This is definitely a feedback loop in play, as already LLMs emulate bland SEO optimised text very well because most of the internet is already full of that crap. It's just a bunch of sites, and mostly other sources that serve as source of K though, is it not? So the feedback loop exposes to more people that they shouldn't see 'the internet' as the source of all truth? And is this feedbackloop not pointing to people simply stopping to take this stuff in (the writing part does not matter when there's no reader for it)? Unless curated, filtered etc by verifiable human actors? Are we about to see personal generative agents that can do lots of pattern hunting for me on my [[Social Distance als ordeningsprincipe 20190612143232]] en [[Social netwerk als filter 20060930194648]]

Multi-modal output can for now still suggest there's a human at work, not a generative agent. But multi-modal output can soon if not already also be generated. This still seems to focus on the output being the thing authenticated to identify human making. output that is connected to other generated output. There's still no link to things outside the output, into the authors life e.g. Can one fake the human process towards output, which is not a one-off thing (me writing this in a certain way), but a continuous and evolving thing (me writing this in a certain way as part of a certain information process, connected to certain of my work processes etc.). Seen from processes multi-modal output isn't a different media format, it is also work results, projects created, agency in the physical world. In those processes all output is an intermediate result. Because of those evolving processes my [[Blogs als avatar 20030731084659]]. Vgl [[Kunst-artefact is (tussen)uitkomst proces 20140505070232]] There was this article about an artist I can't find back that saw all his outputs over time as intermediate and expression of one narrative. This https://www.flickr.com/photos/tonz/52849988531/in/datetaken/ comes to mind to. Provenance and entanglement as indicators of authenticity.

This hints towards prompt-engineering, and the role of prompts in human interaction itself [[Prompting skill in conversation and AI chat 20230301120740]]

High Q use of generative AI will be about where in a creative / work process you employ to what purpose. Not in accepting the current face presented to us in e.g. chatGPT: give me an input and I'll give you an output. This in turn requires an understanding of one's own creative work processes, and where tools can help reduce friction (and where the friction is the cognitive actual work and must not be taken out)

If you use LLMs to improve your mediocre writing it will help. If you use it to outsource too much of your own cognitive work it will get you the bland SEO texts the LLMs were trained on and the result will be more mediocre. Greedy reductionism will get punished.

I wonder about reading after this entire section about writing. Why would I ever bother reading generated texts (apart from 'anonymous' texts like manuals? It does not negate the need to be able to identify a human author, on the contrary, but it would also make even the cheapest way of generating too costly if noone will ever read it or act upon it. Current troll farming has effect because we read it, and still assume it's human written and genuine. As soon as that assumption is fully eroded whatever gets generated will not have impact, because there's no reader left to be impacted. The current transitional assymmetry in judging output vs generating it is costly to humans, people will learn to avoid that cost. Another angle is humans pretending to be the actual author of generated texts.

Is this not merely entering an 'arms race' against our own tools? With the rat race effect of higher demands over time?

What about moving sideways not up? Bringing in the richness of the layering of our (internal) reality and lives? The entire fabric that makes up our lives, work, communities, societies, indicately more richly in our artefacts. Which is where my sense of beauty is [[Schoonheidsbegrip 20151023132920]] as [[Making sense is deeply emotional 20181217130024]]

On a web with many generative agents, all actors are going to be assumed models until it is clear they're really human.

Maggie Appleton calls this 'passing the reverse Turing test'. She suggests using different languages than English, insider jargon etc, may delay this effect by a few months at most (and she's right, I've had conversations with LLMs in several languages now, and there's no real difference anymore with English as there was last fall.)

The web is conversation (my blog def is), texts are a means to enter into a conversation, connection. For algogens the texts are the purpose (and human time spend evaluating its utility and finding it generated an externalised cost, assymmetric as an LLM can generate more than one can ever evaluate for authenticity). Behind a generated text there's no author to connect to. Not in terms of annotation (cause no author intention) and not in terms of actual connection to the human behind the text.

