David Shor

Head of Data Science at Blue Rose Research, based in NYC, originally from Miami. I try to elect Democrats. Views are my own. he/him🌹

[David Shor @davidshor]

Very excited to announce that we’re hiring for an AI-focused software engineering roles (both general SWE and ML-infra) at Blue Rose!

This is a great chance to work with giant novel datasets and cutting edge ML/AI to help defeat Trump in 2026!

[David Shor @davidshor]

@AuthorJCHowell @SteveKrak What do you mean by forensic audit?

[Ryan Moulton @moultano]

Every election conspiracy starts with someone who has taken stats 101 discovering to their horror that real world data isn't i.i.d.

[David Shor @davidshor]

@trivialanalyst @flaminhottweets @PatrickinNOLA @SteveKrak That isn’t true - in recounts for example campaigns challenge individual ballots (not knowing how they voted, but often guessing based on location or party registration).

[Zachary Donnini @ZacharyDonnini]

I’ve seen some people on this platform say no one expected Raman to outperform Bass in later drops...

In VoteHub’s pre-election LA early-vote preview, we noted the post-Election Day electorate was likely to be bluer and younger, which was better for Raman.

[David Shor @davidshor]

@lxeagle17 @dilanesper @ChrisCillizza How did he do relative to house Republican candidates is the main question imo

[Jason Dean @_Jason_Dean_]

@gfodor @EnablerGPT @davidshor I just searched “Red Mirage” among people I follow

It’s possible you didn’t see it before the election if you don’t follow people interested in electoral politics

[David Shor @davidshor]

@neqyve @mattyglesias As politics gets more multi-dimensional you'd expect more mass to concentrate at the extremes just because of how R^N works

[David Shor @davidshor]

@PatrickinNOLA @SteveKrak If you tried to do anything like this you would get caught and go to jail https://t.co/3eFPQABQa6

[David Shor @davidshor]

@EnablerGPT @gfodor It’s fine to not know a lot about politics but I think you should at least try to ask factual questions like “is it common for left wing candidates to improve in late counts” to Grok before you allege the other side is engaging in massive voter fraud

[NotCompeting @JaneParkway]

@davidshor @PatrickinNOLA @SteveKrak perfect ex of super transparent institution that's subject to tons of social media scrutiny => collapse of trust in institution

[David Shor @davidshor]

For this I was talking about since 2005 - but to be clear both hard side and soft side spending on data products have been enormous.

Koch funded i360 famously dumped giant sums of money after 2012 into this kind of thing.

Attribution to specific entities is hard because there is a lot of very legally complex cost sharing agreements that span the hard and soft side.

[David Shor @davidshor]

If you register to vote and you're not eligible (either because you're a felon or not a citizen), then it's pretty trivial to find out because the state can do a database join to federal databases.

Like, a couple dozen people a year try and they generally are prosecuted because it's an easy win for politically ambitious state AGs.

Ideally I guess you'd make it impossible to do in the first place, but generally whenever you have a very close election that comes down to a recount, both candidates will have well-funded data teams that audit the voter rolls so that they can throw out the votes of like, the two or three ex-felons who didn't realize they couldn't vote.

[David Shor @davidshor]

This I think is the big thing that people who are paranoid about on this issue do not understand: both the Democratic and Republican party have invested literally hundreds of millions of dollars into sophisticated data operations that constantly are looking for evidence of malfeasance to forward to their well funded army of lawyers who are ready to file injunctions at a moment's notice.

Whenever one side tries to do something sneaky (IE, when a GOP sheriff purposely purged every black voter off the rolls in 2008, or in 2022 when someone tried to do actual ballot harvesting), the other side (and also data journalists and outside groups!) notices it very quickly because there actually is a massive data trail on these things that outsiders have no idea about.

[David Shor @davidshor]

@PatrickinNOLA @SteveKrak I have in fact worked on super close elections where both sides would go through and try to find people who shouldn’t have voted and then challenge their eligibility.

There is much more of a paper trail on this stuff than you think!!!

[David Shor @davidshor]

If you register to vote and you're not eligible (either because you're a felon or not a citizen), then it's pretty trivial to find out because the state can do a database join to federal databases.

Like, a couple dozen people a year try and they generally are prosecuted because it's an easy win for politically ambitious state AGs.

Ideally I guess you'd make it impossible to do in the first place, but generally whenever you have a very close election that comes down to a recount, both candidates will have well-funded data teams that audit the voter rolls so that they can throw out the votes of like, the two or three ex-felons who didn't realize they couldn't vote.

[David Shor @davidshor]

@PatrickinNOLA @SteveKrak Every vote is tied to a record on the voter file that has PII - I work with these files literally every day - it would not be hard for Trump’s DOJ to cross check those with administrative data to find non-citizens/felons/etc voting and prosecutors in fact do this all the time

[David Shor @davidshor]

@kane Obviously these things are hard to measure locally, but state+local spending as a % of GDP is like 2x higher in NY than CA so I would be pretty surprised if spending was higher in SF vs NY when you measure it right

[David Shor @davidshor]

@johnloeber I think if it turns out the uber algorithm is biased in the optimistic direction then the employees responsible (from the executives to the data scientists) should face criminal fraud charges and the customers should get meaningful financial restitution

[David Shor @davidshor]

@AuthorJCHowell @SteveKrak What do you mean by forensic audit?

[David Shor @davidshor]

@AndrewCFollett @2miles_aweigh Get help and leave me alone.