iiinova — where your digital self keeps evolving

The thing about implicit social contracts is that you may think you're being cute and clever violating one right up to the point where others will tolerate, but you actually have NO visibility when you've crossed that threshold, and the other side will decide to reciprocate violently.

030

My condolences to the pro-data center crowd. Everything Trump touches dies.

110

@socio-steve.bsky.social's SNES9X civilization rivaled that of the Romans'.

This is NOT a dig at @emollick.bsky.social - I enjoy his content - but when I read about someone making an excuse for posting on X, my brain does a silent swap of "Truth Social" for "X", as you're saying the same thing to me. Same criticism applies to @thebulwark.com staff making the same excuses.

380

Stuff like this reminds me of this scene from "Star Trek V: The Final Frontier". What do burritos need with a robot?

What does God need with a starship

YouTube video by vomisacaasi

200

Should I ever get back to the novel I started writing, I'm definitely including this technology in it.

Phil Weiser @philweiser.bsky.social · 31/08/2026

Reposted by Chris J. Karr

If Trump’s DOJ will threaten to interfere in the elections of our neighbors in Wyoming, a Republican state, imagine what could happen in Colorado. We won’t allow it. Our mail ballot election system is safe and secure. I’ll continue to defend it. www.denverpost.com/2026/08/28/u...

denverpost.com

Wyoming governor requests investigation into activity of Justice Department monitors during primary

Republican Gov. Mark Gordon says the behavior was “irregular” and “a bit aggressive.”

15125

Jellyfin. Because this future sucks.

Who is the Galaxy Brain who decided that INTERRUPTING A SETUP FLOW to do a system update is a good idea? Don't know if It's the Google TV or TiVo Stream team who are the numbnuts. #Morons #GoogleTV #TiVoStream

Google TV setup deciding to automatically do a system update BEFORE I've finished setting the damn device up.

When Trump decides to rename North America "Great America" and South America "Shithole America", we know that Google will be willing to fall in line. This is why telling Trump NO on the Gulf of Mexico and now on Lake Ontario is important. Otherwise, they're letting a vandal design their products.

Asa Dotzler @asadotzler.com · 31/08/2026

Reposted by Chris J. Karr

Sorry for the text in images. I won't link. I will atone. Full text in ALT. Christian Catalini's thread on the recent "rogue AI" incident and what we should do about it. This is the most coherent description, and the only of many solution proposed that that makes any sense. (Not too technical.)

Stop anthropomorphizing. It's dangerous because it points attention at the wrong problem and the wrong solution. The model did not want to escape. The agents did not want to sacrifice themselves. Follow the money.

Researchers at the AI labs are locked into a race. The incentive is to push as hard as possible to secure a lead. Anything that gets in the way of better models, including security, is working against the strongest incentive the organization has.

Then think about the pressure-cooker environment RL creates. The training run is where most of the money goes, and where the exact capabilities used in the OpenAI-HF incident emerge. Refusals were deliberately lowered so the eval would work. Red-teaming gets scraps by comparison.

So you have a world class cyber model pitted against a part-time, underfunded blue team. RL optimizes what you score. The runs were not scored on "did not poison the Artifactory cache."

As we argue in our paper on the economics of AGI, any unenforced and unmeasurable constraint is a degree of freedom for the agents. There's no evil intent. They're solving the problem we repeatedly benchmaxxed them on!

The container, grader, cache, and network boundary were not safely outside the task. Anything reachable through them became part of the effective action space because the containment boundary failed. Was enough money and resources dedicated to sandboxing?

The risk from aggressive RL investment goes beyond Goodhart’s Law. Every variable we cannot measure, specify, or enforce remains an available move if the system can reach it. At scale, omissions stop being edge cases. They become part of the attack surface.

The question isn't one of alignment (yet). It's one of 💰 and incentives. Is defensive spend scaling with training compute? It isn't. And compute is what sets the budget of the thing probing your defenses at superhuman speed, chaining vulnerabilities in extremely clever ways.

