SafePen 🦑

If you’re not using LCM (lossless-claw) from @jlehman_ on your @openclaw setup you are missing out, it is awesome! I’ve done 5 upgrades of openclaw (including rolling back 2026.3.22) multiple system reboots & my claw hasn’t missed a beat, full memory of what we’ve been doing. 🦞

Claude Code deletes your context every time it compacts. I ported the DAG architecture from @jlehman_'s lossless-claw into a Claude Code plugin that fixes this permanently. Every session is preserved. Zero token waste. Free. Full write-up: x.com/dewaldt_h/stat… github.com/GodsBoy/lossle…

i envision a world without duplicates

@kail_designs Is this open source?

Insane🤯 Ascii effect made using ascii-magic.com

That baby gonna poop poop.

Lossless Context in OpenClaw feat. Josh Lehman x.com/i/broadcasts/1…

@jpt401 that is a very good strategy lol. yes.

@SafePen Try spamming FABs, with occasional drone swarms to kill straggler ships.

github.com/jpt4/hormuz A Strait of Hormuz tower defense game, programmed primarily by machine, but with enough human input, especially play-testing, to qualify as a hybrid/centaur development style. I consider it now playable, and at least interesting, but no guarantee as to being fun, bug-free, etc.; comments/critique/improvements welcome. Desktop recommended.

El Fish is Spanish for "The Fish"

I realize that I forgot to mention that I’ve joined the @openclaw maintainers team (as of a little over two weeks ago)! Too busy drinking from the firehose. Incredibly honored and excited to be working alongside one of the coolest and most competent teams in OSS.

@Geoffiam @hewliyang mine also did google docs which are way more fiddly.

@hewliyang @SafePen Doesn't Microsoft words own copilot plugin not even do this?

OpenWord v0.0.1 is out it is competent at most text editing operations, can insert images into your doc, has vision, can add & reply to comments, and even works with Track Changes (redlining)

@Geoffiam @hewliyang Correct - also I built something that did 80% of this about 5 month ago

@sigilante I'm trying to decompile el fish and then rebuild it for the modern era.

Some things I am interested in doing lately: DSLs for AI pipelines, formal verification of AI outputs, constrained generation, better primitives, full stack (down to metal) secured code verticals.

A story in three parts. LCM "genuinely is going to change my life."

New background for video calls.

You wanted it, you got it. Here is the first ever conversation between Nick Land and John Michael Greer! As the Lofty Powers ordained by solemn providence, this discussion took place today on Pi Day, which just so happens to be Nick’s birthday. Enjoy! Link in comments.

I am interested in how you would describe this: A long-living computer that only, slowly, computes primes. We'll call it C-prime. The death of the creator species after it is set off computing. C-prime computes at least 1 additional prime since death of creator species. An alien species discovers C-prime and eventually interprets its function, understanding it is computing primes, and understanding the last valid answer it found since the death of its creators. Was C-prime computing during the interval between the death of the creators and the eventual interpretation? Was it only so retroactively upon interpretation? Did it start computing again upon interpretation but, in between, it was just physics?

Since publishing "The Abstraction Fallacy," the most common pushback relies on a deeply ingrained but flawed metaphor: the assumption that the brain is a computer. Here is a brief clarification on why it is not, and why this matters for computational functionalism. 1/5 philpapers.org/rec/LERTAF