vessenes

29 ms·

661.
by vessenes 9mo ago
Tao’s doing a lot of related work in mathematics, so I can say that first of all literature search is a clearly valuable function frontier models offer. Past that, A frontier LLM can do a lot of critiquing, a good amount of experiment desig
662.
by vessenes 9mo ago
Hmmm. NFL revenue continues to grow, with something like half of it from media licensing. So, I think you’re wrong. My son watches NFL on his laptop, but there’s little to distinguish how he watches it from the experience he would have on a
663.
by vessenes 9mo ago
I wish it were chess boxing. I love chess boxing.
664.
by vessenes 9mo ago
I think of football as one of the killer apps for tv. Baseball is one for radio: baseball is almost always better over radio; there’s lots of space to do something else at the pace of most baseball games.
665.
by vessenes 9mo ago
Don’t forget replication!
666.
by vessenes 9mo ago
You can read about this basically everywhere - the term of art is agent orchestration. Gas town, Claude’s secret swarm mode, or people who like to use phrases like “Wiggum loop” will get you there. If you’re really lazy - the quick summary
667.
by vessenes 9mo ago
? They prefer the branding. The license just says you have to say it was them if you make > $250mm a year on the model.
668.
by vessenes 9mo ago
Steve runs beads across like 100 coding environments simultaneously. So, you need some sort of coordination, whether that's your db or a daemon. Realistically with 100 simultaneous connections, I would probably reach for both myself.
669.
by vessenes 9mo ago
Good summary. Upshot: Steve thinks he’s built a quality task tracker/work system (beads), and is iterating on architectures, and has gotten convinced an architecture-builder is going to make sense. Meanwhile, work output is going to im
670.
by vessenes 9mo ago
Yep, it works. Like anything getting the most out of these tools is its own (human) skill. With that in mind, a couple of comments - think of the coding agents as personalities with blind spots. A code review by all of them and a synthesis
671.
by vessenes 9mo ago
It's 100% both sides. We haven't had a president work to roll back his own power, since ... Hmm. Maybe Gerald Ford? I guess Carter was fairly principled on some of this. This part of the system - executive power grabs - is suppose
672.
by vessenes 9mo ago
A lot of this depends on your workflow. A language with great typing, type checking and good compiler errors will work better in a loop than one with a large surface overhead and syntax complexity, even if it's well represented. This i
673.
by vessenes 9mo ago
Those are sort of like 2D splats - if only they'd thought to make it all differentiable!
674.
by vessenes 9mo ago
Yes he is on an extended manic episode right now - we can only sit back and enjoy the fruits of his extreme labor. I expect the dust will settle at some point, and I think he’s right that he’s on to some quality architecture.
675.
by vessenes 9mo ago
Looks like solid incremental improvements. The UI oneshot demos are a big improvement over 4.6. Open models continue to lag roughly a year on benchmarks; pretty exciting over the long term. As always, GLM is really big - 355B parameters wit
676.
by vessenes 9mo ago
Probably the wrong attitude here - beads is infra for your coding agents, not you. The most I directly interact with it is by invoking `bd prime` at the start of some sessions if the LLM hasn’t gotten the message; maybe very occasionally ru
677.
by vessenes 9mo ago
It not only sort of works, the 10% of the time it works surprisingly well at scale! Tantalizing.
678.
by vessenes 9mo ago
Counterpoint - you can go much faster if you get lots of people engaging with something and testing it. This is exploratory work, not some sort of ivory tower rationalism exercise, (if those even ever truly exist), there’s no compulsion inv
679.
by vessenes 9mo ago
It is worth an install; it works very differently than an agent in a single loop. Beads formalizes building a DAG for a given workload. This has a bunch of implications, but one is that you can specify larger workloads and the agents won’t
680.
by vessenes 9mo ago
Not just this, but I’ve been thinking that naming things with aggressively strong connotations might help Claude get out of ‘nice/helpful’ mode. “You are the Deacon, grrrrr”. So there might be actually be a bit of effectiveness added b
681.
by vessenes 9mo ago
Very minor nit -- crew could be a person also - in fact that's how you're supposed to hack on a codebase in gas town directly - add yourself as crew. Other than that, this is a helpful list especially for someone who hasn't b
682.
by vessenes 9mo ago
Reviewing Hugos last 20 years or so: Ann Leckie is great, Tamsyn Muir is gothic horror lesbian necromancer-ish and super cool, Andy Weir is a full humanist in the best sense, Arkady Martine is a thoughtful writer I enjoy, Yoon Ha Lee was a
683.
by vessenes 9mo ago
Salvatore - this is cool. I am a fan of using Steve Yegge's beads for this - it generally cuts the markdown file cruft significantly. Did you run any benchmarking? I'm curious if python's stack is faster or slower than a pure
684.
by vessenes 9mo ago
Some problems are ‘uninteresting’ in that they show results that aren’t immediately seen as useful. However, solutions may end up having ‘interesting’ connections or ideas or mathematical tools that are used elsewhere. More broadly, I think
685.
by vessenes 9mo ago
… “but I guess it was able to formalize it in Lean, so…”
686.
by vessenes 9mo ago
Thanks for the curious question. This is one in a sequence of efforts to use LLMs to generate candidate proofs to open mathematical questions, which then are generally formalized into Lean, a formal proof system for pure mathematics. Erdos
687.
by vessenes 10mo ago
my bad - I read but didn't watch. okay now I want the latter
688.
by vessenes 10mo ago
One rough edge for me: the cowork interface seems to have turned off “extensions” - my first ask was to read some emails and compare with some local documents and draft a document. It kept trying to use claude chrome to navigate to gmail. I
689.
by vessenes 10mo ago
I wasn't sure whether this was emulating the sewing machine, or driving an actual sewing machine from the emulator -- it's the latter. Followup request - digital emulated sewing machine! Should be a breeze based on the blog!
690.
by vessenes 10mo ago
As a user of a lot of coding tokens I’m most interested in latency - these numbers are presumably for heavily batched workloads. I dearly wish Claude had a cerebras endpoint. I’m sure I’d use more tokens because I’d get more revs, but I don

More ›