Amplifying dominant perspectives, a feedback loop that ignores all of humanity falling outside the original trainingset, which is impovering itself, while likely also extending the societal inequality that the data represents. Given how such early weaving errors determine the future (see fridges), I don't expect that to change even with more data in the future. The first discrepancy will not be overcome.

Appleton points to the inherent severely limited trainingset and hence perspective that is embedded in LLMs. Most of current human society, of history and future is excluded. This goes back to my take on data and blind faith in using it: [[Data geeft klein deel werkelijkheid slecht weer 20201219122618]] en [[Check data against reality 20201219145507]]

Algogens aren't fixed social entities/identities, but mirrors of the prompts

Conversation / social interaction / contactivity is the human condition.

Algogens are disconnected from reality. And, seems a key point, our own cognition and relation to reality is not just through language (and by extension not just through the language center in our brain): spatial awareness, embodiment, senses, time awareness are all not language. It is overly reductionist to treat intelligence or even humanity as language only.

generative AI as very smart and super incompetent at the same time, which is hard to reconcile. Is this a [[Monstertheorie 20030725114320]] style cultural category challenge? Or is the basic one replacing human cognition?

yes, all generated content is devoid of an author context e.g. It's flat and 2D in that sense, and usually fully self contained no references to actual experiences, experiments or things outside the scope of the immediate text. As I describe https://hypothes.is/a/kpthXCuQEe2TcGOizzoJrQ

Having the internet swarmed by generative agents is unlike current bots and scripts. It will be harder to see diff between humans and machines online. This may be problematic for those of us who treat the web as a space for human interaction.

AgentGPT https://agentgpt.reworkd.ai/nl is a version of such Generative Agents. It can be run locally or in your own cloud space. https://github.com/reworkd/AgentGPT

Generative agents have a database for long term memory, and can do internal prompting/outputs

Generative agents are a sort of indefinite prompt chaining: an NPC or interactive thing can be LLM controlled. https://www.youtube.com/watch?v=Gz6mAX41fs0 shows this for Skyrim. Appleton mentions a paper https://arxiv.org/abs/2304.03442 which does it for simlike stuff. See Zotero copy Vgl [[Stealing Worlds by Karl Schroeder]] where NPC were a mix of such agents and real people taking on an NPC role.

Prompt chaining is when you iterate through multiple steps from an input to a final result, where the output of intermediate steps is input for the next. This is what AutoGPT does too. Appleton's employer Ought is working in this area too. https://www.zylstra.org/blog/2023/05/playing-with-autogpt/

generative AI currently follows the pattern of 1 input and 1 output. There's no reason to expect it will stay that way. outputs can scale : if you can generate one text supporting your viewpoint, you can generate 1000 and spread them all as original content. Using those outputs will get more clever.

Mentioned copy writing algogens * Jasper * Wordtune * copy.ai * quillbot * sudowrite * copysmith * moonbeam

Appleton posits that the waves of generative AI output will expand the dark forest enormously in the sense of feeling all alone as a human online voice in an otherwise automated sea of content.

Appleton sketches a layering of dark forest web (silos mainly), cozy web (personal sites, newsletters, public but intentionally less reach), and private chat groups, where you are in pseudo closed or closed groups. This is not searchable so any knowledge gained / expressed there is inaccessible to the wider community. Another issue I think is that these closed groups only feel private, but are in fact not. Examples mentioned like Slack, Discord and Whatsapp are definitely not private. The landlord is wacthing over your shoulder and gathering data as much as the silos up in the dark forest.

Cozy Web is like Strickler/Liu's black zones above. Sounds friendlier.

Strickler compares this to black zones as described in [[Three Body Problem _ Dark Forest by Cixin Liu]], withdraw into something smaller which is safe but also excluding yourself permanently from the greater whole. Liu describes planets that lower the speed of light around them on purpose so they can't escape their own planet anymore. Which makes others leave them alone, because they can't approach them either.

Personal websites like mine since mid 90s fit this. #openvraag what incentives are there actually for people now to start their own site for online interaction, if you 'grew up' in the silos? My team is largely not on-line at all, they use services but don't interact outside their own circles.

