iiinova — where your digital self keeps evolving

Andreas Keller @nannus.bsky.social · 36m

Like all MAGAs they have a distorted world perception. That also applies to their own faces.

Andreas Keller @nannus.bsky.social · 06/10/2026

From the small sketchbook of my great-great-aunt Mathilde Pajeken (1842–1913) #12: A houseboat at a wooden boarding pier. I decipher the inscription as “Am Sande, Steg 5” (?) ("At the Sand, pier 5"), but I am not sure about that. I think this is somewhere along the Weser river.

A houseboat at a wooden boarding pier. In the background are trees and bushes, reeds, and a dike.

020

Andreas Keller @nannus.bsky.social · 04/10/2026

Just listening to: Amadinda music from Uganda. Both players play parallel octaves on a xylophone tuned in an equidistant pentatonic, interlocking their beats. The listeners hears fast inherent patterns combined from both players: www.youtube.com/watch?v=Wh9D...

youtube.com

Buganda Music Ensemble - N'agenda Kasana - The Singing Wells project

YouTube video by Singing Wells

020

Andreas Keller @nannus.bsky.social · 03/10/2026

The way to make "Trickle down" actually work would be high taxes on high incomes and large forunes. It does not come by itself, it must be enforced!

000

Andreas Keller @nannus.bsky.social · 02/10/2026

"Learning algorithms" are algorithms. As such, they are UNCREATIVE by definition, so the reshuffling they do with the input data is uncreative and thus always PLAGIARISM. If the input was open domain, the result should be open domain as well. If not, it belongs to the people whose data went in.

011

Andreas Keller @nannus.bsky.social · 01/10/2026

The quote is a little bit out of context, but it fits 🤣.

000

Andreas Keller @nannus.bsky.social · 01/10/2026

Oh, we have a dive again! 😀Hello Lisa!

110

Andreas Keller @nannus.bsky.social · 27/09/2026

The work of a true nerd! 😀

170

Andreas Keller @nannus.bsky.social · 26/09/2026

Is the green looking thing in the lower part of the body the content of the stomach? And the structure above that the mastax?

110

Andreas Keller @nannus.bsky.social · 26/09/2026

"Trickle-down" has never worked. To make it work, you have to implement it. Implementation involves much higher taxes on large fortunes, high incomes, and large inheritances. Without enforcement, the rich will grab everything they can for themselves, i.e. they will implement “stream-up” instead.

020

Andreas Keller @nannus.bsky.social · 24/09/2026

Thanks. Where I live, C is the standard. Interesting recipe. I'll have to try it sometime.

010

Andreas Keller @nannus.bsky.social · 23/09/2026

In ROV dives (Schmidth Ocean, EV Nautilus etc.), you can quite often see larvacean houses (or the remains of them), sometimes with the animal at home. Fascinating creatures.

000

Andreas Keller @nannus.bsky.social · 23/09/2026

I guess because he cannot bend since his spine is short, movements like that are a more efficient way for him to turn.

000

Andreas Keller @nannus.bsky.social · 23/09/2026

You can still do it. We Europeans are waiting for exactly that to happen.

000

Andreas Keller @nannus.bsky.social · 23/09/2026

Der Unterschied ist, in solchen Fällen wurden Bürger für die Interessen von großen Firmen enteignet. Jetzt sollen große Firmen im Interesse von Bürgern enteignet werden. Das gefällt Merz und anderen Lobbykraten nicht.

050

Andreas Keller @nannus.bsky.social · 23/09/2026

My argument is connected to Gödel. It comes out of computability theory, which one could also call formalizability theory since it describes the limits of what can be formalized. I think formalization does not reach as far as some analytic philosophers think. Cognition is unformalizable.

010

Andreas Keller @nannus.bsky.social · 23/09/2026

What you need is an entity that produces such an outside reference and is not controlled or triggered from inside the algorithm. It must run asynchronously to it. You can compare it to mutations in genetics. Producing the reference from outside, the true meta level, is like a mutation.

020

Andreas Keller @nannus.bsky.social · 23/09/2026

As long as you can describe that as an algorithm, it is necessarily limited. What an algorithm cannot do, in principle, is producing a reference to itself as a whole, from where it could describe and modify every aspect of itself. Somehow, we seem to be able to do that.

211

Andreas Keller @nannus.bsky.social · 23/09/2026

The idea of the proof I have sketched there only depends on the system being an algorithm. I guess it is possible to build entities that go beyond that but they would no longer be algorithms and every formal theory about them would be incomplete. Today's AI does not seem to be such a thing.

110

Andreas Keller @nannus.bsky.social · 23/09/2026

I think humans are not limited in this way, but creative in the sense that we can move out of the scope of any formal theory about our cognition, i.e. human cognition is not formalizable/describable in terms of a single algorithm. This is a hypothesis but I think there's a lot of empirical evidence.

