iiinova — where your digital self keeps evolving

Ivo @velitchkov.eu · 01/10/2026

Fair offer for a boat trip. I wish LLM services could offer something similar: "100% correct answer or your money back." In fact, it should be "money back doubled". With the boat, it's still a boat trip even without dolphins.

010

Ivo @velitchkov.eu · 01/10/2026

[[Stafford Beer]] 100 Festschrift book is out. I contributed a chapter on Autonomy & Cohesion, for which you can use my Metaphorum slidedeck as a TL;DR : kvistgaard.github.io/slides/ac/me... The electronic version of the book is free to download: staffordbeer100.com/festschrift

162

Ivo @velitchkov.eu · 30/09/2026

A wife-generated alla-prima oil portrait of @anneapplebaum.wsocial.eu

020

Ivo @velitchkov.eu · 29/09/2026

Programming used to be like literacy. Once you learned it, nobody could take it away. Now the craft comes on lease, metered, and on terms that can change overnight. This essay looks at seven orders of software makers, defined by what the lease did to them. www.linkandth.ink/p/the-new-co...

linkandth.ink

The New Coding Orders

Programming as a leased craft

021

Ivo @velitchkov.eu · 25/09/2026

Recent essays on Link & Think: www.linkandth.ink/p/air-rights www.linkandth.ink/p/second-law... Next up: The New Coding Orders

linkandth.ink

Air Rights

Stack capture through layer insertion

011

Ivo @velitchkov.eu · 25/09/2026

And the book is out. 📘 I have contributed a chapter on Autonomy and Cohesion (all new work, beyond what was in the A-C chapter of essentialbalances.com ) The e-book is free to download: staffordbeer100.com/festschrift

essentialbalances.com

ЕSSENTIAL BALANCES

022

Ivo @velitchkov.eu · 25/09/2026

Now you can also pin classes on top.

010

Ivo @velitchkov.eu · 24/09/2026

I meant design *idea*, of course.

000

Reposted by Ivo

Aaron Rupar @atrupar.com · 24/09/2026

Netanyahu after the UNGA emptied out ahead of his speech: "If there any other moral cowards who have not left the hall, please do so now."

4062388584

Ivo @velitchkov.eu · 24/09/2026

Credit for the multi-pane design goes to @andymatuschak.org . When I saw this UX in 2020, I thought it was already cool for wikilinks/backlings browsing (digital gardens and suchlike), but just imagine what it would do for RDF graphs

120

Ivo @velitchkov.eu · 24/09/2026

Opsis is a one-file faceted browser for [[RDF]] graphs. Every click is a SPARQL query. Fully data-centric. Even its settings are part of the graph. I'm using it daily to browse my local personal knowledge graph. Now it's available for everyone to use. #SPARQL #PKG github.com/kvistgaard/o...

140

Ivo @velitchkov.eu · 23/09/2026

While the research and analysis tracks have been progressing modestly but steadily for a year, the situation with software is different: an SOS (summer of software).

000

Reposted by Ivo

Luke Ontario @coolhand.bsky.social · 07/07/2025

Everyone stop, this is important

A black and white drawing of three dachshunds arranged vertically, each holding a different optical illusion object in its mouth, and the dogs themselves drawn as if a single line optical illusion. The top dog holds a Penrose triangle, the middle dog holds a Necker cube, and the bottom dog holds a Penrose trident. The dogs’ bodies are elongated in an exaggerated cartoon style. The artwork is signed “@IndieTheIntrovert” in the bottom right corner.

9354761701

Ivo @velitchkov.eu · 23/09/2026

Happy to be flying back home after a tiring conference limbo in the UK.

010

Ivo @velitchkov.eu · 16/09/2026

Even the Wikipedia article for AI Alignment (which is strange way to persue AI safety BTW) is heavily anthropomorphic, high frequency of "deceptive" etc, so the editorial norms and the distributed peer review failed. If scientific papers and W do that, we can't expect anything better from the rest.

040

Reposted by Ivo

Andy Matuschak @andymatuschak.org · 15/09/2026

✍️ Playing with programmable highlighters, e.g. * green highlight on a citation -> finds and prints the cited paper * orange highlight on a claim -> prints a skeptical fact-check

5676

Ivo @velitchkov.eu · 15/09/2026

My apps (some of them still idiosycrapps), if anything, have at least cool names: nodica, fractalino, bobinet, argus, opsis, krino, docket, quotawake

020

Ivo @velitchkov.eu · 11/09/2026

Fine, good points but why "misalignment"? And this is from mathematicians. mathandai.org

mathandai.org

Declaration — Math and AI

Read the declaration and add your name.

000

Ivo @velitchkov.eu · 08/09/2026

When a proprietary layer is inserted on top of a free protocol, the whole coordination stack is captured. It's like air rights over commons. Facebook and Amazon claimed the air rights over HTTP, GitHub – over git, FIFA – over the rules of the game. www.linkandth.ink/p/air-rights

linkandth.ink

Air Rights

Stack capture through layer insertion

031

Ivo @velitchkov.eu · 08/09/2026

One by one, I’m including different datasets stored on my computer into my [[PKG]]. One of the ways to access it is a simple faceted browser working only with SPARQL queries over Oxigraph endpoint. Now I added my HTML slidedecks, which was the easiest part since they were already JSON-LD annotated.

100

Reposted by Ivo

Melanie Mitchell @melaniemitchell.bsky.social · 08/09/2026

Hey @ericlipton.nytimes.com 👋 AI is not “evolving” on its own. *People* are choosing how to build it. *People* are choosing how to train it. *People* are choosing how to run it. Maybe the problem is that they are the *wrong people* to make these choices.

28635174

Ivo @velitchkov.eu · 05/09/2026

Of course, it is that. But I use anthropomorphisms as a shortcut, in this case of: "Could it be that the geometry of short-term thinking is such a pervasive pattern that it resulted in stable probabilities in the model?"

100

Ivo @velitchkov.eu · 05/09/2026

Opus tends to cheat more than Fable. When asked to solve a rather trivial problem with labels in the faceted browser of my personal knowledge graph, instead of properly querying the rdfs:label and displaying them, the first solution was to cut-and-stitch them from the local name 🤦‍♂️

130

Ivo @velitchkov.eu · 03/09/2026

LLMs are a Gaussian technology working in a Pareto world.

100

Ivo @velitchkov.eu · 02/09/2026

Doing a personal data catalogue based on DCAT is probably overkill, hopefully useful, but certainly fun.

010

Ivo @velitchkov.eu · 02/09/2026

@melaniemitchell.bsky.social I somewhat hoped that SFI publications would offer sharing on Bluesky, if not instead of X, at least in addition to, but no.

010

Ivo @velitchkov.eu · 31/08/2026

The decision graph starts to pay off after a couple of months of use. Three different ways of accessing depending on the need: Mentor notebook, currently containing 30 SPARQL queries, a faceted browser, and for visual exploration, I use Reactordia. I'll share more soon.

010

Reposted by Ivo

The Big Meowski @catovitch.bsky.social · 29/08/2026

Stop using this unsafe garbage before the entire world starts pwning your machines and putting backdoors in everything. It will never be safe. It cannot be safe. It's part of the stupid design of this stupid bullshit. Just stop using it.