Stacy's Journal

- Soup For My Family on the stove
- Pot roast in the oven
- Pink on the sound system
- Candle lit for cheery mellow
- Cozy pants on

Tangelo season has started. E brought in a bunch of windfall from last week's weather event, and I juiced them this morning. They were all on the small side and FULL of seeds (because we love bees) but even though some were slightly green the juice isn't too tart, not the overwhelmingly sweet E seems to find objectionable, and with that unique tangelo tang. The half full juicer pitcher is in the fridge and my new juice glasses I got for Christmas are just the right size. Excellent start to the citrus season.

As with most fields, the Library keeps pushing for us to find ways to implement AI tools to increase productivity. Given how much of Cataloging is analyzing materials to understand and summarize themes in the content, AI is a crappy option - it doesn't actually UNDERSTAND words, it just knows which ones it sees most often and which other ones are most commonly paired with them, so as to suggest other potentially relevant words instead of having the context to sift through and knowledgeably discard those that don't quite fit. It can scan a bunch of spreadsheets and tell you which words show up most often, but it can't actually do math with the numbers on the spreadsheet, or even reliably articulate a count of word occurrences (even though its back end algorithms have an integer recorded somewhere). But people keep using the word "analyze" for both the contextual understand and numbers comparison versions of analysis, and AI can't do both. The conflated, ambiguous way people use the term in discussing AI drives me bonkers - and culturally we don't acknowledge that it's a huge problem with how we're talking about the technology.

Today the tool under discussion was Microsoft's Copilot, which can be used on any Microsoft data, including Outlook email contents. There was discussion of it analyzing multiple disparate spreadsheets and spitting out a summary report without us having to go through the trouble of concatenating or crosswalking the data involved. I immediately cringed. Granted, in our context data is usually titles and authors rather than prices and quantities, but still, I use Excel because it CAN give me exact numbers data and do the math behind them.

Other suggestions involved figuring out which titles/packages of e-resources give us the most trouble, meaning the emails have to be understood as complaint or compliment. Realistically, if an email is generated there's a complaint, so that's fairly perfunctory, but just because an incident involved 30 emails to resolve doesn't mean it was more trouble than the subsequent title that only took 5, and how is Copilot going to distinguish between those? (Hopefully Re: headers come into play in its understanding of email threads, but this is why some folk are going to experiment.) And will the analysis really be more useful than the off the cuff response my colleague made in the meeting saying "Hands down, it's always [X title]." Does anyone need a full listing of all titles that gave us problems? Are we going to do anything that recognizes they're all with a given publisher/content provider and apply pressure to that org? No, we're just going to fix things as they pop up.

And nowhere that I can find are people asking if the environmental impacts are worth any productivity gains we might see. (I hate editing so almost all AI implementations are anathema to me because they eliminate the initial generation of content which is what I prefer to do and am better at anyway - it's not a boon for me.) I am working very hard to not be That Guy about the environmental implications, because there's LOTS of work-related environmental implications I can't eliminate, so why harp on just one? But it rankles.

The number of fire-related crowd-funding appeals circulating is crushing. Of course because of the sheer number. But also, because they may be inadvertently depriving people of resources.

According to FEMA, "If you receive money from a GoFundMe page for a specific disaster-related expense--such as home repairs, funeral expenses or other emergency needs—you may not be able to receive FEMA aid for the same expense."

Crowd funding can be a very immediate source of funding, and FEMA is notoriously not immediate, so liquidity is a huge factor in the decision. But if evacuees receive $1K that eventually prohibits them from receiving $10K, was it worth it? Sometimes - it's a very case by case question that depends on what kinds of in-kind resources support networks can provide. My fear is that the decisions are being made without understanding of the potential consequences. While many locations are trying to hold workshops and other forms of info sharing about FEMA options, showing up in person to ask detailed questions of FEMA staff can be challenging and official FEMA centers weren't scheduled to open until today:

UCLA has upped remote work for the entire week as of today, so tomorrow's remote login starts with checking with my team about whether they need me onsite for anything, or if I can spend the week catching up on email.

CalTrans still has key exits on the 405 closed, so me going in that way or or via an alternate canyon route would be potentially hazardous obstruction anyway should fires develop/spread over there.

🚧CURRENT HIGHWAY CLOSURES as of 9 PM Sun. 1/12/25 - Stay informed about the latest highway & ramp closures due to ongoing fires & high winds in District 7.
⭐⭐NEW: SR-2 (Angeles Crest Hwy) is OPEN at forest entrance to Upper Big Tujunga Rd #PalisadesFire #EatonFire #lafires pic.twitter.com/3qA52aisaA

— Caltrans District 7 (@CaltransDist7) January 13, 2025

UCLA has announced remote classes and work from home on Monday, extending precautions started Thursday in response to the Palisades fire reaching for the 405.

Southbound (western, fire side) and some northbound exits of the 405 have been closed between the 101 and the 10 freeways. During the Skirball fire 2017 embers did jump the freeway, but hot spots to the west were quickly quashed, often inhibited by the Getty's preventive capacities.

(Twitter's embedding codes for images don't actually allow image loading here)
https://x.com/CaltransDist7/status/1878102595589542023

https://x.com/CHPWestValley/status/1877936259408220389


Meanwhile, I'm still doing laundry, working to keep myself fed, and trying not to read too many comments on the internet.

I went out to water my apricot mallow for the first time in a week. She'd put on so many leaves - she's doing great!

The skies are clear - can't even see smoke plumes in any direction right now. It doesn't smell smokey, but being outside for a couple minutes did make my nose hurt - it may just be a humidity thing, but also I know air quality is currently crap even if you can't smell it, so I had no intention of staying outside.

Running laundry and showered, so improved cleanliness in the works. Now to work on blood sugar...