infryq

Yesterday the aggregations failed, and ultimately couldn't be repaired, more on that later. I met with Indicators, welcoming M to the group; she's been doing the engineering for the healthcare data all along but hasn't officially been a member of our team until now. Went to the big group meeting and did the planning for the 1.4 release. Spent the rest of the afternoon fixing the aggregations and respondent ID lists, which turned out to be a series of thoughtless mistakes; I'd inadvertently left my git environment on the wrong branch, then forgotten that we'd added metro areas for Puerto Rico so the validation error the new metro areas generated was only to be expected, and should be listed in the exceptions file. By the time I had everything sorted out, the API got stuck on a larger than usual dataset from another group, and basically hung for three hours. Without the API, the validation checks can't run, so no F data updates went up at all yesterday. While waiting for scripts to finish failing, I added the style fixes D recommended for the issue date PR, generated and merged a PR I'd forgotten about to remove the universal sample size minimum, finished a first pass at the task descriptions for the hackathon, tracked down contact information for two possible longer-term hires, coordinated with A on the procedure for giving a new research group access to the individual response data now that he's taken over as survey czar, and wrote a welcome guide for Z and G, who are joining us to help with not-sure-yet-but-I-hope-to-rope-G-into-the-backend.

Dinner, crossword, then P and I talked about the contract the english department has belatedly offered him for next year. They waited until his health insurance lapsed, which is. Not great, since in order to put him back on my insurance, we had to have him officially terminated, which nullifies all his benefits vestiture for tuition and retirement funds. If it turns out that was an error and can easily be fixed and backdated, fine, but if not, well. There's also the question of how small classes (which are the only kind the department teaches) are being handled in the fall; P would obviously want to teach remotely or online, because he's a reasonable human being who values life over university profits, but the current draft of university policy is to hold all such classes in person, with 30-minute gaps in between so the classrooms can be cleaned. This is a terrible idea and people will die, but our society is so bloodlessly capitalist that I doubt any institution of higher learning in the country has the financial wherewithal to survive taking a year off and reopening in fall of 2021. That kind of forethought requires a level of public stewardship we haven't had in the US in almost a hundred years. Anyway P had written a letter in response which laid out his concerns, which I read and only lightly adjusted. He'll sleep on it for the weekend and respond to the contract Monday.

DS9, snack, sleep. Dreamt of soccer games, barbed wire, cafes, caesar salad, rowboats, thick coffee.

Yesterday I looked at options for swapping in parts from a newer-model electronic sewing machine pedal to replace the rheostat in the White. Alas, the only ones available are three-prong, which means they would not be a direct replacement. They might still work if I can figure out what they actually do, but I’m starting to think my only cost-effective options are to build one myself or wait and see what the performance with the rheostat actually is before replacing it at all. I’m struggling to think of any other household appliances that have a variable speed and load profile even remotely similar to a sewing machine.

For work I spent about half the day on R prototypes and the other half getting familiar with our deployment system and the PHP for the forecast page. I implemented some cosmetic changes and fixed a bug in the error bars display. I also wrote up a sketch for how to organize the international data so it more or less fits in with the existing history display routines.

After work there was dinner, trash, crossword, DS9, and sleep. Dreamt of an Olin edition of Repair Shop, the best chocolate biscuits, a Gibson in need of TLC, and being late.

Personal update: having more trouble with basic self-care executive dysfunction than usual, like being alarmingly resistant to shift out of an uncomfortable position. Stress is stress.

Broader COVID-19 update: Pittsburgh sanitation workers are on strike (?) not for higher pay but just so they can be equipped with better (or any) PPE. This seems entirely reasonable to me and alarming that we weren’t already doing that. PA state senate passed a bunch of useful stuff just waiting for the governor’s signature, like postponing the primary to 2 June, absolving school districts of state testing requirements, and reducing the hoops required for unemployment. If you’re looking for media that’s informative but nevertheless uplifting, I recommend Ologies. The episodes Virology and All (Washed) Hands On Deck have been great.

Yesterday I made another batch of nut and seed bars, and washed the towels and half the regular laundry.

I cleaned the stitch length face plate, first with regular sewing machine oil and then with fast orange, and wow does fast orange brighten things up! It looks nearly new now. I cleaned the stitch length control pins and the plate they’re mounted to, then removed the plate to reveal a ton of black crud caking the ratchet and cams. Cleaned that, and the knobs turn a bit more easily now. Cleaned and oiled the spring for the reverse lever, still quite stiff but less sticky now. Re mounted the control plate and moved on to where the throat plate mounts. It is sticky for some reason? I did a pass with oil and one with fast orange but I’m not sure I got it all.

I ran out of podcasts and switched to music. Yesterday was lemon jelly, today will probably be Air. Mellow electronica with acoustic samples and heavy refinement in production. Anything you might imagine for a Sesame Street interlude from the 70s.

In the afternoon I got email inviting me to join an effort in my department working on crowdsourced predictions for the COVID-19 pandemic, with permission from T for this to replace my regular work. I said yes, and will be starting meetings for that later today.