The toxicity made me leave FB and reduce my LinkedIn and Twitter exposure. Strickler calls remaining nonetheless the bowling alley effect: you don't like bowling but you know you'll meet your group of regular friends there.

Hardly a 'theory', a metaphor re-applied to experiencing online interaction. (Strickler ipv Striker)

The internet feels lifeless: ads, trolling factories, SEO optimisation, crypto scams, all automated. No human voices. The internet unleashes predators: aggressie behaviour at scale if you do show yourself to be a human. This is the equivalent of Dark Forest.

Yancey Strickler https://onezero.medium.com/the-dark-forest-theory-of-the-internet-7dc3e68a7cb1 https://onezero.medium.com/beyond-the-dark-forest-a905e2dd8ae0 https://www.ystrickler.com/

A specific proposed solution to [[Fermi Paradox 20201123150738]] where is everybody? Dark forest, it's full of life but if you walk through it it seems empty. Universe seems empty of intelligent life to us as well. Because life forms know that if you let yourself be heard/seen you'll be attacked by predators. Leading theme in [[Three Body Problem _ Dark Forest by Cixin Liu]]

True for me (and E) too. For me it largely was because the internet became a thing right around when I entered uni in the late 80s, and it always was about connecting. Blogging esp early in the years 2002-2009 led to a large part of my personal and professional peers network.

'18th c. men of letters' I've sometimes thought about it like that actually, and treat meet-ups etc like the Salons of old vgl. [[Salons organiseren 20201216205547]]

https://web.archive.org/web/20230503150426/https://maggieappleton.com/forest-talk

Maggie Appleton on the impact of generative AI on internet, with a focus on it being a place for humans and human connection. Take out some of the concepts as shorthand, some of the examples mentioned are new to me --> add to lists, sketch out argumentation line and arguments. The talk represents an updated version of earlier essay https://maggieappleton.com/ai-dark-forest which I probably want to go through next for additional details.

Ought makes Elicit (a tool I should use more often). Maggie Appleton works here. A non-profit research lab into machine learning systems to delegate open-ended thinking to.

https://web.archive.org/web/20230503191702/https://www.rechtenraat.nl/artikel-10-evrm-en-woo/ Caroline Raat start zaak tegen RvS om toepassing uitzonderingsgronden WOO.

Argumentatie: - 2017 EHRM arrest stelt dat public watchdogs (journo's, bloggers, ngo's, wetenschap) direct beroep kunnen doen op art 10 EVRM voor toegang tot documenten van de overheid. - Als het zo'n watchdog is die toegang vraagt, gaat EHRM boven WOO/WOB. - Watchdog hoeft alleen aan te tonen dat info nodig voor publieke info-voorlichting - Watchdog hoeft geen bijzondere omstandigheden aan te tonen - Weigering kan alleen als daar dringen maatschappelijk reden toe is - Weigering kan alleen op basis genoemde uitzonderingen in art10lid2 EVRM zelf, en weigering moet als noodzakelijk voor de samenleving gemotiveerd worden - Andere weigeringsgronden in WOB/WOO zijn niet van toepassing.

Dit zou bijv heel ander verloop van Shell papers kunnen geven.

ICs as hardware versions of AI. Interesting this is happening. Who are the players, what is on those chips? In a sense this is also full circle for neuronal networks, back in the late 80s / early 90s at uni neuronal networks were made in hardware, before software simulations took over as they scaled much better both in number of nodes and in number of layers between inputs and output. #openvraag Any open source hardware on the horizon for AI? #openvraag a step towards an 'AI in the wall' Vgl [[AI voor MakerHouseholds 20190715141142]] [[Everymans Allemans AI 20190807141523]]

Very [[Stealing Worlds by Karl Schroeder]] chatgpt wired into NPCs in Skyrim.

https://web.archive.org/web/20230503153010/https://subconscious.substack.com/p/llms-break-the-internet-signing-everything

