iiinova — where your digital self keeps evolving

lunaria/sapphire @hackquantumcpp.bsky.social · 29/09/2026

Likely due to the recent Hart Research poll (Dem partisan poll sponsored by House Majority PAC) showing Leonard up by 1 pt

000

lunaria/sapphire @hackquantumcpp.bsky.social · 29/09/2026

OH-15 is a tossup, apparently (57% chance of Rep win)

101

lunaria/sapphire @hackquantumcpp.bsky.social · 22/09/2026

Correction: my forecast rates NY-17 as Lean D, giving Conley a 65% chance of winning the seat.

000

lunaria/sapphire @hackquantumcpp.bsky.social · 17/09/2026

This change in probability is likely due to the recent (Dem partisan) GSG poll (image below from 50+1 site)

000

lunaria/sapphire @hackquantumcpp.bsky.social · 15/09/2026

I am rather fond of the Waterman butterfly projection

Map using Waterman butterfly projection, centered on Pacific Ocean, with Antarctica cutout

100

lunaria/sapphire @hackquantumcpp.bsky.social · 11/09/2026

You can now see time series charts for forecasted odds and vote shares for individual congressional districts in my forecast! Here is the chart for MI-10, the current most likely tipping point as per my model:

Time series chart of MI-10 win probability over time as per the SnoutCounter forecast

101

lunaria/sapphire @hackquantumcpp.bsky.social · 09/09/2026

I mean isn't that really only 2022 and 2024? I realize that doesn't contradict your statement here, bc a lot of people out there are really young, but this didn't happen in 2018 to my knowledge. Certainly didn't happen in the 2018 generic ballot in particular

closeup of SnoutCounter generic ballot polling averages. Disclaimer, I run SnoutCounter

000

lunaria/sapphire @hackquantumcpp.bsky.social · 09/09/2026

Hell, Republicans have already been hitting Talarico hard on trans issues, and it simply hasn't worked. He continues to lead in the polls (50+1 polling average below)

FiftyPlusOne polling average showing Talarico at 47.4% and Paxton at 44.9% - Talarico leads by around 2 pts

100

lunaria/sapphire @hackquantumcpp.bsky.social · 08/09/2026

For reference, my model gives Republicans a 72.6% chance of winning the seat, versus the Democrats' 27.4% chance.

Snapshot of zoomed in choropleth map showing KY-06 forecast data

Tooltip:
Zach Dembo (D) vs Ralph Alvarado (R):
Rating: Lean R
Dem chance: 27.4%
Rep chance: 72.6%
Posterior mean margin: R+4.4

120

lunaria/sapphire @hackquantumcpp.bsky.social · 27/08/2026

Seriously the current national environment is an *8-pt swing* from 2024 (R+2 --> D+6), and it'll likely get bluer between now and November. Abdul is already leading in the 50+1 polling average by 1.6 pts. A Rogers win is not entirely inconceivable atm but it's just hard to believe that it's probable

100

lunaria/sapphire @hackquantumcpp.bsky.social · 27/08/2026

Underrated point re: MI Sen is that Mike Rogers already ran for Senate two years ago and lost. One might argue that AES is a weaker candidate than Slotkin, but is he *so* weak that he would allow Rogers (a one-time loser) to win in a far less favorable nat'l environment? I don't think so.

100

lunaria/sapphire @hackquantumcpp.bsky.social · 26/08/2026

They're calling it the worst table ever made for a scientific article. Sebo and Long must be convicted of crimes against probability and statistics

Table with rows being various conditions required for consciousness, with the last row being the chance of AI consciousness by 2030, and the columns being "Necessary" (for Consciousness), "Not Met by 2030", and "Necessary and Not Met"

100

lunaria/sapphire @hackquantumcpp.bsky.social · 25/08/2026

Interestingly, Minnesota does not see a spike in ICE arrests in recent months - not even a small one like in some other states (e.g. Oregon, Illinois) There were two spikes (a smaller one in late 2025 and a big one around January-February 2026), and then arrests collapsed

