My name is John Kelly, and I am not a political expert. I am a veteran technology professional with a focus on Linux, automation, and system administration. About halfway through the term of the previous President-- in about 2019 --I found myself much more engaged politically than I had ever been, and as someone who's spent over 20 years writing scripts and code to find and arrange data that's hidden or obscured in logs, command output, error messages, or other seemingly random massive blobs of text, I suddenly saw what seemed to me to be a largely unexplored data domain: elections, voting, polling, predictions, demographics, geography.
Without knowing the word for it, I had discovered in myself a love for the study of these things. That study is called psephology, and like the basis of my career, there is a mountain of psephology data, and in that mountain, as with other agglomerations of data I have spent my career handling, there are messages. Messages that are going unnoticed and unheeded. This website is the most complete effort of mine, to date, to try to find, isolate, and cast a light on those messages.
Currently, this website has combined data from several sources to present as complete a picture as I can assemble with that data. Sources like:
The Cook Partisan Voting Index (also known as the Partisan Voting Index, Cook PVI, or just PVI) is a product of the Cook Political Report, a nonpartisan thinktank and paragon of organized political analysis, without which we would be feeling blindly toward the future without insight or purpose. They have done and continue to do the hard work to generate this data. I make no claim to have created any of the data marked with "PVI" on this site. I did not. The CPR made it and owns it, and is allowing me to arrange it next to related data. The world is an immeasurably better place because of the analysis they provide, and I encourage anyone and everyone to support them with a subscription.
Also provided by the Cook Political Report is the CPR Race Ratings assessment, which allows for the mathematical scenarios and summaries I've assembled on the Projection page. Obviously these ratings are simply a starting point for analysis, and not a guarantee of election results. They're usually right, but they've been wrong before, and they will again. But these assessments are invaluable components for analysis, and I personally cannot imagine diving into any analysis without them.
Congress itself provides free access to a comprehensive quantity of data about current and past members of both the House and the Senate. It took some doing, but correlating this data with that of the Cook Political Report has been a big part of the success of this site. There's always something new to discover. I recommend that anyone who's interested in what Congress is doing spends some time diving into the information about past and current bills under consideration. It's a goal of mine to start integrating data about pending legislation into this project.
The Census Bureau compiles per-district numbers for numerous categories of data, with the current latest batch dated from 2022. I hope to refresh this with data from 2023 or later, before long. But as it is, this information gives us fantastic insights into the demographic makeup of each of our 435 House districts and 50 states.
Also from the Census Bureau and used heavily on this site, is the Cartographic Boundary data for drawing House districts, arranged in KML/XML format, and freely downloadable. Publicly released Census resources are in the Public Domain. The quantity of resources available through the Census Bureau is positively staggering, and I cannot recommend enough losing yourself in it for at least a while.
The Federal Election Commission, or FEC for short, governs federal elections and campaigns, and through their excellent API it is easy to see the financial data about various campaigns, as reported to them(pursuant to federal law). This takes a lot of the emotion out of the consideration of each race, but it's nevertheless an indispensible component for analysis.
This site itself is hosted on a Linode VPS running Alpine Linux. The code is perl, and the backend is redis. It's ugly, and maybe you're interested in helping with that. If I can only get that assistance on a paid basis, then the site will probably stay ugly. Don't get me wrong-- I believe strongly that creative work should be fairly compensated. I just cannot afford to pay for commercial creative services for this free service I'm providing. I built it from scratch and maintain it on my own time, and out of my own pocket. There are no ads, no cookies, no trackers, no sold data, and no profit vectors for me whatsoever. Maybe I'll try to defray costs with a patreon. We'll see.
It is my intention that this site remains forever free, as it is also my intention to provide a service to which most people do not have access, not being employed by large news organizations, political parties, campaigns, advocacy groups, corporate entities, educational institutions, or other well-resourced entities that are able to hire consultants and firms to aggregate and present this data to them. I believe very strongly that the public are entitled to public information, and indeed our societal outcomes are better when they have it. So if this site assists with that in an even nanoscopic portion of a measure, then I will consider all the work I put into this completely worthwhile.
The value of an informed electorate cannot be overestimated, and I would consider it the most important work I've ever done if this site serves in any way to move us closer to that goal.