Stephen Downes

This is pretty basic but I liked the idea of analyzing the phenomenon of 'becoming a group member' from the perspective of joining a trail running group. Most of us have had similar experiences with other groups. In true social sciences tradition, the author gives us a taxonomy of five stages of membership: learning formal rules, learning informal rules, learning the language, forming connections, and status symbols within the group. Image: The Nature of Group Membership.

We may have forgotten about the metaverse but the metaverse hasn't forgottn about us. "Android XR is Google's new operating system aimed at powering devices like headsets and glasses and making possible new experiences, a.k.a. apps, running on them. Android XR will integrate Gemini, Google's AI assistant, to enable understanding user intent, defining a plan, guiding through tasks, and more." My thinking: this isn't intended to be a product. The hope is probably that enough people will buy them that the data collected can be used to train robots equipped with Android XR to navigate through the world, collecting their own 'experiences', vastly increasing the data that Google AI can use in the future.

Nothing to do with online learning, but it's cool, so it's going into the newsletter. The surface of the Sun is known as the Alfvén surface. "Above this surface, the solar wind becomes supersonic, so no disturbances in its flow can affect the Sun below." The Parker Solar Probe aoared below the Alfvén at a speed of 690,000 kilometers per hour (in other words, from Montreal to Vancouver is a bit less than a second). We've never been thi fast or this close to the Sun before.

In philosophy there is something called the type-token distinction. A token is (say) the actual instance or use of a word, while the type is the abstract concept or idea it represents. This Meta paper offers a version of the distinction, saying that large language models are based on tokens, while human reasoning is based on concepts, ie., types. I think there are issues with this, but it doesn't matter, because what they're actually doing is something different again: "The human brain does not operate at the word level only. We usually have a top-down process to solve a complex task or compose a long document: we first plan at a higher level the overall structure, and then step-by-step, add details at lower levels of abstraction." Breaking things down into parts, though a very useful heuristic (known as the Cartesian method) has nothing to do with types and tokens. But that's OK, because that's not what they're doing either. Looking more closely at the paper (49 page PDF), it appears that they're setting up basic representations like frames or models and looking at a word's role within that. But that's not how humans think at all.

The story: "Anthropic recently released their Model Context Protocol (MCP), an open standard describing a protocol for integrating external resources and tools with LLM apps. The release includes SDKs implementing the protocol, as well as an open-source repository of reference implementations of MCP." It's important because this is the sort of way learning technology will integrate with AI engines.

When I studied history in school, it was told from the perspective of Europeans, as they gradually unveiled the mystery that was new world and the darkness that was Africa. The perspective of Indigenous people, who already knew these places existed, was not represented. Over time this perspective has shifted, but it needs to be more than just additive, write the authors of this paper (32 page PDF). They offer five recommendations: "(1) challenge hegemonic narratives, (2) value Indigenous ways of knowing and being, (3) reflect on privilege and positionality, (4) engage in the ethical dimension, and (5) focus on the future." I note that 'ways of knowing', as presented here, is probably better characterized as 'ways of experiencing'. Image: Donna Ward.