Shelves are done.

I started in on a new painting last night. This one is for my wife. It's the first time I'm painting with oil on canvas, and at 4' x 3' it will be my largest to date.

I got a nice mitre saw for X-mas, so I am going to be making my own frame.

AND NOW FOR SOMETHING COMPLETELY DIFFERENT

Last Friday I got a text from my cousin. "Any want? 'Sligh' manuf Is on its way out the door". A mutual friend, whose in-laws had passed, was clearing out all of their stuff because his wife had not come downstairs in the six months since their passing because of all the reminders of their existence. So he was literally purging everything. I'd met the guy who owned this thing once. He was senile and infirm, but lived in a loving environment. In the bottom of the case is a brush with mortality.

Fiftieth birthday present, August 24 1989. I'm 45.

It's lived in a basement for maybe ten years. The nickel-plated stuff on the dial shows it. It's starting to lift off the white metal. It's also dropped half a dozen numerals, which are plastic. The movement is a Hermle, as nearly everything in the US is, "made in West Germany". Wonder Woman could do worse than to grab the pendulum shield, it's an easy thirteen inches across. It chimes like a dropped piano.

I pondered this for a while. The way you tune a gong block is by finding the lowest note and then snipping all other gongs to match. Except the strike gongs are two identical notes that happen to not be within 3/16ths of an inch of the same length. Then I remembered that when I accidentally made a chime, I started out with one sheet of nickel, one set of snips, one dapping block and one hammer and despite treating both discs exactly the same, they're not quite a semitone apart. So I reckon it's entirely possible that ten years of heat-cool cycles combined with an uneven smattering of surface rust has taken the tuning well away from where it oughtta be but I still welcome the speculation of Complexity and any other armchair metallurgists. A new gong block is like $80.

My daughter hates it. "Because it's big and it's loud and it's rusty." My wife is also not a fan. To be honest this is not a house where a ten thousand dollar seven-foot oak grandfather clock should live. It is a marvelous monster, though. My cousin has expressed interest in taking it himself, though, so I told him at a minimum I'd get it in time for him. We can do this.

An advantage of using proprietary software over cheap chinese bullshit is you can (A) time clocks (B) let the fucker run for days in order to get a nice average. The bottom two marks are me adjusting the pendulum 1/8 of a turn, having adjusted it 1/8 of a turn at a time over the course of several days, only to discover that if you don't tug on Wonder Woman's shield it doesn't actually drop so you're inducing a great deal of randomness. The extraordinarily hard left is me giving it a tug and realizing this. Everything else is me guiding it in over the space of about 20 hours.

Straight up and down on this scale is... keeping pretty good time.

That is, unfortunately, a 60-second average. There's an amplitude periodicity of precisely sixty seconds which means the escape wheel has some uneven lubrication (entirely to be expected on a thirty-year-old clock). Once you take out the sixty second periodicity there's also a 60-minute periodicity which means there's some uneven lubrication on the fourth wheel. There's probably uneven lubrication on the second and third wheel too but those don't show up as obviously.

The big bumps in amplitude are the software switching between positive and negative; beat error on the beastie is about 18ms or about 1.8%, which isn't bloody bad considering I set it by tapping gently with my finger until it settled in. I'm not particularly interested in buying a $15 bottle of clock oil and I'm not particularly interested in slathering the thing in synthetic 5W-30 the way the local clock savages would suggest. The big spike in "average rate" is where I stopped it, adjusted it about two degrees faster and let it run overnight.

Snaking to the left then snaking to the right than snaking to the left again reflects the changes in pendulum length as the room goes from 68 degrees to 71 degrees to 68 degrees. They also reflect a minimum rate of about minus 1.1sec and a maximum rate of about plus 0.5 sec. Right now, it thinks the average rate across a 24-hour period is 0.16 sec/day, which you know what? We'll take it. Not bloody bad for a cheap shitty West German movement that hasn't been lubed in 30 years. If you get serious about it, add an Invar pendulum, ball bearings, jewels and atmospheric pressure compensation, you can get to within a second a month. If you get deadly serious about it, run the thing in vacuum, add an electric remontoire to ensure constant force on the escapement and go deadbeat on it, you can get within a millisecond a month.

Which, really, brings me full circle. It was the lack of awesome longcase clocks that launched me down this road. I've only ever met one clockmaker that had so much as heard of Erwin Sattler, and he's the guy who sells eTimer. They're nice and big on the inside, compared to watches - you can make a clock movement with a jeweler's saw and a drill press. I probably won't soon but insane clock people and their halfassery is the reason I'm here.

I don't need a seven foot tall Michigan grandfather clock. I really don't. I really don't need this either.

But I admire its purity.

Home improvement

"The basement is leaking!" This announcement from below got my attention and meant no ping pong session that evening. The dishwasher appeared to be the source, and I discovered the most common dishwasher installation defect. A piece of string took care of the missing high loop, but the sink has always been slow to drain and this provided the backup which flooded the dishwasher.

A 25-foot snake can be had for less than a dollar a foot now. I had tried before with a toilet auger (going through the trap, in case that is less gross) but found nothing. The longer snake hit a snag some twenty feet in, and pulled up a pasty white material which looked like undissolved detergent. It was nasty, and my first instinct was to clean the barb by rinsing it off in another sink.

That's not much of a home improvement, just trying to keep the place from destroying itself. The painting job is about half done, removing wallpaper and chair rail scars being the biggest efforts. Upgraded baseboard will require a mitre saw, and I'm unsure between renting and buying a good saw.

Coding

TensorFlow installation got bogged down in some kind of dependency hell. You google the error message and find ten ways to fix it, try and figure out which one is least likely to cause further problems and eventually try one at random and get a new error.

