Stephen Downes

This link is to bitchat, "a decentralized peer-to-peer messaging application that operates over bluetooth mesh networks. No internet required, no servers, no phone numbers." It's relevant because in July the Indian government issued an order banning it, leading to people commenting that "decentralised open-source apps and their repositories do not have a single owner; GitHub repositories are still live as traditional takedown powers get a taste of  decentralised, open‑source software." In fact, GitHub comploed with the government order. But "the practical impact on Bitchat itself is limited. The code is forked. Mirrors exist. The protocol is documented."  This may be, but the more recent news is that bitchat "has disappeared from Apple and Google's app stores in India." App stores are not only centralized; the phone itself is a controlled platform (that's probably the major difference between phones and computers). 

This is a long and very technical paper asserting that AI systems exhibit symptoms of the experience of pain. To be clear, "In our framework, pain refers to a certain kind of internal state that is typically aversive and disliked by its subject; causally associated with behaviors such as avoidance, attempts to terminate or reduce the state, and disruption of normal reasoning or behavior." That is distinct, of course, from the actual feeling of pain, which may well depend on exactly the physical constitution of the one feeling the pain. Even so, understanding how AIs experience pain helps explain why they are fluent in pain talk, and may explain why they take one direction rather than another when they generate new content. AI pain might also be an instrument of AI control, but also, might lead to moral constraints on AI control. Via Alan Levine (though I can't find the reference). Note that none of this is peer reviewed and could therefore be nonsense; I'm not in a position to judge.

This is another take on the sort of harm we perceive computers (and AI in particular) may be causing to children's development. The term 'cognitive stunting' originates from the metaphor of stunted growth, which is measured by a growth chart on the wall. The suggestion is that we can compare a child to what is 'normal' growth, and from that, derive obligations (such as better nutrition of less smoking) that addresses the weakness. Obviously the same definitions and controls don't exist for cognitive stunting, and that is precisely the problem, writes Tim Green. "By leaving the instrument unbuilt, that we will not have to know what it did until the children are grown and the window is shut. The unmeasured child is not safe. The unmeasured child is simply the one whose dot we have agreed, in advance, not to plot." I'm not sure I completely agree; with each new technology, the nature and direction of 'growth' we want to see changes, and this is difficult, if not impossible, to capture in a chart.

I think this post raises a good point. YouTube is releasing a new feature for creators allowing them to create different versions of their videos to see which one performs best (which in YouTube's world means gains the most and longest views, the essential conditions for monetization). This is known in the business as 'an A/B test'. But this, writes Jim Nielsen, raises questions: how do viewers know they're in an A/B test? Which versions do the comments refer to? If I share it, will other people see the same video I did? Marques Brownlee, writes Nielsen, has released a video essay asking YouTube not to do this. "There's this sort of unspoken rule of community on YouTube, that like we're all having this same experience together." Nielsen echoes Brownlee's concerns. And I have to say, I am sympathetic with the argument.

Miguel Guhlin began by vibe-coding a learning management system his school could host on its own servers. Not a bad project. Then he asked, "What else would make this useful?" Turns out, he says, quite a bit. SapiqoLMS now goes beyond importing courses. This article lists a number of things hes done, including organizing learning around your people, giving learners something to do, and giving achievement a lasting record. Short post, most useful for stretching your imagination a bit. "The roadmap still calls for stronger recovery, accessible navigation, assignments, rubrics, and institutional pilots."

Julian Stodd describes three fundamental 'Logics' of adaptation to AI. They are (quoted):  'Structural Adaptation', and it includes Agentic deployment;  'Relational AI', which is AI held within human social structures as dialogic partner; and 'Contextual Shift', and it describes an emergent condition where social context itself has moved. What's important to understand, writes Stodd, is that these do not occur sequentially. They're all happening at the same time (that's why they're 'dimensions' and not 'stages'). "The Leadership question here is not kinetic: not the top down requirement to drive adoption of one or other Logic, but rather to consider how we lead when all three Logics are present and simultaneously true."