Gordon Brander on how Maggie Appleton's point in her talk may be addressed: by humans signing their output (it doesn't preclude humans signing generated output I suppose, which amounts to the same result as not signing) Appleton suggests IRL meet-ups are key here for the signing. Reminds me of the 'parties' where we'd sign / vouch for each others SSL certs. Or how already in Threema IRL meet-ups are used to verify Threema profiles as mutually trusted. Noosphere is more than this though? It would replace the current web with its own layer. (and issues). Like Maggie Appleton mentions Dead Internet Theory

https://web.archive.org/web/20230503151906/https://maggieappleton.com/ai-dark-forest The essay from which the talk I also saved came. The talk is a good approach to the storyline, expect more details / arguments to be found in here.

https://web.archive.org/web/20230503103156/https://www.bluelabellabs.com/blog/deceptive-dark-patterns-stats/

Overview of darkpatterns in how brands treat people online. Vgl [[Digital Ethics Malpractices 20190529071619]] en [[Software transparantie over data opslag 20200109075817]]

https://web.archive.org/web/20230430194301/https://netzpolitik.org/2023/longtermism-an-odd-and-peculiar-ideology/ The EA/LT reasoning explained in this interview, in a way that allows easy outlining. Bit sad to see Jaan Tallinns existential risk path taking this shape, CSER seemed to be more balanced back in 2012/13 when I briefly met him in the context of TEDxTallinn, with climate change a key existential risk, not a speed bump on the road to advanced AI to provide future humanity.

https://web.archive.org/web/20230502181156/https://notebooks.vires.services/notebooks/01a__intro-jupyter-python

Seems a useful description of what Jupyter books are.

https://web.archive.org/web/20230502113317/https://wattenberger.com/thoughts/boo-chatbots

This seem like a number of useful observations wrt interacting with LLM based tools, and how to prompt them. E.g. I've seen mention of prompt marketplaces where you can buy better prompts for your queries last week. Which reinforces some of the points here. Vgl [[Prompting skill in conversation and AI chat 20230301120740]] and [[Prompting valkuil instrumentaliseren conversatiepartner 20230301120937]]

https://web.archive.org/web/20230426184932/https://www.mattprd.com/p/the-complete-beginners-guide-to-autonomous-agents via Stephen Downes https://www.downes.ca/post/75148

off grid / prepping is a LARP, very striking observation.

In narrative inquiry I ask people to title the experience they shared after sharing. Similarly I write my own titles usually after the content of a blogpost or a notion. Although when it comes to the internal branching highlighted above I usually start with a temporary title, which captures the jumping off point from the originating note.

I have this regularly, whenever I spend a bit of time on usually two or three related notes. Usually it annoys me because it sometimes feels like the branching goes faster than I can keep up with noting. That's from a 'production' perspective. Here I was aiming to finish a note, reducing the unfinished corpus by one, only to add a bunch of new beginnings to the heap to go through. The internal branching is a more positive phrasing for an effect I regularly treat as 'more work'. Good switch of perspective, as I have a mental image of external explosion that I can't contain, whereas internal branching is like fractals within the same general boundary. Good image.

https://web.archive.org/web/20230421183849/https://zettelkasten.de/posts/why-single-note-matters/

https://web.archive.org/web/20230420053314/https://howtosavetheworld.ca/2023/04/14/how-ai-conditioning-and-human-conditioning-differ/

not just trained with, also trained by. is it fully transparent though? Perhaps from the trainers/tools standpoint, but users are likely to fall for the tool abstracting its origins away, ELIZA style, and project agency and thus morality on it.

https://web.archive.org/web/20230411095546/https://www.reuters.com/technology/germany-principle-could-block-chat-gpt-if-needed-data-protection-chief-2023-04-03/

On the temporary ban of ChatGPT in Italy on the basis of GDPR concerns.

Italian DPA temporarily bans ChatGPT until adequate answers are received from OpenAI. Issues to address: 1. Absence of age check (older than 13) of ChatGPT users 2. Missing justification for the presence of personal data in trainingsdata of ChatGPT. 3. OpenAI has no EU based offices and as such there's no immediate counterparts for DPAs to interact with them. The temp ban is to ensure a conversation with OpenAI will be started.

