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
