On our way
Remote
Dmytri
@dk@tldr.nettime.org
Venture Communist
759
Followers
971
Following
48
Posts
Joined November 30, 2022
Bluesky:
email:
hello@dmytri.to
Posts
So Berlin people, this is my last Tuesday until next summer, so if you could stop by, I would be very happy to see you -- Dmytri
On Tuesday, August 11, 1942, Hedy Lamarr and George Antheil patented frequency-hopping communication.
Today is Stammtisch
Today!
Repeated retrieval tool calls create many unnecessary model round trips. Yoink it all at once into one structured multipart bundle.
~> https://github.com/dmytri/yoink
npx skills add dmytri/yoink
npx @dk@tldr.nettime.org/yoink
The six most terrifying words in the AI era are "Let me try an alternative approach."
If your agent even thinks this, TINA takes its tools away.
TINA: There Is No Alternative.
On Tuesday, July 21, 1925, a Tennessee jury found substitute teacher John Scopes guilty of teaching human evolution in the famous "Monkey Trial."
Today is Stammtisch
(2 of 3 before I move to Canada for a year)
On Tuesday July 8, 1947, the Roswell Army Air Field (RAAF) in New Mexico issued an official press release announcing that the military had recovered the remains of a "flying disc" from a ranch near Roswell.
Today is Stammtisch
(1 of 3 before I move to Canada for a year)
On Tuesday, August 22, 1995, the United States Patent and Trademark Office granted US Patent 5,443,036 to inventors Kevin T. Amiss and Martin H. Abbott for a method of exercising a cat with a laser pointer.
Today is Stammtisch
Use this one cool trick to make every agent work better. Add this instruction:
**The problem is always you.** When something doesn't work, assume you are doing it wrong first. Do not blame the tools, APIs, docs, or environment until you have found and fixed what you messed up.
Over on LinkedIn: on Context, Custody and Traceability in agent assisted software development.
good friend and stammtisch alumuni Aly soliparty:
Working Title: Democracy 2.0
Logline: In a stable, post-democratic society, every citizen believes their preferred party is in power. A neural-link system personalizes everyone's news, political speeches, and policy explanations. The laws governing everyone remain identical.
Replying to
@dk@tldr.nettime.org
Easy emotional boosts in a world where human strokes are hard to come by warp the superego in a very strange way; the human loses grounding, tends toward delusion, and the parent voice becomes hallucination.
3/3
Replying to
@dk@tldr.nettime.org
A chatbot is not a nurturing or critical parent, but a probabilistic one, optimized for engagement rather than care.
As people become more isolated and more stroke-deprived, AI becomes a kind of automatic stroke dispenser, always available with no risk of rejection.
2/3
Psychologically, a Stochastic Parent is much worse than a Stochastic Parrot.
The parent ego-state is formed via positive and negative strokes from authority figures, developing the superego and entrenching ideology.
1/3
Off-the-rack fits like a sack; go bespoke and go broke. Made-to-measure brings the treasure.
SaaS is an important sustainability model for free software. The future of SaaS is T-shaped and made to measure.
Replying to
@koinonikos@mastodon.social
@koinonikos@mastodon.social Kaschk, it's a bar right beside the Rosa Luxemburg U-Bahn station
I made a thing to eliminate version bump commits for npm packages, and included simple content integrity as a bonus.
git tag v1.0.0
git push origin main --tags
npx @dk@tldr.nettime.org/hipp
Package is published as v1.0.0. No version-bump commits. No merge conflicts.
Every time I read AI proponents say, “A well-defined specification upfront is crucial for a successful outcome,” it feels like we're betting on improving the stage that the computer industry has always failed at: a well-defined specification upfront.
Replying to
@evan@cosocial.ca
@evan@cosocial.ca yeah, like who cares what their secret identity is? It's like this person you never heard of is Banksy? So? Unless it's someone who is already famous for something else besides being Banksy why would that be interesting?
Replying to
@guyjantic@infosec.exchange
@guyjantic interesting, I don't know much about psychology, from what little I do know, feels like jungian stuff like "shadow work" with it's fixed archtypes feels a lot more ptolemiac that CBT which from my limited understanding is "you are not your thoughts" type exercises. Would love to know more about your view here..
Replying to
@dk@tldr.nettime.org
@Npars01 @DaveMWilburn @pluralistic eg, this underappreciated joke from July
https://tldr.nettime.org/@dk/114851322710541553
Replying to
@odrotbohm@chaos.social
@odrotbohm for sure, ai just majes devs more productive in ptolemaic code producion
Replying to
@Npars01@mstdn.social
Replying to
@dk@tldr.nettime.org
The Ptolemaic model of the solar system, though completely incorrect, allowed sailors to navigate the globe for centuries, its failure was not practical but explanatory, similar to how Ptolemaic Code works until deeper issues arise.
6/7
Replying to
@dk@tldr.nettime.org
This is a trap: while the code works within its parameters, its internal model is incoherent. When it inevitably fails, the incoherent basis for debugging or correction will lead to additional 'epicycles' being added. This process increases system complexity and brittleness.
5/7
Replying to
@dk@tldr.nettime.org
It attempts to fit solutions from potentially very different domains together with corrective code "epicycles" until it satisfies its context, the user, and tests.
4/7
Replying to
@dk@tldr.nettime.org
this code passes all its tests and satisfies, its specifications, yet is built on a fundamentally flawed logic.
AI code generation, which relies on examples, will likely produce significant amounts of Ptolemaic Code.
3/7
Replying to
@dk@tldr.nettime.org
Like the ancient Ptolemaic model of the solar system, which tried to overcome its fundamentally incorrect understanding of the solar system by adding complex "epicycles" to force its Earth-centred model to match observed reality,
2/7
I hereby coin the term "Ptolemaic Code" to refer to software that appears functional but is based on a fundamentally incorrect model of the problem domain. As more code is generated by AI, the prevalence of such code is likely to increase.
1/7