I stepped onto the balcony this afternoon to capture the sound of our jellies clattering in the breeze. I must have startled the wind, because it hushed up the moment I started filming. The silence left behind was soon filled with other phonic creatures bending around the alley – a train calling in the distance, cars whispering down the road, a bird chirping its electric rhythms.
There’s a sense of immersion when you stop and listen like that, as if you’re a forest floor and can feel dozens of different species crawling and bouncing and sprinting across you, leaving each of their unique footprints across your length.
a different kind of buzz
Our reward for waking early is the low, bright light – a light crisp and sharp, so different in tone from the low light of sunset. It puts a bit of life in everything, turning water droplets in the shower into pearls with long shadows, making ghosts of the steam rising in swirls from our breakfast, setting us aglow as we sit to eat.
Good morning, good morning. Wake early and enjoy the light – you can always catch a nap later on.
right here with you
We had a meandering sort of day today, playing video games and buckling down to work, standing in the shade and deciding we should move to the sun, watching a stranger practice his abstruse Tai Chi moves in the park, walking across town in search of quesadilla components, risking our lives shaving or climbing.
This was perhaps an ideally balanced sort of day, with a little of everything we need – it’s rare to have everything fall together like that, but it’s lovely when it happens.
I’ve been taking a machine learning course for the past few weeks. It’s just an online class, something I can work on at a leisurely pace, but it’s good to have something pushing me to keep learning.
This week’s assignment had me building a neural network – an algorithm to allow computers to learn and develop hypotheses about data, designed in a simplified fashion after how actual neurons connect and process information. This is how technologies like Google’s clever search suggestions and self-driving cars come from – the guided crunching of data, phenomena reduced to matrices, ground down and sifted until patterns emerge from the chaos. I’ve gotten the impression, somehow, that given enough time and information, a good neural network could work out any problem, be it structural engineering or musical composition. We’ll see, in time, if real artificial intelligence comes of this someday.
My own neural network here was asked to learn to understand handwritten numbers. From a sample of 5000 digits, it learned to read with an accuracy of about 95%; with the hundreds of thousands of samples available, it might become even better.
These pixelated swirls are a representation of the models the neural network keeps in mind when it hypothesizes what certain numbers looks like. Each swirl is meant to represent some different archetype of handwriting, a certain stroke or other pattern it can search for when it reads. This is still a bit of a black box to me, but I’m gradually getting better at understanding it myself – building my own mental models, I suppose.
Devising these artificial neural networks might give us some idea about what the brain is actually doing when it perceives something. Peering into brain activity via MRI or EEG lets us learn a few things, but the details of our thoughts remain a mystery; perhaps we might learn more about the brain by trying to reinvent it.
the black cap brotherhood
We dozed off early tonight, but I can never call it a day before a photo is captured and written. Whenever I wake like this, it’s a bit like existing in a stray bubble-universe floating between two more substantial days.
I shouldn’t break the silence now. Back to sleep – I can see morning in the distance.
we used to be friends
The thought that I really wanted to put here, at least in the shape I had in mind, has melted away, but I will try anyway.
Because every moment feels like such a permanent thing – and like something of very finite supply – I have been feeling more and more like I need to choose each action with great care. Progressing through life feels a little like trying to craft some delicate precision instrument – a musical one meant to produce a very specific chord, or a scientific one meant to learn something subtle about the universe.
I should realize that it’s impossible to fit your life precisely to any plan you have, but there are still certain things I’m certain I want. Perhaps if I’m careful I can balance them all just right.
It always feels like the saddest thing to think that some lovely thing will Never happen again – or that a small floating dream you had will Never happen at all. The future is an infinite tree of possibilities, you could say, and as you move forwards every possibility not taken is sheared away forever.
I’m speaking some melancholy nonsense here, I’m sorry. I’ve been making a list of what I want to do, and it seems to grow much faster than I can check it off.
But I’ll keep at it.
Cake gets along with the haus folk quite well, and is happy to climb to their porch by roof or stairway whenever they congregate outside. Hannah opened the door to find Cake waiting there comfortably after we went inside; the cats have made the place a second home.
These interspecies friendships are important. It’s good for the heart to get a little nonhuman perspective.