The trigger was a 9 hour cybersecurity breach where user's financial information and content of their prompts/generated texts leaked over into other accounts.

https://web.archive.org/web/20230406100224/https://leanpub.com/langchain/read#using-llamaindex-to-search-local-documents-using-gpt-3

This looks like a useful guide to try and run my own 'chatpkm' on top of BLOOM through Huggingface. Vgl [[Onderzoek selfhosting AI tools 20230128101556]]

AI tools yes, not ChatGPT though. More contextualising and specialisation needed. And I'd add the notion that AI might be necessary as temporary fix, on our way to statistics. Its power is in weighing (literally) many more different factors then we could statistically figure out, also because of interdependencies between factors. Once that's done there may well be a path to less blackbox tooling like ML/DL towards logistic regression: https://pubmed.ncbi.nlm.nih.gov/33208887/ [[Machine learning niet beter dan Regressie 20201209145001]]

More ELIZA. The opposite of searching on the internet for your symptoms and ending up with selfdiagnosing yourself with 'everything' as all outliers are there too (availability bias), doing so through prompting generative AI will result in never suggesting outliers because it will stick to dominant scripted situations (see the vignettes quote earlier) and it won't deviate from your prompts.

Things missing in a prompt will not result from a prompt. This may reinforce one's own blind spots / omissions, lowering the probability of an intuitive leap to other possibilities. The machine helps you search under the light you switched on with your prompt. Regardless of whether you're searching in the right place.

This is where complexity comes in, teasing out narratives and recombine them into probes, probing actions that may changes the weights of narratives and mental models held about a situation. Not diagnostics, but building the path towards diagnostics. Vgl [[Probe proberend handelen 20201111162752]] [[Vertelpunt 20201111170556]]

I am coming to see this as a pitfall of generative AI texts. It seduces us to anthromorphise the machine, to read intent and comprehension in the generated text. Removing the noise in generating text, meaning the machine would give the same rote answers to the same prompts would reduce this human projection. It would make the texts much 'flatter' and blander than they currently already are. Our fascination with these machines is that they sometimes sound like us, and it makes us easily overlook the actual value of the content produced. In human conversation we would give these responses a pass as they are plausible, but we'd also not treat conversation as likely fully true.

Machines will do well in scripted situations (in itself a form of automation / codification). This was a factor in Hzap 08 / 09 in Rotterdam, where in programming courses the problems were simplified and highly scripted to enable the teacher to be able to grade the results, but at the cost of removing students from actual real life programming challenges they might encounter. It's a form of greedy reductionism of complexity. Whereas the proof of the pudding is performing well within complexity.

https://web.archive.org/web/20230406083125/https://inflecthealth.medium.com/im-an-er-doctor-here-s-what-i-found-when-i-asked-chatgpt-to-diagnose-my-patients-7829c375a9da

A comparison of ChatGPT responses to actual ER case descriptions. Interesting experiment by the author, though there shouldn't be an expectation for better results than it gave.

Ha. Annotation Kalir/Garcia positions annotation as genre, and as (distributed) conversation. [[Annotatie als genre of als middel 20220515112227]], [[Annotation by Remi Kalir Antero Garcia]] and [[Gedistribueerde conversatie 20180418144327]]

The human condition in its entirety is an infinite conversation I suspect.

https://web.archive.org/web/20230404092627/https://newsletter.mollywhite.net/p/feedly-launches-strikebreaking-as

This entire 'feedly goes into strikebreaking' headline at first didn't make any sense to me when E first mentioned it. First because it sounds extremely out there, in terms of 'service', and second, it's RSS, which I think is hardly suited for the type of claims this service makes (and the article shows that too imo). RSS content hardly shows emergent patterns if you've not defined the network/group you're drawing from imo (e.g. media are not useful for it), and it works at a slower pace than 'let's see if this protest turns violent'. I've worked for orgs that had a 'keep our employees save' coordination centre, and they defintely didn't tap into RSS. They'd send me an sms to avoid a certain part of certain city because of a disease outbreak for instance, or warn met of specific types of crime to watch out for when embarking on a mission, or real time weather warnings for my location.