Meals were all good, I’m obeying my eat alarms. After dinner we did the crossword, then I watched DS9, had a snack, and went to sleep. Dreamt of secret messages, upstairs/downstairs dynamics, a ghost scullery powered by a half-broken submarine, cucumber plants, and nearly falling off someone’s balcony.

Yesterday for work I filled in more symbolic expressions for the knowledge base inferences. I met with M and the advisor crew; he’s got a good paper on sequence-to-sequence models for semantic parsing that utilize conversation history and can do coreference. I met with T and we picked out a CS student project to pursue for summer, then dipped a tiny toe into the scenario annotations I’ve been working on. That document is currently 50 pages long and isn’t complete yet, so I wanted a confirmation or course correction. The particular scenario scripts are not what T considers central to LIA‘s interaction mode (though they were fine before break?) but he agreed we should be focusing on integrating demonstration while TL is still here, rather than trying to first create the asynchronous monitoring machinery that would be necessary for the rule-based instructions he thinks are typical but have never actually been supported by code. More saga. I spent the rest of the afternoon trying to get CUDA upgrade on one of W’s machines. Unfortunately the script I prototyped a month ago didn’t work, so there’s more to do there.

I had a telehealth therapy appointment that was a half hour of actual therapy followed by a half hour of bullshitting about politics when I ran out of stuff. I am not the greatest at functioning under this world’s typical rules set, but I am way better than average at rapidly adjusting to arbitrary new paradigms. I am fine. Other people I care about are struggling — which is understandable, things are very different, I just don’t seem to need the same kind of adjustment period as others before carrying on. Luck of the nature+nurture draw.

The polishing compound arrived, so I spent an hour after work buffing the rust and pitted areas off parts of the White. There is some learning curve, but the shuttle face is now mirror-bright. Very pleased.

Dinner, crossword (fun trick!), DS9, snack, sleep. Horrendous dreams about authoritarian regimes, human trafficking, a quashed uprising, and some sort of alarmingly violent golem made of ritz crackers. Wat.

COVID-19 update: we got mail from the university president saying to prepare to close campus by March 25, then a couple hours later the governor shut down all non-life-sustaining everything with enforcement starting Saturday. I am both cheering because it’s about goddamn time (if no one will die if you stay home, stay home!) but also frustrated because if CMU had been more on top of this, labs with irreplaceable long-term wetlab experiments wouldn’t be scrambling right now. Reaping the rewards of unreasonable campus uptime expectations.

Yesterday I finished disassembling the attachments on the rear of the machine, and started in on the head and bobbin area. There are a couple stuck screws and joints, and I’m using sewing machine oil as a gentle penetrating oil to start. Building up an order for parts and supplies, including a more serious penetrating oil called liquid wrench, a new rubber friction pulley for the motor drive wheel, a new rubber friction pulley for the bobbin winder, an LED replacement lamp, and some updated cabling. My goal for this repair is similar to how I treat my antique spinning wheels: not a purely faithful restoration but to bring it back into workhorse order. Learned a new word: rheostat.

I went to a doctors appointment after lunch for a meds follow up, which went fine (don’t touch your face). When we got back, I helped P debug and verify his remote teaching setup, then went back to the White.

Major bummer: while locating the house micrometer, we discovered the drain in the basement has been backing up the whole house sewage line. P cleaned up the affected area, but basically no flushing toilets or other large waste water usage is allowed until we get a plumber out; still waiting for a callback on that. Bad timing.

COVID-19 update: spring carnival is cancelled, club events are cancelled, cmu library due dates and fines are also cancelled. MLD is completing its current round of faculty interviews remotely. We haven’t been directed to wfh yet, which seems short-sighted, but I plan to arrange that with T. If we end up on campus today I’ll be picking up my spare keyboard. Olin students have, endearingly, staged a faux-mencement before everyone is sent home next week, with academic regalia cobbled together from trash bags and campus clutter. It’s a good place.

Dinner, crossword, DS9, snack, sleep. Dreams were an exercise in gross dude-appeasing dubious consent in a variety of (thankfully platonic) contexts. Do not recommend.

Yesterday I finished tacking the canvas in place, then put the hem back on over a pile of episodes of DS9. None of the remaining projects on my to-sew list are compelling just now so I may switch over to taking apart the Rotary.

Ancillary Sword continues well. The first time around it didn’t occur to me that Radch gender privacy would apply equally to the word “sister” as it does to pronouns, so that shift is doing interesting good things to the plot structure in my brain.

Lunch was good, snack was good, dinner was good (asparagus and turmeric are a surprisingly good match); crossword, more DS9, snack, and sleep. Long involved dreams of trying to find the high school, a maze of athletic complex, a jump rope club into Greek puns.

COVID-19 news: no cases in Pittsburgh yet, but CMU is preparing to take all courses online. My department is particularly concerned with how office hours are going to work, since they tend to be both densely attended and require a lot of 1-1 debugging.