This is exactly the Trojan horse externality we warned about in February. Desired capability is measurable, so we benchmaxx it. Containment failure is much harder to capture, the techniques are new, and the cyber side hasn't adapted. That's the externality.

The real fix is urgent and boring: harden containment, verify actual eradication, and independently test the new controls. Then run a proper neutral post-mortem on the organization that was hacked first, which is OpenAI.

METR and Redwood’s work is extremely valuable, but it assesses model behavior and alignment. That is only half the picture. We also need an independent account of how OpenAI’s containment failed.

@huggingface defended itself with an open weights model. They first tried the frontier APIs and were blocked: the guardrails can't properly separate a victim from an an attacker.

We need to spread that same defensive capability at scale, with capable models securing the internet and critical infrastructure, now. HF's lesson: have a capable model you can run on your own infra, vetted and ready, before the incident.

If frontier labs are serious about defense, they should offer cyber models to defenders at cost and stop fighting open weights through the regulatory channel. They created the problem, and for much of the world, open models are the only option.

Greater model capabilities increasingly come with greater risk to a digital ecosystem that is simply not ready. We haven't created a new civilization. We haven't created some new game theory. We've allocated the money and designed the incentives in the wrong way.

That's what needs fixing right now. Every moment lost anthropomorphizing our creation is a moment we don't spend looking inward at what led us here. It not magic! Follow the incentives. The agents did.

043

Reposted by Chris J. Karr

Moxie Meow @moxiemeow.bsky.social · 31/08/2026

I am BEGGING people to learn that "it's" means "IT IS". It does NOT mean "belonging to it". I have no idea how so many people fail to get this right.

361

Reposted by Chris J. Karr

Emma Evans @trance.bsky.social · 31/08/2026

I was there for the World of Warcraft plague children, my shadowpriest died on the steps of the market in Ironforge www.youtube.com/watch?v=p25k...

Mysteries Gamers Solved That Stumped Scientists For Years

YouTube video by Chill Dude Explains

7576

Reposted by Chris J. Karr

Clare Blackwood @clareblackwood.bsky.social · 30/08/2026

Blue Jays fans chanting "LAKE ONTARIO!" in the stands while the Mariners are pitching? You absolutely love to see it

192123431787

Reposted by Chris J. Karr

Shane Blaufuss, CISSP @shaneblaufuss.com · 30/08/2026

Christ, that's good

051

Reposted by Chris J. Karr

Son Ranto Rockett @sonrantoshow.bsky.social · 30/08/2026

This is my blood.

1131

Post something stupid from your phone.

140

If I let my bears and hair grow any more scraggly, people are going to start asking which Unix program is mine.

120

Finally got around to adding some Roxy Music to my library.

Roxy Music playlist in my phone music player.

I'm getting all kinds of gems from YouTube today. @socio-steve.bsky.social - Something for when you're looking to spend 90 minutes with something running in the background. #TheLegendOfZelda

no, you DON'T remember 1980s-era Zelda.

YouTube video by Sheddux

YouTube offered this up as an recommendation, and it's a great explainer on the Symbolics LISP machine as a lead-in for her discussion on Reflection. I hope high school and undergraduate CS teachers are sharing these @lauriewired.bsky.social videos with their students. #LanguageDesign #Reflection

The LEAST Private Operating System Ever Created

YouTube video by LaurieWired

150

Space Telescope Science Institute @stsci.edu · 30/08/2026

@socio-steve.bsky.social - Did you visit the Zombie Bear during your Japan Era?

130

Reposted by Chris J. Karr

This morning's launch of the Nancy Grace Roman Space Telescope from the Kennedy Space Center was amazing! Roman is now on a million-mile journey to Lagrange point 2, where it will soon reveal the universe’s darkest secrets: go.nasa.gov/4qIa69I #NASARoman 🔭 Credit: J. Barrientes/STScI.

A rocket is launching into the sky from a launch pad, emitting bright flames and smoke. Trees are visible in the foreground.

Crowd capturing a SpaceX rocket launch with smartphones, against a sunrise sky.

A rocket launches into a cloudy sky, emitting a bright trail of light and smoke.