I haven't used Feedly, I only mentioned it once on my blog in 2019, because my hoster blocked it as 'bad bot'. Foresight? https://www.zylstra.org/blog/2019/06/feedly-blocked-as-bad-bot-by-my-hoster/ I think that blocking feedly might be not as bad as I thought in 2019

Being crap at doing harm doesn't allow you to claim innocence of doing harm.

https://web.archive.org/web/20230404050349/https://greshake.github.io/

This site goes with this paper <br /> https://doi.org/10.48550/arXiv.2302.12173

The screenshot shows a curious error which makes me a little bit suspicious: the reverse Axelendaer is not rednelexa, there's an a missing.

The mechanism needs a consent step from the user: to allow Bing Chat to see currently open websites. And one of those websites already open, needs to contain the promptinjection.

Wait, what does this mean. #openvraag That previously it did, but now doesn't in response to this? Or that Bing Chat never did so in the first place? In the latter this paper is dealing in hypotheticals at this stage?

Somewhat suspicious of timing, but listen to those soundfiles. We're surrounded by Triffids!

Timing: the info about these sounds is known since 2012 https://gizmodo.com/plants-communicate-with-each-other-by-using-clicking-so-5919973 but this new paper turns to learning models to derive info from the sounds made.

Running BLOOM on a VPS or locally even is either expensive or very slow. Mostly because of the sheer size of the 176B model. Share it in a group of similar users on a VPS set-up? Use the Hugging Face API for BLOOM?

Donald points rightly to some of the classic monsterisation responses to AI. Although he imo misrepresents the EU AI Act (which in its proposal carefully avoids static tech regulation).

Vgl [[Monstertheorie 20030725114320]]

https://web.archive.org/web/20230329045841/https://digital-strategy.ec.europa.eu/en/library/cyber-resilience-act

proposal Cyber Resilience Act (read together with CSA, April 2019)

https://web.archive.org/web/20230329045405/https://berthub.eu/articles/posts/eu-cra-practicalities/

Bert Hubert on the CRA part 2

https://web.archive.org/web/20230329045249/https://berthub.eu/articles/posts/eu-cra-secure-coding-solution/

Bert Hubert on the CRA proposal, part 1.

Generative AI abstracts away the open web that is the substrate it was trained on. Abstracting away the open web means there may be much less incentive to share on the open web, if the LLMs etc never point back to it. Vgl the way FB et al increasingly treated open web URLs as problematic.

replacing the incentives to share on the open web are not a mere by-effect of it being abstracted away by generative AI, but an aimed for effect. As it may push people to seek the gains of sharing elsewhere, i.e. enclosed web3 services.

https://web.archive.org/web/20230328130227/https://librarian.aedileworks.com/2023/03/27/the-push-to-ai-is-meant-to-devalue-the-open-web-so-we-will-move-to-web3-for-compensation/

blogged https://www.zylstra.org/blog/2023/03/22930/

Via Stephen Downes https://www.downes.ca/post/75035

openvraag is er een link met [[Monstertheorie 20030725114320]] ook uit #2003/

PDF in Zotero

psychologische veiligheid als belangrijkste voorwaarde voor team succes. Verwijst naar Google onderzoek hiernaar, https://rework.withgoogle.com/print/guides/5721312655835136/

on both short term and long term risks in AI

https://web.archive.org/web/20230316192144/https://garymarcus.substack.com/p/should-we-worry-more-about-short?utm_source=twitter&sd=pf

https://web.archive.org/web/20230316103739/https://subconscious.substack.com/p/everyone-will-have-their-own-ai

Vgl [[Onderzoek selfhosting AI tools 20230128101556]] en [[Persoonlijke algoritmes als agents 20180417200200]] en [[Everymans Allemans AI 20190807141523]] en [[AI personal assistants 20201011124147]]