Time series chart of ICE arrests in Minnesota from Feb 2025-August 2026 by criminal status (Convicted Criminal, Pending Charges, or No Criminal Record)

000

lunaria/sapphire @hackquantumcpp.bsky.social · 25/08/2026

Arrests of undocumented immigrants without a criminal record have skyrocketed across the nation, in multiple states

Time series chart showing ICE arrests in California from Feb 2025-Aug 2026 by criminal status (Convicted Criminal, Pending Charges, No Criminal Record)

Time series chart showing ICE arrests by criminal record or lack thereof in Texas from Feb 2025-Aug 2026

Time series chart of ICE arrests in Florida

Time series chart of ICE arrests in New York State

000

lunaria/sapphire @hackquantumcpp.bsky.social · 25/08/2026

I'm gonna dig into this data a little more when I have the time, especially the new joined arrests + detentions dataset, but it seems that arrests of undocumented immigrants - particularly those with no criminal record - have skyrocketed in the last few months

Time series chart made with matplotlib showing 7-day rolling average of ICE arrests over time from Feb 2025 to August 2026

000

lunaria/sapphire @hackquantumcpp.bsky.social · 23/08/2026

I've been running @snoutcounter.bsky.social for some time now, and I thought I might as well make an official dedicated account for that! If you're interested in elections, polling, or just statistics and data in general, consider giving it a follow and checking it out!

100

lunaria/sapphire @hackquantumcpp.bsky.social · 23/08/2026

This is really funny considering the precinct that contains the Church of Scientology voted (with a majority) for Spencer Pratt in the primary. Bass isn't even second-place in the precinct! (source for map below: VoteHub)

000

lunaria/sapphire @hackquantumcpp.bsky.social · 23/08/2026

Yet the Scientologists didn't vote for her back in the primary. Sad!

010

lunaria/sapphire @hackquantumcpp.bsky.social · 21/08/2026

Speaking of the spreadsheet discourse, fun fact: the logo for my site is inspired by spreadsheets. The rectangles framing the syllables are supposed to be cells

000

lunaria/sapphire @hackquantumcpp.bsky.social · 21/08/2026

All of these areas have flipped. Wahab is now winning these areas, though she's leading in the single digits in most precincts

000

lunaria/sapphire @hackquantumcpp.bsky.social · 21/08/2026

The LLNL precinct has flipped, with Hernandez leading by around 2 pts

000

lunaria/sapphire @hackquantumcpp.bsky.social · 21/08/2026

With most of the results in, we can check back on the precinct-level map, and there's quite a few interesting things here. I'm surprised at Wahab's strong performance in Pleasanton and Livermore; if I had to guess she probably outright won both. The late mail greatly boosted Wahab in dwntn Livermore

000

lunaria/sapphire @hackquantumcpp.bsky.social · 21/08/2026

Seems like the Hacienda complex precinct was disproportionately late mail. With the new drop, it has flipped to Wahab

000

lunaria/sapphire @hackquantumcpp.bsky.social · 21/08/2026

Wahab still will go against Hernandez come November but she'll have the advantage of incumbency. The late mail broke hard for her, giving her a 7.2 point lead

010

lunaria/sapphire @hackquantumcpp.bsky.social · 19/08/2026

With how much centrists have panicked over Michigan and have basically dismissed it as lost, you'd be forgiven for not knowing that most models still have Democrats more likely to win Michigan than Texas, Ohio etc Below: Silver Bulletin (Deluxe model), 50+1, Split Ticket, VoteHub (Complete model)

000

lunaria/sapphire @hackquantumcpp.bsky.social · 19/08/2026

Wahab is overall strong in Union City and Fremont, and from a cursory look there does seem to be some correlation between racial demographics and vote choice - the more Asian precincts seem to be leaning towards Wahab, whereas the more Hispanic and white precincts seem to be leaning Hernandez

100

lunaria/sapphire @hackquantumcpp.bsky.social · 19/08/2026

Wahab is winning the LLNL (Lawrence Livermore National Laboratory) precinct by 13 votes

