will receive larger vote shares of that group compared to a candidate who doesn’t. However, when multiple candidates share similar positions, the game divides those shares down the middle. This led to many instances where same-party candidates were statistically tied in every county.
Los Angeles County polls on turn 1 without name recognition activated.
To fix this, I’ve added a variable for primaries only called Name Recognition. Name Recognition scales from 0.01 to 1, and is designed to create more variability in the starting position of each candidate during a primary. Candidates have a base name recognition that’s set pre-election, which will be filled for each state, and then a candidate’s home state will receive a +0.3 boost from their base score. These base scores are also supposed to represent the general knowledge of a candidate and their positions, not necessarily whether you've ever heard of them or not. These scores are set subjectively (by me, of course), but will be editable in the election settings before a game begins.
Here’s the result of implementing this variable with just their base score.
Los Angeles County polls on turn 1 using national name recognition.
As you can see, Bernie Sanders takes a significant leap due to being the most well-known out of the three here. Now, here’s how it looks when you factor in the candidate’s home state advantage.
Los Angeles County polls on turn 1 using home state name recognition.
Harris maintains her positioning, but Gavin Newsom gets a +13% bump due to being the state's governor. This sends Sanders down to 26.7%, who despite his national reputation, is up against two California natives.
From here, it’s up to how each candidate campaigns. In a larger primary, name recognition has an even larger impact. If you want a real world example, just look at 2016. Jeb Bush, with a well known family name and decent track record in Florida was the clear leader of the Republican primary until, well…you know what happened.
What’s Next
Lots of work. I’m getting there. This project has been very rewarding to work on and I can’t wait to release it, but I also really want to make sure it's polished. So, please bear with me as I work diligently over the next few months.
Thanks,