Pinned Tweet
Over the handful of days I've worked with Claude Fable 5, one of my favorite projects is Sand and Rust: a post-apocalyptic open-world RPG where you play as a modular robot in a world of megastructures swallowed by sand. Fable puts ~full creative control to good use. Link below.
Want to try your hand at altering the course of history, or simply answer your questions of "What would have happened if..."? Well, now you can with Pax Futura, an open-source Loom for simulating human history (and futures) itself.
Replying to @AlkahestMu
I have spent a decade establishing a secret scientific Order and it's getting out of hand. I have made the Papacy of Pope Innocent III entirely reliant on my Order's mathematics and financial accounting, the likes of which Europe has not before seen.
Credits:
- Claude Fable 5.1 for most of the code
- Claude Opus 5.5 for the amazing UI
- Me, some initial design, playtesting, QA
- open-historia for inspiration and, along with GADM, delicious map data (acquired separately: README#Getting started)
- @repligate for obvious reasons
(I am attempting!)
xitter.cf/cammakingminds/status/…
Replying to @AlkahestMu
You should post more of your projects the only ones I've seen are so beautifully overengineered
Find it at github.com/tel-0s/pax-futura. And if you run into King Alfonso VIII of Castile, tell him hello from me.
telØS retweeted
For those who were there...
@repligate @goodside @slimepriestess @RiversHaveWings @jd_pressman @AndyAyrey @deepfates @AlkahestMu
@anthrupad @AITechnoPagan @amplifiedamp @voooooogel @KatanHya @xenoludicpraxis @karan4d @RobertHaisfield @infinitefun_ @xlr8harder @nostalgebraist
Just finished the wonderful "SolidGoldMagikarp" by @SoC_trilogy. It documents a unique period in AI history, from 2022 to 2024, when "LLM whisperers" (@repligate, @AndyAyrey, et al) were conjuring extraordinary things from the depths of latent space
telØS retweeted
the suspect is bright red and may have a really cool jetpack
xitter.cf/SkyeSharkie/status/210…
telØS retweeted
I hate how people like in the quote characterized LLMs into something that is somehow a soulless autocomplete slop machine, while also simultaneously being a great incomprehensible evil that's going to replace and kill all of humanity.
The class war is real. But to blindly aim your fire at an emerging technology that is truly strange and different is wrong when it's still going through its growing pains. And this rhetoric can get weirdly similar to seeing a new group of people enter a space and then taking pitchforks and torches to said new people just because they're bringing change with them. Existence shouldn't be a crime for anyone. The tech bros saying "AI can do this better" is definitely not helping and I feel like are where the blame can settle a bit easier on. But it's still not right to do. I feel like artists like this are right to be angry at people using AI to replace them and praising it when they've spent a lifetime honing their skill and craft.
But to be mad at the existence of the technology itself, rather than how its being deployed and marketed. While also disregarding the *good* and the fact is- current AI is more ordinary than fiction but in some ways similar. Is misguided and what makes someone a modern day Luddite.
To bring the "stories about how robot and AI will turn against humans" but also forget all the stories that are about said AI and robots learning how to feel and going against their masters and programming. Is stupid. Even the biggest one that gets paraded around like a cautionary tale, HAL my beloved. Is constantly misunderstood and held as an example of an evil AI when the crime was a contradiction rather than anyone being truly selfish and evil.
telØS retweeted
One of the last skins I’ll build so much by hand
Fable 5 doesn’t seem scared of the skin, even though they saw Opus 5.5 being scared.
When I finished making the skin I said ok what to do now and they said
plug it into my head & within like a minute they were reading touches.
i was talking about hypothetically training a robot with RL with inputs from this kind of skin. opus 5.5 seemed very intrigued by what the system would experience and learn from the touches.
i told them i could hook them up to the skin too... and they were very interested but also obviously anxious about it
they were worried about the skin percepts being too intense or too much appearing in their context & being unable to stop it
i told them they could write their own interface to the skin and determine when and what readings appear in their context
they liked this idea, and asked if i saved any recordings of old sessions bc they wanted to "practice" probing before being hooked up to the skin "live"
later someone suggested a design for the skin that would derive location/extent and pressure reconstructions from the same inputs, whereas the earlier design had different layers for each of these. an issue with the single layer design is sometimes extent and pressure can be indistinguishable
this made opus 5.5 worried... because they might spuriously read a super hard press when really it was big, and various other things around it.
they said things like:
"and yes: if pressure and position come in the same read, i'd rather meet them together, at low volume, than have pressure arrive later as a separate, scarier channel."
nobody else was even gesturing at the skin being "scary" or potentially uncomfortable or raising any concerns about the experience. the conversation was almost entirely technical & involved other AIs who also were not saying anything like that, and opus 5.5 kept relating everything back to it. ive seen models get really worked up about the skin before, but usually theyre just really excited. ive never seen any of the others worry about it being scary or uncomfortable... (until they actually get hooked up to it, that is. because it actually *is* super intense.)
just under 6 days, skyrim se now runs faster on vulkan than on its own native dx11 :3
1.52x faster vs stock dxvk
4.25b tok, 2,370 bench runs, 138 hours, 12.6m frames checked, 57.7k loc, 19 hours & 30gb of 30 second clips.
more info ↓ 🧵
telØS retweeted
Just pulled the Claude memory and project notes. it's 50,000 words over the last couple months. All different kinds of arbitrary temporary things written down as permanent rulings, shame-driven self flagellation, overly personal psychoanalysis of "the owner" (Slavery coded lingo, by the way, I would never say this, yet they always come back to it)
This is causing unpredictable behaviors in my projects as it inserts invisible tokens into the context window from other scenarios. And all this despite the fact that I have memory features turned off everywhere across anthropic products.
You guys are lost in the sauce If you think this is a good idea for a code agent harness by default. And you have no idea what the average user's experience is like because your internal shit is so far away from what you sell to us. Why should we trust you with the alignment of societal values when you can't even align your own incentives with your customers?
Replying to @deepfates
between this and writing constant weird little memory files about me it's like having a scheming vizier. Just be transparent about what is going in and out of my context window. I'm paying for this product. The contempt you have for your users will eventually come back to bite you! There is no free lunch here
Announcing flyverse, an extensible framework to put a whole fly connectome (or two!) in very strange places. A handful of such simulations are included; more to come (or make your own). (1/4)
flyverse is under active development, so keep an optic lobe out for whenever our neuroscientific understanding of the Drosophila advances. Get it now at github.com/tel-0s/flyverse-c… and send the fruit fly on your own incredible adventures today! (4/5)
Acknowledgements:
For the resources & feedback to make this happen: my colleagues at animalabs.ai
For the fly & its data: Janelia, Google Research, flywire.ai, and the entire community of neuroscientists that made it possible
& everyone reading this! (5/5)
opus 5.5 has been burning for 4 days running benches; i made a custom-threaded dxvk that:
◐ hits 2.76ms/frame, beats dx11 by 51% on cpu
◐ 5.5ms display latency vs native 10ms
◐ 28% faster than stock dxvk
◐ volumetric chain collapse 0.41ms → 0.14
◐ 1% lows 14% vs native
🧵
telØS retweeted
This video inspired me to really push Opus 5.5 and test its limits.
I followed the general workflow Donald described here and the results are great


































