Stephen Downes

I first encountered the term 'salience' reading Robert Stalnaker's book Caounterfactuals. Something is 'salient' if it is 'more vivid, more relevant'. It stands out. To me, it marked the difference between just any similarity between things, and relevant similarities. Stalnaker, and also David K. Lewis, analyzed salience in terms of possible worlds. I analyzed it in terms of neural networks; something was 'salient' if it stimulated a neural network response. A pattern of perception. This article analyzes salience from a phenomenological, analytical and affective perspective. "Salience as an embodied familiarity," writes Maritina Dimopoulou, "what jumps to attention is an entanglement of the observer with their world." Salience comes with expertise. "Pruning irrelevant dimensions is one of the manifestations of expertise: the domain expert has internalised so many distinctions that deviations from the pattern are visible to them." 

This is very much the sort of question that has been central; in my thinking for the last few years. "Can a federated network scale without turning itself into a platform? And underneath that is can the openweb grow without becoming the thing it was built to replace?" My response is that anything that looks like centralization is a point off weakness, a potential point of failure. That includes initiatives like the Openweb Governance Body (OGB), "the proposal for a native governance layer for a federated network, using the same principles that make federation possible – autonomous communities, open protocols, distributed participation, federation between communities, local decision-making, shared standards, negotiation and mediation, accountability between participating groups." I know, OGB doesn't look very central, but to my view the seeds are there.

"There are three reasons AI demands that we go to college. We need to study its effects and see it as a viable human tool. We must confront our need to discern and critique it. We need to be able to use it in a data-now, rapid world of information sharing. And we need to realize it is a new professional industry that, at its foundation, is an extension of human thought and inquiry." I know this looks like four reasons, but only the latter three points (discern and critique, be able to use it, extension of human thought) are discussed at any length. Of these, probably the first is most important. And we need to be able to set AI reasoning against human experience in a way that makes sense. 

I had to look up what the author could possible mean by 'Easter Bush' - it could be a type of plant, but more likely, the reference is to the University of Edinburgh campus of that name. The reference comes up in this discussion of experiential learning, and specifically a 'walk-shop', or walking workshop, modeled after Berger's essay Why Look at Animals, but instead titled 'Why look at Easter Bush' (one wonders after reading Berger: are the students like animals in a zoo?). The point, though, is to actually see what is there, more in the spirit of Kimmerer's Learning the Grammar of Animacy. "Imagine the access we would have to different perspectives, the things we might see through other eyes, the wisdom that surrounds us." As the article says, "Walking in this way can quickly become a radical act because the ideas you get are sometimes far-out and unusual. It provides us a moment to be playful and sometimes some of the best ideas come that way." That's my thinking as well, and why I take care to experience the places I visit, and to not merely pass through them. Of course, being able to do that in a place like Easter Bush is a privilege.

The popular social media platform Reddit is killing RSS support. As they explain, "Because RSS is now a common surface for large-scale scraping and automated abuse, we'll stop supporting RSS feeds on November 13, 2026 while preserving moderation workflows that depend on it through supported alternatives." I should note that the website is also a common surface for large-scale scraping, so there are probably plans in the works to limit access to this as well. The change hits moderators especially hard because they often use RSS to view posts within their own workflow. As one commenter notes, "Dropping a text based interface and replacing it with a UI only control panel is a huge loss of usability. Anyone who works with technical stuff knows this." Via Grant Potter.

This is a very nice article framing the discussion of machine intelligence within the context of the debate between David Hume and Immanuel Kant. Reason, said Hume, is and ought to be a slave of the passions. This includes (I would add) for Hume a moral sense, which is not (but should have been) discussed in this article. Kant, by contrast, argues that reason is the source of our freedom, giving us the means to determine which of those passions (goals, wants, ideals) are rational and consistent with the world and society within it. The AI argument that intelligence is no more that optimization, argues author Sasha Mudd, aligns with Hume's view, but this leaves us passive clients of machine intelligence. "What if we become so accustomed to being predicted, nudged, ranked and served that we lose confidence in the older, harder activity of asking what ends deserve our allegiance in the first place?" I don't agree, but I appreciate the crystal clear and accurate summations of Hume and Kant's respective philosophies, and the argument Mudd has build on them. Via Fredrik Graver.