Vintage zines and research woes (Learning Things: May 8, 2023)

Love 2 complain about LLM-generated garbage

Here's a rundown of what I've been up to this week:

Paranormal research

Over the weekend, I spent a while looking through old UFO newsletters and zines. I love the aesthetic of the typewritten, xeroxed 1950s-1980s publications (I definitely have a bit of anemoia when it comes to 'em.)

I especially enjoyed looking through this August/September 1979 issue of Skywatch and this 1968 issue of Saucer Scoop (just look at that delightful cover!)

Other than that, my research into remote viewing continues. I finished reading Limitless Mind by Russell Targ. A few odd synchronicities have happened this week related to remote viewing and the Gateway Tapes. I'll probably write about that later this week.

I also finally finished reading Operation Trojan Horse by John Keel. I feel like my pride requires me to mention that I read, like 10 other books while slowly working my way through this one. It just took ages because I was taking copious notes (in fact, the Obsidian graph for my notes on the book are the background of today's art).

DIY paranormal gear

As of a few days ago, I now have all the components I need to finish up the biodata sonification device build, but I just haven't had the energy to do that yet. (Either I'm still recovering from being mildly sick a week or so ago--seems unlikely--or I've been totally thrown off by this eclipse season, but I've been experiencing a lot of malaise lately.) Hopefully I'll get to it next weekend.

Art and paranormal investigation

This week, I've been playing around with automatic drawing using oil pastels. I did three sessions this week, at home, with two types of oil pastels. I used a similar process to my past attempts, but since I used pastels, there was an added element of color, which I liked. (I chose the colors with my eyes closed and switched between different pastels whenever the vibe felt right.)

There are definitely some downsides to using oil pastels, including the fact that they don't glide very nicely over paper. Also, even with my eyes closed, I can feel when I'm drawing on top of an existing line, because the sensation drawing over the texture of oil pastel lines is so different from the feeling of pastel on paper. While automatic drawing, I try not to pay attention to where my hand has been on the page, to avoid tipping the scale and switching from automatic drawing to normal, self-conscious drawing. So the medium makes that a bit harder.

I'll probably try colored pencil this week and see how I like that. That'd retain the easy glide across paper that a pen has while also having the element of color. I just don't think that colored pencils are as expressive as oil pastel, so I'm slightly less excited about them. (As an artist, I've always been unfairly biased against colored pencil.)

Other than that, I've also been poring over my new Hilma af Klint books. It's been helpful to flip through them, and I'm looking forward to doing a deeper dive into them.

Writing

I'm feeling pretty good about the current round of revisions on my queer solarpunk fantasy romance novel. I'm working on filling in worldbuilding gaps and writing those pesky scenes that I'd marked "to write later." Welp, it's later.

I did have one unpleasant research experience, though. Since I'm working out technical details, that means digging into things. (Since this is science-fantasy adjacent, there's only so much handwaving I can do before actually having to come up with semi-plausible scientific explanations for things.)

Anyway, I was exploring some mycology details and information about the logistics of growing mushrooms, and, while reading articles, started to realize that some of them were a little . . . off.

I had some basic questions about the process of growing, propagating, and caring for mushrooms in a greenhouse, and even though I was ostensibly reading guides on the topic, the answers kept evading me. I was getting frustrated. I was definitely reading and paying attention, but I felt like I wasn't, because actual meaning seemed absent from the text.

Then I realized that the prose in the articles kept repeating the same sentiments over and over. (For example, a slightly rephrased opening line started every subsection of one article.) Then I saw the dates of the articles topping the search results--early 2023. And I realized I was reading garbage articles churned out from ChatGPT or other LLMs.

So . . . that's a new thing to look forward to when researching. I've been reading articles warning about this for months, but this was only the second or third time my research has been stonewalled by bullshit LLM texts, so it still feels novel and worth commenting on.

I tried filtering results by date, eliminating everything from the past year, but for whatever reason, the advanced search was still giving me stuff from 2023. I usually use DuckDuckGo as my primary search engine and Google as a backup (I search most queries twice and then compare results between the two platforms), but I gotta play around with some other search engines and see whether they have a stricter date filter.

In the meantime, I have several books on mycology that I've been meaning to read, so I'll probably crack those open and see if I can get my answers the old-fashioned way.

There's something deeply ironic here: LLM tech is supposed to be an innovative technology of the future, but it's muddling search results to such an extent that it pushes researchers to older tech that can't be gamed as easily, like books.¹ It's so frustrating and typical that this supposed big step forward is actually a shove backwards.

1. Of course, books can and often are gamed. I'm a big fan of self-published genre fiction, but for years, there's been a whole wave of cash-grab self-published nonfiction. We can expect a lot more of that with the accessibility and popularity of LLMs; I've heard about plenty of passive income/get-rich-quick types churning out eye-watering numbers of books using LLMs.