Verdant Labs

Yes, absolutely. The vast majority of engineers are fairly right-leaning. Some good reading: http://www.businessinsider.com/charts-show-the-political-bias-of-each-… https://www.nytimes.com/2016/10/07/upshot/your-surgeon-is-probably-a-r… http://verdantlabs.com/politics_of_professions/ CS is very left-leaning compared to most engineering professions, so if you include CS it throws the overall numbers towards left-leaning, but I personally wouldn't consider general CS to be an engineering field. Certain subfields, certainly, but they're a minority. My uni doesn't even have it in the same department as engineering, so that my inform my opinion there. t. MechE/Aero grad student

I’ve never met a conservative teacher. Here’s some (probably low quality) data that confirms that idea: https://verdantlabs.com/politics_of_professions/ Here’s an article about how college professors used to be much more Republican: https://nationalaffairs.com/publications/detail/the-disappearing-conse… Here’s one about Russian infiltration of a Western uni: https://www.propublica.org/article/why-russian-spies-really-like-ameri… Admittedly that last one does not describe an infiltration on the scale I’m talking about. But my thing happened around the 1920’s - 1930’s. Just look at all the Confucian Institutes on campuses and academics getting into trouble for illegal ties to China. If you’re an ideological enemy of the West and seek to undermine it in the long term, this slow infiltration of the teaching institutions is exactly what you would do.

My assumption is that the political affiliation of police officers will largely match the ratio of the area. Majority blue areas will see more Democrat police, majority red (rural) will see more Republican sheriffs. This is born out by the data I was able to find, but I can't vouch for the veracity: http://verdantlabs.com/politics_of_professions/ Sheriffs and State Troopers are very majority Republican, police are barely majority Democrat. I suspect that if we had individual data per region, the affiliation would track with the local politics, maybe with +2 R across the board, or something like that.

> our infrastructure is unmaintained Bernie supporters are generally the least likely to get their hands dirty and do the kind of work necessary to build and maintain infrastructure. Pretty much everyone I meet that does that kind of work is Republican. Not saying this is a universal truth, but it's a somewhat accurate generalization. Source for party affiliation by a sampling of occupations: http://verdantlabs.com/politics_of_professions/

A lot of Republican voters are farm owners [1], and this law is something many owners of trailers, such as John Deer, have been pushing for [2]. [1] Only good source I can find. (Mostly sourced from observation.) http://verdantlabs.com/politics_of_professions/ [2] https://www.wired.com/2015/04/dmca-ownership-john-deere/

As you can see here: http://verdantlabs.com/politics_of_professions/ The engineering profession is about 2:1 Democrat to Republican, and software engineering in particular is about 3:1.

I had added to the title: "79% of software engineers lean D" Per your feedback I have removed this part, as more accurately this would be described as "According to FEC data, out of 100 software engineers who made political contributions, 79 were to democrats and 19 to republicans" In any case this is strong evidence software engineers lean overwhelmingly democrat rather than republican, compared to other professions. Per http://verdantlabs.com/blog/2015/06/02/politics-of-professions/ and http://www.washingtonpost.com/blogs/the-fix/wp/2015/06/03/how-democrat… the data source for the second chart is the same, and refers to the self-reported occupations from the 2012 and 2014 cycles.

http://www.verdantlabs.com/blog/2014/12/30/names-by-profession/ Direct link to the blog that goes in to a few details of how they work out the names.

"In our sample of two and a half million people, a whopping 1.9% of Arnolds are accountants. Contrast that with just 0.55% of Shanes. Arnolds therefore appear to have a much higher tendency to be accountants than Shanes." http://www.verdantlabs.com/blog/2014/12/30/names-by-profession/ You see where this is going. If you correlate the data with the popularity of names in general, you'll find that Arnold is a much more popular name than Shane ...

http://www.verdantlabs.com/blog/2014/05/23/nametrix-2/ There they list the actual top 5 for some professions. Having read that, I am inclined to agree with you here. The top 6 for car salesmen in that graph has literally only one name (Clay) that is in the actual top 5 and even that was only 5th. The top 4 names (Emmett, Luther, Emanuel, Morton) all got replaced with stereotypical white working class guy names. The top 6 for surgeon in the graph has no female names yet the actual most disproportionately common name for surgeons is 'Vivienne'.

The finding here is not so much gender bias but rather a generic leftward bias. Although the headline result is a large bias in favor of women, there's also a bias towards people who put preferred pronouns on their CVs. This is a problem that's been widely known about in the AI industry for a long time now. It's easy to assume that this is deliberate, because of incidents like Google's notorious "everyone is black including popes and Nazis" faceplant. But OpenAI/Altman have commented on it in public, Meta (FAIR) have also stated clearly in their last Llama release that this is an unintentional problem that they are looking for ways to correct. The issue is likely that left-wing people are attracted to professions whose primary output is words rather than things. Actors, journalists and academics are all quite left-biased professions whose output consists entirely of words, and so the things they say will be over-represented in the training set. In contrast some of the most conservative industries are things like oil & gas, mining, farming and manufacturing, where the outputs are physical and thus invisible to LLMs. https://verdantlabs.com/politics_of_professions/ It's not entirely clear what can be done about this, but synthetic data and filtering will probably play a role. Even quite biased LLMs do understand the nature of political disagreements and what bias means, so can be used to curate out the worst of the input data. Ultimately though, the problem of left-wing…

Kinda surprised this isn't common knowledge here, or that people think its localized to SV? Here's a link: http://verdantlabs.com/politics_of_professions/ Software Engineers: 79 Democrats for every 21 Republicans. Similar story for engineering overall (The most "conservative" engineering profession being chemical engineering) IT: 71 Democrats for every 29 Republicans. You can blame it on SV but the truth is our industry's age skews young: We are 38 years old on average compared to the workforce average of 43. Source: https://www.visier.com/clarity/four-common-tech-ageism-myths-debunked/ Young voters and college-educated tend to vote more Democratic: https://www.nytimes.com/interactive/2016/11/08/us/politics/election-ex… You could argue the fact that the profession is predominantly male might mitigate some of this but the problem there is that the gender gap closes once you look at genders on a college-education basis: https://www.pewresearch.org/fact-tank/2018/11/08/the-2018-midterm-vote… What I'm trying to say is that if we assume political affiliation does not correlate with skill (it doesn't AFAIK), and you hire based on merit, and you hire nation-wide, your software development team will skew left. And if your hiring focuses on major urban hubs it will skew even more left. If you want to introduce artificial "balance" you are going to have to do affirmative action for political beliefs. This is challenging, as it's…