Coté

Taken at some analyst forum in October, 2010. It was probably Adobe MAX that year.

I’m always trying to find how many apps are written (and, of course, running) in Java. Here are recent numbers I have on hand: “[N]early 70% of respondents say that more than half of their applications are built with Java or run on a JVM.” Azul survey, 2025. “98% of companies we surveyed use Java, with 57% saying it is the backbone of most of their application and infrastructure estate.

When I’m solo roleplaying D&D, or just farting around, I like to make pictures of characters and scenes. Sometimes I ask the robot to make me pages from artists sketchbooks, concept art. Here’s some recent pictures: Above based on this one from the Xoth player’s guide. Look how it made the new one’s décolletage more modest. For the next two, I wanted to use the ship map from the 2024 DMG, but put it on a very big, 100x100 ocean.

Most of these are human-made, two were given to the robot as guides. The video is me as a kid, maybe six or so. I gave it to Twitter’s video thing, and there you go. Fantastic and freaky.

Tasks should be approached with a balance of urgency and diligence to avoid mistakes and achieve optimal long-term results.

blogging as art!!! why not? i am drowning in information. all i want is a little fun. a respite from postmodernity. folk say the internet used to be fun. i was there, it kind of was. we can do better! 🔗 blogging as art

Here are some thoughts after a week of using Claude Skills. JasonJ in the SDT Slack says: I’d tl;dr Claude CLI skills as “low/no/english code MCPs'”. Seems helpful in the way that local utility scripts are today. That is a good way of putting it. I haven’t done heavy experimenting, but I think there’s a distinction between: (1) The Skill has no code, except maybe code fragments. Most of it is just a SKILL.

Excellent comparison of various LLMs doing AI image generations. Cool UI for looking at the output too. 🔗 GenAI Image Showdown

There’s a lot of knowledge work that can be automated: ‘Where Altman’s comment holds water is in what it hints at, even if it doesn’t spell it out. Most jobs aren’t fake, but many have accumulated layers of automatable junk: compliance checklists, reports nobody reads, emails summarizing meetings that could’ve been Slack threads. That’s the kind of “game-playing” work LLMs are already good at. When Altman says these models will wipe out tasks, not just roles, this is what he likely means.

What exactly does “thinking” mean? “Thinking harder”? Robin says: prospecting new analogies; sending your inquiry out away from the gravitational attractors of protocol and cliché; turning the workpiece around to inspect it from new angles; and especially bringing more senses into the mix." And, also that it is silent. We even have a phrase that shows the exception that makes the rule: “thinking out-loud.” Playing D&D with LLMs surfaces how different “thinking” is for AIs.