brookst

17 ms·

61.
by brookst 12d ago
To conspiracy theorists, a particular theory making no sense is strong evidence that it’s true. It’s the sensible things that are obviously false.
62.
by brookst 12d ago
If you find evidence that these models are capable of stateful, long-term strategic planning… please post links.
63.
by brookst 12d ago
“Supposed to” by who? Claude code communicates between sessions. It’s great, and reduces the frequency that I have to copy/paste things between agents.
64.
by brookst 12d ago
Some people thinks it makes them sound smart when they always have the inside line on what’s really going on. With these people, it’s never just a power outage during a windstorm, it’s proof that [insert far more complex and unlikely scen
65.
by brookst 12d ago
Not to mention the noise-over-signal of asserting that anyone who disagrees is a shill / sheeple / whatever who is “falling for marketing” as if it is literally impossible for a knowledgeable person to disagree on good faith. I r
66.
by brookst 12d ago
I don’t think stalking laws cover it. Laws are largely about intent and I don’t think anyone could say that Google is planning to attack everyone. Law isn’t code. You can’t reduce stalking laws down to “it’s illegal to track people” and the
67.
by brookst 12d ago
Reminiscent of race to sleep: not suitable for sustained workloads, but for bursts ones it’s a good approach.
68.
by brookst 13d ago
Huh? Can you link to where condescending sycophancy and hallucination were intentional design choices?
69.
by brookst 15d ago
You’re more right than you probably realize!
70.
by brookst 15d ago
Some of us are motivated by interest in facts. I get that others are financially motivated, or focused on sports-style love/hate for organizations. I can’t imagine coloring my perceptions or actions to ensure they only attack big compa
71.
by brookst 16d ago
Sure. But thinking that means every article is a complicated psyop is classic correlation / causation. If you’ve hot any occam’s razor left in you, consider: 1. Apple is a masterful media manipulator, carefully planting stories like th
72.
by brookst 16d ago
Sure, that’s why TM-30 is better. Black body will probably always be better, but people unhappy with LED often have CRI < 80 LEDs when even 95+ isn’t necessarily great.
73.
by brookst 16d ago
I love this. Thanks for sharing. Would be great to generalize to documentation in general, not just slides. My LLM workflow is documentation-heavy, and that works great for getting projects off the ground and managing very large projects. B
74.
by brookst 16d ago
What an insightful response. I thought about that for a while, and its a great way to reframe some of this debate about the desirability of using AI.
75.
by brookst 16d ago
Someone else hit on it: a lot this discussion is about whether we as individuals see collection and maintenance of pure knowledge as the goal, or achievement of outcomes. I’d say both are valid. I’m definitely an outcome person. I like maki
76.
by brookst 16d ago
I appreciate your reply and framing, but it strikes me as a spirited opposition to delegation. Or at least a strong assertion that AI use is similar to managing an organization. The reins metaphor is also interesting and cuts both ways: rid
77.
by brookst 16d ago
It hugely depends on the quality of the light. Look at CRI (and the newer, better TM-30). Low quality LED lighting is just terrible, the room appears light but colors and details are poor.
78.
by brookst 17d ago
To the extent code is artistic expression, sure, I get what you're saying. But when you sit down at an awesome synth and start jamming, do you care whether the firmware was written by a real live human? It's forest/trees.
79.
by brookst 17d ago
No, waterfall is an engineering process. What I’m describing is thorough documentation of requirements (acceptance criteria, if you like), and then encouraging Claude to be agile in execution. As long as the outcome is well-defined, it is e
80.
by brookst 17d ago
I’ve never been a cram for the test kind of person. I do best when using knowledge in an ongoing basis. Regardless of AI, things I studied intensely a decade ago but never use are by and large gone, whereas things I use frequently stay and
81.
by brookst 17d ago
This, after decades of phone trees, mail-in rebates, specialist pre-approvals, and other processes specifically designed to create friction for typical people when interacting with large companies and organizations. I have no moral judgment
82.
by brookst 17d ago
Hear hear. I do not understand the people who bemoan the loss of slide rules and books of trig tables in the era of graphing calculators. I’m with you — I’m producing more and better art, with a deeper understanding of the art which comes f
83.
by brookst 17d ago
It’s funny, I’m the exact opposite. I’ve tackled projects at scales I could never have done before, and learned more than I would have. My actual coding ability is probably lower, but I’ve learned so much about music theory and production,
84.
by brookst 17d ago
Next year has been the year of the Linux desktop for as long as I remember.
85.
by brookst 17d ago
And it always will be!
86.
by brookst 18d ago
I see a significant difference between making a security feature available but opt-in during a teething phase, and removing the feature altogether. Am I missing something?
87.
by brookst 18d ago
Yep. I do so much work with Claude on product requirements and architecture before a single line of code is written, and test coverage is extreme. If I need to understand a specific line of code it means I did something wrong in planning or
88.
by brookst 18d ago
I’m genuinely curious why it matters to make abstract moral judgments about the differences between remembering function names and deep understanding of the syntax used in a function. Why don’t we just care about results? Code quality, pe
89.
by brookst 18d ago
There’s certainly an analogue to package managers, which made it easy to pick a random package rather than writing the code. We survived. Some upsides, some downsides, life continues.
90.
by brookst 18d ago
Yep, even with cache hits that like submitting uncached 80k-token prompts over and over. My Claude.md + framework stuff has it tell me when it recommends /clear before the next prompt (and write itself handoff notes) and I rarely see &

More ›