010

lunaria/sapphire @hackquantumcpp.bsky.social · 19/08/2026

Expectedly, Wahab is doing well in parts of Downtown Pleasanton and one Pleasantonian precinct along I-580. What's more surprising is her strength in lower-density Livermore - it's actually flipped from my expectations. The lower-density areas are stronger for her than downtown, despite the latter

100

lunaria/sapphire @hackquantumcpp.bsky.social · 19/08/2026

Hernandez is winning most precincts in the heavily Hispanic unincorporated CDPs on the coast - Cherryland, Ashalnd, San Lorenzo - but it's between 50-60% Hernandez. Wahab meanwhile is doing quite well in much of lower-density Castro Valley - also 50-60% in that case.

000

lunaria/sapphire @hackquantumcpp.bsky.social · 19/08/2026

The map looks about what one would expect (CA-14 special, early VBM + E-Day ballots only). Wahab is doing better in the coastal areas - Hayward, Union City, Fremont, etc - while Hernandez is doing better in the inland Tri-Valley suburbs. But there's a lot more heterogenity than one might expect

000

lunaria/sapphire @hackquantumcpp.bsky.social · 19/08/2026

Fishback literally under-performed the polling averages lmfao

100

lunaria/sapphire @hackquantumcpp.bsky.social · 05/08/2026

This is another reason why one shouldn't dismiss general election polling out of hand based on the polling misses in primary elections. We can't weight for past vote in the same way we do for two-party general elections!

100

lunaria/sapphire @hackquantumcpp.bsky.social · 05/08/2026

Ftr: South Dakota's RED Score (a metric developed by Split Ticket measuring the potency of independent runs in red states) is comparable to Nebraska's.

100

lunaria/sapphire @hackquantumcpp.bsky.social · 04/08/2026

Incredible things are happening in the election nerd community on Wikipedia

Wikibox with map of 2025 New Jersey governor election results by Catholic diocese

011

lunaria/sapphire @hackquantumcpp.bsky.social · 04/08/2026

I need everyone even remotely following the 2026 midterms and dooming over Michigan/Maine to do a sanity check right now (images below from FiftyPlusOne's forecast)

200

lunaria/sapphire @hackquantumcpp.bsky.social · 09/07/2026

I keep track of Trump's issue-specific approvals, and while there isn't a lot of polling on his handling of DEI that matches the required wording, the polling I have suggest that most Americans disapprove of his handling of DEI

100

lunaria/sapphire @hackquantumcpp.bsky.social · 09/07/2026

Though, on the other hand, another poll from Pew in April 2025 found that most Americans opposed Trump's efforts to end DEI programs - 53% disapproved, while 44% approved.

100

lunaria/sapphire @hackquantumcpp.bsky.social · 09/07/2026

This poll is from October 2024, btw, when the national environment was significantly more right-wing. Obviously not every poll finds this support - an NBC News/Hart/POS poll from March 2025 found that support for DEI was roughly evenly divided, for instance.

Views of DEI programs in the workplace

Registered voters are divided over the future of DEI programs in the latest NBC News poll (from March 2025, actually, not the latest)

We should continue DEI programs because diverse perspectives reflect our country, create innovative ideas and solutions, encourage unity and make our workplaces fair and inclusive. - 48%

We should eliminate DEI programs because they create divisions and inefficiencies in the workplace by putting too much emphasis on race and other social factors over merit, skills and experience. - 49%

Notes: The poll was conducted March 7-11, 2025 and surveyed 1,000 registered voters nationally. The margin of error is plus or minus 3.1 percentage points.

100

lunaria/sapphire @hackquantumcpp.bsky.social · 09/07/2026

Ftr, this sort of messaging around "DEI" is bad messaging if you're trying to target anyone outside of the diehard GOP base. Polling writ large seems to indicate that Americans are actually supportive of DEI!

Pew Research Center poll from October 2024 finding that 52% say DEI is a good thing, 21% say it's a bad thing, and 26% say it's neither good nor bad

100