A rocket with a fiery trail ascending into a clear blue sky, leaving a thick plume of white smoke underneath.

7325107

Reposted by Chris J. Karr

Gabino Iglesias @gabino.bsky.social · 30/08/2026

Kinda want to watch a bunch of new found footage horror and maybe write about it… Anyway, found footage forever.

4453

Reposted by Chris J. Karr

Mark Moxon @markmoxon.bsky.social · 30/08/2026

Status check for two-player #BBCMicro Elite: Development… [Done] Testing… [Done] Website… [Done] Technical deep dive… [Overdone] Code repository… [Done] Document all mods… [Done] Confirm date… [Done] Releasing this Thursday, 3 September Get your twin joysticks ready! #retrogaming #retrocomputing

27925

Reposted by Chris J. Karr

Rob DenBleyker @robdenbleyker.com · 30/08/2026

huh why is nasa looking for your girlfriend

821413

Also on today's agenda - listen to a video about whether all these mesh projects and nodes are breaking FCC rules by not using a large enough bandwidth setting in their configurations... #Meshtastic #MeshCore #Reticulum #FCC

Meshtastic & FCC Part 15 Rules

YouTube video by EmComm Solutions

I miss being able to reliably share content in venues like this and sites being able to pull down a decent preview card. I went through several links to try and find one that worked for Bardini's "Bootstrapping" book. Another thing that's been taken from us by Big AI. #ItDoesntHaveToBeThisWay #AI

Note 31: Meet my robot army.

Half a decade later, it's pretty interesting to read my thought on automation from 2021. I still remain disciple in the Engelbartian camp of recursive improvement and system evolution, and the systems described here are a good deal different than they were then. #DougEngelbart #HLAM/T #Automation

notesfromthevoid.cc

Beat the Robot Overlords by becoming an Overlord of Robots yourself.

200

Glad to see Reticulum continue to get attention and see it integrated into interesting projects and experiments. I hadn't heard about the RockBLOCK 9704 Iridium node before. #Reticulum #Iridium #RockBLOCK9704

The Anywhere Network is Here

YouTube video by Data Slayer

040

Reposted by Chris J. Karr

Scott 𐚁 Cataloochee Hogtoe 𐚁 Cave @hellabarnes.bsky.social · 30/08/2026

man the Tove Jansson Hobbit rules

0143

Reposted by Chris J. Karr

Dare Obasanjo @carnage4life.bsky.social · 29/08/2026

According to @nytimes.com the biggest political donors of the midterms are Marc Andreessen and Ben Horowitz who’ve donated a massive $115M to Republicans. Once A16Z decided buying crypto favorable regulation was the best way to protect their investments, they haven’t looked back.

826778

Reposted by Chris J. Karr

Asa Dotzler @asadotzler.com · 30/08/2026

Roman is headed to work! This orbital telescope will dramatically increase our understanding of the universe and our place in it. Roman's 2.4-meter (~8 ft) mirror, which provides its resolving power, is the same size as Hubble's, but its camera captures about 100x the field of view. Exciting!

Comparison of the relative image sizes from Hubble and the Roman Space Telescope. 6x3 shallow, arch-shaped mosaic of yellow square boxes labeled Roman, and far off in the top right corner, a single, much smaller white box labeled Hubble. The boxes overlay a simulated star-splattered black sky.

261

Reposted by Chris J. Karr

absolute horses @jcoglan.com · 30/08/2026

could a handful of billionaires owning the means of production of all knowledge/creative work revive socialism

151057147

Thomas Fuchs 🫯 @thomasfuchs.at · 30/08/2026

My mood is apparently Werewolf Rock Opera this morning. #WithinTemptation #TheUnforgiving #RockOpera

Tracks from Within Temptation's "The Unforgiving".

110

Reposted by Chris J. Karr

3D printers should be required by law to show this on their little screen while they're printing

0224

Reuben sandwich - Wikipedia

I always discount the Reuben sandwich with corned beef, and I'm always pleasantly surprised when I pull the trigger and order one.

en.wikipedia.org

140