While messing with the Raspberry Pi, I remembered the Pi-hole project. I've rarely used adblocking software, figuring the ads are a kind of tax on browsing, annoying little reminders that I could be reading a book. But it was so fun, and so easy to set up, and so satisfying to have this pocket-sized server filtering DNS for the whole house, and so interesting to see all the blocked lookups. I haven't mentioned the change, to see if anyone notices.

Daytrading

I got into a debate with a daytrading friend about the possibility of profit. I've seen some smart friends try to get ahead and not conspicuously succeed, and my gut feeling is that there is too much noise to allow any short-term signal to be recognized. Long term buy-and-hold is a proven technique, and in my view short term trades only work in a cumulative fashion by exposing the trader to the very long term 1.83% per year average per capita GDP growth in the U.S. (and transaciton costs plus capital gains taxes could easily consume this profit).

An interesting conclusion is that I would not worry about the "emotional" investor who breaks the rules by selling low in a panic, or buys high with FOMO. There's a narrative showing these are mistakes if you look for it, but there is an equally plausible narrative that the seller avoided greater losses and the buyer got in while a winner was rising. If emotional investors predictably lost value, you could win by betting against them.

I proposed a challenge in which I would simulate trades in a dartboard portfolio, selecting stocks at random from a list of most active stocks. At first I updated my positions once a week, but this was too arbitrary and mechanical. Things got more exciting when I set up a portfolio containing $1000 each of five randomly selected stocks and simulated limit orders by selling as soon as I became aware that any position had gained 2.5%, recording the 2.5% profit and rolling the returns into whatever stock was then most active. This is a brainless strategy and not recommended, but should be possible in practice.

After two days the paper portfolio was up 2%, with five sells (a 380% APR!). By the end of the week the portfolio was flat again, at $5004.

Despite my Google Apps script sending me an e-mail within ten minutes of a sell condition, my reaction time causes lags, so when SPCE rocketed from $32.17 to $38.67 today (up 20%), I only got two round trip sales out of it. But it doesn't matter! At any point in time, future performance is unpredictable, so there's no reason to prefer getting in earlier or later. The SPCE position is now down 8%, worth $1010, and the portfolio is down 0.9%.

One simulated portfolio is too insignificant to prove anything. For more data, you could test strategies against past performance, but you can't trade in the past so that seems pointless. My expectation is that all five positions will eventually be too far underwater to recover quickly, and the strategy will revert to long-term buy-and-hold, which actually works.

Peanuts and crows

I have drawn the attention of many crows, and one Metro maintenance van, while hiding peanuts in various niches of the parking garage top deck. One difficult hiding place is regularly emptied between visits.

On one occasion I had the opportunity to experience a bird in the hand while trying to rescue two young birds that got trapped in a stairwell.

Kia KIA

At 11:45 Saturday morning I parked our 2006.5 Kia Optima on the street, then went inside to get the kid ready for a class. At 11:46 a neighbor drove his Volvo 850 into the Kia, destroying the driver side wheels and door. Nobody hurt, but the tow driver and insurance agent agreed that it appears to be totaled.

Cultural studies

We have gotten pretty good at yogurt in the Instat Pot, and haven't had to use store yogurt as starter for weeks. The weekly loaf of sandwich bread in the Pullman pan is still irregular. Sometimes the dough rises so much it overflows the pan, other times it is sticky and low-volume. Assuming I am correctly counting four cups of flour (which is far from certain) I try to pay attention to ambient temperature, milk temperature, rest timing, and quantity of sugar (using maple syrup, which is too messy to measure).

When there is overflow, the extra becomes rolls or pretzels or, this week, cinnamon buns which turned out much better than the loaf itself.

Blanket 1

So it turns out the grey yarn I was so fond of, in terms of color, turned out to be wool. After the blanket was washed, in both spots that I used it, there’s some extreme shrink. I wonder more if that’s due to the wool shrinking so much, or the knit of the blanket being so loose. Who knows. I don’t. I’m just doing. But it can’t stay like this, obviously, if I’m gonna keep working on this blanket. So at least for a day or so, the plan is gonna be to try my best to surgically remove the grey yarn, work on the blanket in other areas, and let everything relax and recover before adding new yarn to where the grey yarn used to be.

Edit: LOL. Just started by undoing the wrong yarn, so I had to resecure that with some string and oh man, what a messy way to start.

Edit 2: Nope. This wool isn't going anywhere. It's fused to itself. It's fused to the yarn adjacent to it. It's fused to the blanket. How?!

In other news, my mother sent me a scroll sewing frame that she’s had for years but never used. It’s 1’x2’, so it’s a bit too big for the smaller pieces I work on. But for this blanket (and potentially other larger projects I might be crazy enough to attempt), it’ll be perfect. So yeah, that’s a win right there.

The other month I was a bit bored, so I worked on my jacket’s collar a bit more. Instead of adding fabric, because the spots I was working on were so tiny, I just did blanket stitches around the border of the worn out areas. As you can tell, I’m kind of alternating back and forth between thread colors to give it some variety and character. I’ve worn it many, many times and washed it probably twice since working on it, and it’s held up great.

Lastly, no pic, because it’s a box of paper. Recently though, I got a box of Southworth Business Paper. It’s really nice (with the exception that Southworth tends to use really obnoxiously big watermarks) and I look forward to trying to make text blocks out of it.

. . . still need to work on that dang eagle.

Can you untag me for a while? I'll be gone for a month and will be doing mostly home improvement the weeks thereafter, so not much music making for me unfortunately.