Hi folks
I pushed out many updates to https://abif.electorama.com late last night, and it's becoming a lot of fun to use. Some big changes:
- Tabbed interface to switch between election methods on any given election
- I've added approval voting
- it is MUCH easier to just browse tags to find various elections now.
I haven't added many new elections, but you may not have noticed many of the elections that were already there before, since the tag navigation was so bad.
I did add one notable election: the NYC Democratic Party mayoral race of 2025: https://abif.electorama.com/id/NYC2025-Mayor-Primary-Dem
Interestingly, by my count, it would seem that Andrew Cuomo lost in pairwise comparisons to THREE candidates (not just Zohran Mamdani). The November general election is going to be interesting....
The latest version of abif.electorama.com supports approval voting. An interesting bit: it converts ranked ballots to approval ballots. The algorithm uses a funky way of approximating reasonable-ish voter behavior extrapolated from their rankings, which I call the "favorite_viable_half" algorithm: find the candidate with the most first preferences and then determine the minimum number of figurative seats that would need to be open in order for the candidate to exceed the Hare quota with the given first-prefs (an estimate of viable candidates). Assume then that all candidates that the voter ranks above their respective quota of viable candidates are approved, and any ranked below their least favorite in their quota was not approved. Their quota: half of the viable candidates, rounded up. Of course, if they don't rank a candidate, then that candidate isn't approved.
Since there isn't a jurisdiction (yet, that I know of) that allows voters to draw an approval line, the algorithm above is my best guess at where they would draw the line between approved and not-approved candidates. In the examples that I've looked at, it intuitively seems rational. For example, let's look at the 2024 SF mayoral race: https://abif.electorama.com/id/sf2024-mayor/approval#approval
By the favorite_viable_half algorithm, Daniel Lurie (the winner of the RCV/IRV election) gets 50.1% approval. London Breed (the runner-up incumbent) gets 42.2% approval. That roughly matches what their pre-election approval ratings were, based on my cursory research. Breed had 55% disapproval and 40% approval before the election. It wouldn't be surprising if she netted the 2% difference in the voting booth because voters didn't approve of any of the dozen other candidates running against her. That 42.2% approval is an informed approximation; there were more voters that ranked Breed, but only 42.2% ranked her as one of their two-two favorite viable candidates (of the approximately 4 viable candidates in the election).
The media here in SF had a tough time deciding if 4 or 5 of the 13 candidates were viable. The first publicized mayoral debate had 5 candidates, and the remaining debates had 4 candidates. Speaking as a voter who was asked to rank 13 candidates, I say to all y'all who are jealous: be careful what you wish for.
The update to abif.electorama.com is all part of the upcoming 0.33 release of abiftool and awt. abiftool is the command-line tool for viewing/analyzing elections (and converting between election-result formats), and then awt is the web interface for abiftool that powers abif.electorama.com. It's all free/open source software, so feel free to download and try it out yourself. Patches/contributors welcome! Also, if you find any bugs (there are many), please file a report in the issue tracker: https://github.com/electorama/awt/issues
Don't be bashful about filing feature requests as well. I've got a lot of bugfixing to do before the 0.33.0 release, but after that, I'll be looking for compelling features to implement.
Thanks Rob