Tuesday with Coding and Dinner

I rose late, rolling over at 6ish and 7ish, finally putting on my slippers and leaving my comfortable bedroom to start my day nearing 8. While I do not feel I really start the day until I change out of my PJs, I spend the morning writing the blog, which is done in PJs, as is making breakfast. Breakfast is liberal coffee, half a can of canned pears (juice mostly ignored), and small curd cottage cheese with a dash of sea salt and smoked paprika. I am out of bananas.

While drinking coffee is not a prayer, in some ways, it resembles praying. It brings us focus, especially liberal coffee, which reminds us that we are even helping people with our coffee, and the bitterness reminds us there is so much more to do. Most mornings, we leave behind the shades of sleep, start to sharpen our minds, our day comes into focus, and we sip our brew. It is a ritual. We take it black, with milk or cream, and with sweaters and complex if a barista is around (or you or your partner takes on that role). We believe in coffee! We look into our cup and see the swirls that will soon be inside us and helping us start. Lift up your liberal cup with me, friends, feel that holiness from being granted another day, call out Hallelujah, and drink! My sermon, “Welcome to the rest of your life! I take mine black or with cream.” Amen.

That means I must choose a song for this morning’s caffeinated worship: Leonard Cohen’s Hallelujah. For me, the ‘Secret Cord’ is French Roast, black or with cream, and Equal Exchange brand. We should not forget something from the Moody Blues’ progress rock album Search for the Lost Chord: here, the recessional, if you like.

Returning to the narration, I wrote a long blog and enjoyed writing and being creative on Tuesday morning. It took all morning to write a better-than-usual 1,600 words. Tuesday was quiet, and I coded through much of it. I did my usual reading of the news and switched to the Jerusalem Post for some of my news with the wars heating up in the Middle East. I read the ‘liberal’ press, which is not how I would describe the NY Times. And CNN for events (Ok, they might be ‘liberal,’ but I still don’t see too many stories about helping people), and searched the Internet for information on North Carolina and Asheville. We heard from Glenda and Joyce at Brooks-Howell Home in Asheville today (Susie’s aunts). They are OK, and the retirement facilities took no damage from the storms (they had one before the hurricane). They are managing without power and water to the building as there is a nearby church with power and water. Toilets are being flushed with water from the pool. We, the family scattered all over the USA, are much relieved.

I dress, write some AI Python code, mostly copying from examples, and then go for a long walk, adding about 4,000 steps (I finished Tuesday with 4,600+ steps). Lunch is reheated eggplant parmesan that I have wrapped in plastic wrap and then in foil to freeze. I toss the foil and partially heat the food in the microwave to loosen the plastic wrap. Once unwrapped, I heat the food for four minutes and let it sit, balance, and reach an edible temperature. It is excellent, and the breading stops the eggplant from disappearing in the cheese and sauce.

Corwin got paid, and he is still repaying me for his truck. He stopped by with my money and took one of my good spare headphones–I don’t need two. Mariah asked to have dinner together, and soon Corwin and I were in Air Volvo, stuck traveling 17 miles, mainly with Air Volvo at 12 miles an hour or less, for much of the hour-long trip. We did have to gas Air Volvo, which delayed us enough to push us into rush hour. We arrived after 5 with Mariah waiting for us. We were at Nudi Noodles in SE Portland, a local food joint about good Asian food and incredible but weird drinks, and not about naked dancers. It does make for a disturbing transaction in Quicken! I have the kiwi margarita and the boat noodles with beef and more beef of various types in a heavy beef broth.

Corwin is creating music with an AI tool. I am mostly listening as I am not good with music, mixing, or playing instruments, I am interested in some aspects, but mostly, I am over my head. It sounded exciting. I have not looked into sound and Python as they are not things I understand well.

I could not hear the dessert’s name, but I ordered one. Crème Brûlée Sweet Potato is an excellent finish. Mariah and I shared one. Corwin’s dessert disappeared in a moment. It is a wonderful holiday idea and a good use for yams. Hmmm.

Air Volvo had no problem taking Powell out of Portland to Beaverton and across Beaverton. I was not paying enough attention and was over 75 a few times. I was still going slow for the I-must-get-home-now Oregonians likely headed home to help with homework after a soccer game with the kids. While I saw no lousy lane changes, I did feel like someone was shooting mini-vans and SUVs from a cannon at me, borrowing from Charles Stross. Or, borrowing from Douglas Adams, the improbability drive changed, and it was very improbable, the Oregonians to speed-loving drivers. I am sure the effect has worn off by now.

I showered and dressed early. I read for a while, finishing The Long Way Home: A Chief Inspector Gamache Novel (A Chief Inspector Gamache Mystery Book 10), which was another well-written and excellent story. The author, Lousie Penny, tells a poetic tale somehow mixed with crime and mystery. When I am done, I always think, “How does she do that!” I started Slow Horses after the show and discovered that there are actually eight books, not three, as Amazon stated earlier. Amazon doesn’t update their information as much as they suggest. It is harder to read than the Inspector Gamache books but not as opaque as other spy and crime novels I have tried and abandoned. Surprisingly, indirect storytelling and imprecise language are captured in the look of the TV series. 

I returned to coding after finding that I was scanning and not reading Slow Horses. Switching from the near-perfect prose of The Long Way Home is a poor choice. I discovered that much of the code I used was wrong and built for something else. My AI classifier selects a choice from five options without making a real number value. I instead looked up and remembered that I could ask the classifier model to rate itself. The model was at 59%, terrible, but random would be 20% (with five choices). Likely, the model is overfitting as I have not spent time removing matching features. Generally, you want very different data to find the subtle differences. Using features (columns in the data) that closely match makes the AI model guess these values should be more valuable, thus overfitting. I should drop columns from the training and test. I managed to get just under 61% by tuning the model parameters. I follow Ernest’s style and copy in all the parameters, but I don’t comment on the starting values like Ernest (it is available online). I started making mistakes and stopped after 11. Time to sleep.

I read a few more screens of Slow Horses on my Kindle, put it away, and went to sleep. I managed to sleep past midnight through the night.

Thanks for reading.