2026-01-31 00:31:51
My buddy Paul1 sent me a link to this macOS app a few weeks ago, but I didn’t have a use case for it until yesterday.
I was reading a few different documents relating to a machine learning contract I just started, and I was getting overwhelmed by the mountains of text on my screen.
Monocle is a very simple utility for macOS that blurs out every app window that is not in the foreground. If you want to see everything, you can shake your mouse back and forth2. If you want it blurred again, you do the same shake gesture.
It’s the best $9 I’ve spent on software in years.
2026-01-30 00:09:48
The most annoying thing about corporate surveillance to me is the arrogance of the prediction mechanisms. These algorithms build a model of me based on my clicks from three years ago and then try to trap me in that loop forever. They show me music they think I'll like, and news they think I'll engage with, and videos they think will enrage me enough to keep me hooked to their platforms. They are actively trying to flatten my personality into something easy to monetize.
As most people I've seen say out loud, "Privacy as a concept is way beyond hiding secrets. A part of it also means preserving your capacity to change. To be surprised. To be inconsistent." If I could tell every human one thing, it would be to actively refuse to be a predictable data point.
Mess up their metrics. In whatever way you are capable of.
2026-01-27 22:25:30
Apropos of nothing going on in my backyard right now, these fonts are dope.
2026-01-27 22:21:47
What's going on in those vast oceans of GPUs that enables providers to give you a 10x discount on input tokens? What are they saving between requests? It's not a case of saving the response and re-using it if the same prompt is sent again, it's easy to verify that this isn't happening through the API. Write a prompt, send it a dozen times, notice that you get different responses each time even when the usage section shows cached input tokens.
Not satisfied with the answers in the vendor documentation, which do a good job of explaining how to use prompt caching but sidestep the question of what is actually being cached, I decided to go deeper. I went down the rabbit hole of how LLMs work until I understood the precise data providers cache, what it's used for, and how it makes everything faster and cheaper for everyone.
After reading the Joan Westenberg article I posted yesterday, I decided I’m going to read more technical articles and focus my attention on them.
This post from the ngrok blog was very helpful in explaining how LLMs work up through the attention phase, which is where prompt caching happens.
It also got me to go down a rabbit hole to remember how matrix multiplication works. I haven’t heard the phrase “dot product” since high school.
2026-01-27 07:17:16
Every attempt to discuss the problem becomes another piece of content, another take, another entry in the engagement competition that makes the problem worse. I'm aware that this essay is doing exactly that. I'm aware that you, reading this, are spending cognitive resources on yet another analysis of the discourse when you could be spending those resources on something more important. I'm sorry. I'm not sure what else to do.
What I do know is that the feeling of being overwhelmed, of never being able to keep up, of having strong opinions about everything and confident understanding of nothing, is not a personal failing. It's a predictable response to an impossible situation. Your brain is being DDoS'd, and the fact that you're struggling to think clearly under that onslaught is evidence that your brain is working normally. The servers aren't broken. They're overloaded. And until we figure out how to reduce the load or increase the bandwidth, the best any of us can do is recognize what's happening and try, when possible, to step away from the flood long enough to do some actual thinking.
Find some topic you care about. Just one. Resist the temptation to have takes on everything else. Let the discourse rage without you while you spend weeks or months actually understanding something. Read books about it, not takes. Talk to experts, not pundits. Follow the evidence where it leads, even when it's uncomfortable. Change your mind when you find you were wrong. And when you finally have something to say, something you've actually earned through careful thought rather than absorbed from the tribal zeitgeist, say it clearly and then step back.
2026-01-18 13:18:42
Here’s the paradox that makes this pattern particularly poignant. We’ve made extraordinary progress in software capabilities. The Apollo guidance computer had 4KB of RAM. Your smartphone has millions of times more computing power. We’ve built tools and frameworks that genuinely make many aspects of development easier.
Yet demand for software far exceeds our ability to create it. Every organization needs more software than it can build. The backlog of desired features and new initiatives grows faster than development teams can address it.
This tension—powerful tools yet insufficient capacity—keeps the dream alive. Business leaders look at the backlog and think, “There must be a way to go faster, to enable more people to contribute.” That’s a reasonable thought. It leads naturally to enthusiasm for any tool or approach that promises to democratize software creation.
The challenge is that software development isn’t primarily constrained by typing speed or syntax knowledge. It’s constrained by the thinking required to handle complexity well. Faster typing doesn’t help when you’re thinking through how to handle concurrent database updates. Simpler syntax doesn’t help when you’re reasoning about security implications.