010

Andreas Keller @nannus.bsky.social · 23/09/2026

This shows that AIs, as long as they are algorithms (which today's AIs are) are always incomplete in this sense. It is always possible to expand them, but they cannot do that themselves and the resulting expanded systems are incomplete in turn.

210

Andreas Keller @nannus.bsky.social · 23/09/2026

If we map possible training data as well as input and output data of models to natural numbers (i.e. Gödel numbers), learning algorithms appear as programs enumerating total functions on natural numbers. There is a simple diagonalization proof showing that such enumerations are always incomplete.

210

Andreas Keller @nannus.bsky.social · 23/09/2026

My argument about algorithms/formal theories having systematic blind spots is here: www.cckp.space/single-post/...

cckp.space

BP 7 (2024): Andreas Keller, Intelligence and Paraintelligence, 62-108

110

Andreas Keller @nannus.bsky.social · 23/09/2026

At least the creativity is in the whole, provided by the humans. The AI is an algorithm and it is produced by an algorithm. Each learning algorithm/AI model can be shown to have systematic blind spots. Humans otoh seem to be creative, i.e. can move out of the scope of any formal theory about them.

110

Andreas Keller @nannus.bsky.social · 23/09/2026

In früheren Jahren habe ich Gemüseselbsternte schon jahrelang in Köln gemacht, bei diesem Projekt: gartenglueck.info, auch sehr zu empfehlen.

gartenglueck.info

gartenglück – Selbsternte-Gärten

020

Andreas Keller @nannus.bsky.social · 23/09/2026

Die Hokkaidos waren Teil des Grundprogramms, die anderen beiden Sorten habe ich dazugesäht (Saatgut im Internet bestellbar). Bush Delicata und Sweet Dumpling schmecken ganz anders, eher wie Süßkartoffeln, ein wenig wie Maronen.

010

Andreas Keller @nannus.bsky.social · 23/09/2026

Kürbisernte (oder was davon übrig ist, drei von den Hokkaidos sind schon verspeist).Die anderen Sorten: Bush Delicata (länglich) und Sweet Dumpling.

A green glass bowl filled with pumpkins/squash: several Hokkaido pumpkins, as well as a few small white squash with green stripes of two varieties: Bush Delicata (elongated) and Sweet Dumpling.

162

Andreas Keller @nannus.bsky.social · 22/09/2026

I suggest the 10th boundary is the sustainable size of the economy and our economic growth has gone through that boundary as well already, driving the whole global system across the other boundaries as well. Its the economy, stupid. www.scientificamerican.com/article/9-bo...

scientificamerican.com

9 ‘boundaries’ keep Earth livable. We’ve breached 7 of them

Earth’s annual checkup shows that we’re at increasing risk of moving away from the environment we’re adapted to

010

Andreas Keller @nannus.bsky.social · 21/09/2026

Am Ende dieser Brücke gibt es bei jedem Zug das Kaiser-Wilhelm-Gedächtnis-Quietschen. Wilhelm II wollte, dass die Brücke genau auf den Dom zu führt. Kurz davor biegen die Züge dann ab zum Bahnhof und es quitscht in der Kurve. Eine seiner kleineren Sünden.

110

Andreas Keller @nannus.bsky.social · 20/09/2026

Wenn Merz Anstand hätte würde er jetzt die Verantwortung übernehmen und sowohl als Parteivorsitzender als auch als Kanzler zurücktreten.

000

Andreas Keller @nannus.bsky.social · 20/09/2026

Exactly which of the Schmidt Ocean expeditions was this. Which of the dives? They are all on Youtube, so it would be interesting to know the dive number and the exact date and time of the observation.

001

Andreas Keller @nannus.bsky.social · 20/09/2026

Is that maybe what is at the center of the right-wing "character": fear of women? At least part of it?

110

Andreas Keller @nannus.bsky.social · 20/09/2026

Strange these men seem to be afraid of strong women.

100

Andreas Keller @nannus.bsky.social · 19/09/2026

Is this real? Looks AIish to me.

000

Andreas Keller @nannus.bsky.social · 19/09/2026

Der nächste Merzismus*. Bei solchen Äußerungen von ihm frage ich mich jedesmal: ist das jetzt Dummheit oder ist das Vorsatz, also Bösartigkeit? Beides bedeutet, dass er als Kanzler völlig ungeeignet ist. Wer hatte bloß die Idee, diesen Mann zum Kanzler zu machen? *typisch Merz'sche Fehlleistung.

000

Andreas Keller @nannus.bsky.social · 19/09/2026

We're making a documentary about refugees in Nigeria who fled from war in Cameroon. Why did they flee, and what are their lives like now? You can help us by donating some money. linktr.ee/oneofmanyneg...

Plakat für das Dokumentarfilmprojekt "One of many neglected crises" über Flüchtlinge in Nigeria und den Krieg in Kamerun, vor dem sie geflohen sind.

Poster for the documentary film project “One of Many Neglected Crises” about refugees in Nigeria and the war in Cameroon from which they fled.

010