Newspeak House 2026: Theodore Keloglou Topics

These topics are some of the activities of the one-year programme: Introduction to Political Technology. If interested in these ideas read more and explore participating at: newspeak.house/study-with-us

Table of Contents

How can companies avoid becoming corrupted?

Eric Ries published Incorruptible: Why Good Companies Go Bad—and How Great Companies Stay Great in 2026. This book is about:

Specific points from the book:

Nostr Protocol

Nostr came out in 2020 as a decentralised protocol for social networking and publishing. In nostr, there are no servers. There are only clients and relays: clients post messages, relays transfer them to other clients.

Identity in nostr is based on public-key cryptography: your public key is your identity (like in Bitcoin) and your private key is your password. This means that identities are not tied to particular servers, domain names, or anything else. They are just two keys that sign messages.

This is interesting because it's radically simple, it's just identities and relays. How far can you go with such a simple protocol?

Iroh Protocol

Iroh is a peer-to-peer networking protocol, designed to make direct communication between devices simple.

What's the point of this existing? Peer-to-peer connectivity, with a complete lack of an always on server in the cloud, is not something feasible for today's applications. Their complexity and feature set (even basic features) require an always-on server that will be there to respond at any point.

However, servers are centralised infrastructure that usually do not control. We would love to avoid them. Iroh, using some neat low-level networking tricks, makes this very easy! Which means we can create high-level apps with fancy features without a central server! Magic!

Shelter Protocol and chelonia.js

Shelter Protocol is a protocol for building end-to-end encrypted, federated web applications.

Think Signal's end-to-end encrypted capabilities but instead of just having text messages to exchange, you can have any kind of data; which means you can build any kind of application that supports end-to-end encryption natively.

Group Income is a software that facilities basic Income for friends and family in a decentralised and end-end-encrypted way. It is built using Shelter and chelonia.js (library that implements the protocol in JavaScript).

Multiplayer Agent Harnesses

There is a new category of software, coming now, after agents harnesses the multiplayer harnesses!

(It's always like that with games, isn't it? First they release single-player and then people want multi-player! Pokemon on Gameboy, GTA, Elder Scrolls, etc.)

How does it work? Visualise this: you are part of team that works on a project (either software project or non-software). You want, while inside Claude Code, to mention a teammate to ask them something, let them know about something, or anything else. How do you do it? You can't! Agent harnesses like Claude Code and ChatGPT Codex do not have such a feature.

Actually, Anthropic does offer a version of this inside Slack (they call it Claude Tag).

But in any case, people are building their own open source self-hosted versions of such tools:

This is a rapidly progressing category of software! Things will probably be very different in 3 months.

Blockchain Foundations

I think this course is great: https://www.marabu.dev/

It's a deep dive into blockchain and crypocurrency fundamentals:

This course talks both about the theory behind blockchains and how to implement them in practice. At the end of course we end up with a working cryptocurrency and blockchain network.

Ethereum

Ethereum is a both a (crypto-)currency but also a fully decentralized computing platform. It's like a CPU with memory but fully spread across thousands of computers.

Ethereum has governance that is decentralised and open via the Ethereum Improvement Proposals scheme.

Ethereum has multiple independent institutions that establish it:

Other Ethereum places:

Decentralised Finance

Cryptocurrencies have disappeared from the news but at the same time they are becoming part of global finance.

Western economies are introducing regulatory frameworks (MiCA in the EU, GENIUS and Digital Asset Market Clarity Act in the US, FSMA in the UK) and this paves the way for:

Dark side:

Digital Deliberative Spaces, Infrastructure, & Methods

People collectively thinking, discussing, and deciding what they think about an issue is an important aspect of quasidemocratic polities. Ideally, deliberation enables peaceful synthesis of conflicting preferences and needs. Local assemblies are a simple yet dangerous tool to do this. They are simple because it's just people gathering; they are dangerous because unless there is appropriate structure to them, chaos is likely.

Items to research online:

  1. Polis by The Computational Democracy Project
  2. Community Notes by Twitter
  3. vTaiwan's open consultation platform

Books:

  1. Open Democracy, Hélène Landemore, 2020
  2. When the People Speak, James S. Fishkin, 2011
  3. Against Elections, David Van Reybrouck, 2016

Books from a different point of view:

  1. Against Democracy, Jason Brennan, 2016
  2. The Myth of the Rational Voter, Bryan Caplan, 2007

Finally, there are a number of organisations in the UK (and outside / global) that work in the field of citizen assemblies and participatory democracy:

  1. Involve
  2. Shared Future
  3. Sortition Foundation / 858.org.uk
  4. House of the People / timetoassemble.org

Open Weights Models & Local Inference

I’m fascinated with the AI advancements of the post-ChatGPT world. The world of software engineering is completely revolutionised because of these innovations. Open source has been transcended. I’m excited about these new tools and use them as core tools for my work albeit, indeed, the future is generally substantially unclear.

I've been exploring the Chinese open weights models (Deepseek, Kimi, Minimax, GLM) and local inference AI tools (llama.cpp, ollama, DwarfStar, LM Studio, etc). I find this ecosystem particularly interesting because it currently represents the only meaningful counterweight to the centralized AI future dominated by the three main corporate players (OpenAI, Anthropic, Google).

Starting New Intentional Learning Communities

Universities around the Western world are becoming less relevant. Anybody can learn anything online on their own. Yet learning communities remain crucial. They facilitate people meeting, inspiring, supporting each other. Residential communities are a great way to approach this. You are in one this year. Let's start 20 more like it!

Related topics to start from or research:

Large Scale Encrypted Communications

Encrypted communications is a fascinating topic in human history. With the advent of PGP in 1991 and more importantly Signal app (as Open Whisper Systems when it came out in 2013) we acquired an extremely powerful tool we never had before: provably secure and instant communications available at scale.

Western governments have made multiple attempts to ban these tools in the past, and continue to do so today. Whether that’s correct or ethical is an open question and something we can debate but the tools’ importance cannot be understated; such tools can enable unprecedented coordination and power for small groups of people across the whole world.

Knowledge and Science Infrastructure

There is a significant part of the scientific community who feels concerned about the future of science infrastructure. In the term science infrastructure I include:

This topic is about the communication part. However, the way to improve this is not abstractly promoting communication and collaboration. Scientific communication and infrastructure for communication in science are absolutely vital and core components of how science works: unless we communicate effectively what knowledge we have discovered, we are bound to re-discover it.

We might that maybe it's fine, some inefficiency always exists, doesn't it? I claim no: this is not about inefficiency. If we can’t utilise what exists, we’ll never be able to build on top of what others discover.

Isolated discoveries mean no knowledge accumulation which eventually leads to the breakdown of the global science machinery.

Many paths and approaches to fix this, open to discussion.