MDTO (Metagegevens voor duurzaam toegankelijke overheidsinformatie) lijkt ook de norm te worden voor WOO actieve openbaarmaking. #openvraag hoe zit dat voor passieve openbaarmaking? Wordt metadatering beperkt tot de actieve categorieën?

https://web.archive.org/web/20230309111559/https://www.d4d.net/news/ai-and-the-state-of-open-data/

Tim Davies looks at the bridge between #opendata and #AI. Should go throug the chapter in version 1 of the State of Open Data too. Note: while Tim acknowledges some of the EU data strategy developments (e.g. the dataspaces) it doesn't mention others (e.g. data altruistic non-profit entities) which may fit the call for instutions better. Data space aren't an institution, but a common market

https://web.archive.org/web/20230309061020/https://boffosocko.com/2023/03/07/rules-against-quotes-in-zettelkasten-a-closer-look-at-ahrens-on-quotes-and-collecting/

Chris Aldrich on Ahrens and on quoting and processing by one's own rephrasing as mutually reinforcing. As Ahrens is regularly interpreted as advicing no quoting. Vgl [[Zettelkasten Ahrens 20201025194845]]

https://web.archive.org/web/20230301112750/http://donaldclarkplanb.blogspot.com/2023/02/openai-releases-massive-wave-of.html

Donald points to the race that OpenAI has spurred. Calls the use of ChatGPT to generate school work and plagiarism a distraction. LLMs are seeing a widening in where they're used, and the race is on. Doesn't address whether the race is based on any solid starting points however. To me getting into the race seems more important to some than actually having a sense what you're racing and racing for.

Interesting point by John Caswell. AI prompting is a skill to learn, can we simultaneously learn to prompt better in conversations with other people? Prompting is a key thing in collecting narrated experiences for instance. Or will more conscious prompting lead to instrumentalising your conversation partner? After all AI chat prompting is goal oriented manipulation, what to put int to get the desired output? In collecting narrated experiences the narrator's reality remains a focal point, and only patterns over collections of narrated experiences are abstracted away from the original conversations. n:: [[Prompting skill in conversation and AI chat 20230301120740]] n:: [[Prompting pitfall instrumentalising conversation partner 20230301120937]]

https://web.archive.org/web/20230301110422/https://www.grouppartners.online/blog/critical-conversations

Julian dives into https://medium.com/@ElizAyer/meetings-are-the-work-9e429dde6aa3 on when meetings are productive work (vgl [[Netwerkleren Connectivism 20100421081941]]

https://web.archive.org/web/20230226002724/https://medium.com/@ElizAyer/meetings-are-the-work-9e429dde6aa3 Meetings are regular work, so blindly avoiding meetings is damaging.

Julian Elve follows up https://www.synesthesia.co.uk/2023/02/27/finding-the-real-work-that-meetings-are-good-at/ with lifting out the parts where Ayer discusses the type of meeting that are 'real work' and what they're for. (learning, vgl [[Netwerkleren Connectivism 20100421081941]]

Folgezettel are per the linked https://web.archive.org/web/20220125173712/https://omxi.se/2015-06-21-living-with-a-zettelkasten.html posting also a way to create some sort of initial overview in a physical system. In digital systems network maps serve a similar purpose as initial overview to be able to start with something. The outline Lawson mentions as origin is a thing in itself to me, esp as the connections / place in a system of a note can be reconsidered over time. Physical placement is by def a compromise, the question if it is a constraint that has a creative effect?

presumption: L's Folgezettel are a combination of outlining (as common in US maybe not German law edu) and the note cards used in sociology. Cross linking as a way to escape forced categorisation into exclusive buckets. Is there also in cross linking an element perhaps of escaping established idiom while building new (fields) of knowledge? (Vgl. Richard Rorty's struggle when forced to explain pragmatism in the language of Platonic dilemma's. [[Taal als zicht beperkend element 20031104104523]] )

Yes. Any pkm system or method is (or should be) tailored to one's own needs. Vgl [[% Interessevelden 20200523102304]] as [[Macroscope 20090702120700]]

Lawson thinks L's Folgezettel are a product of his training in law, and they were used by L for synthesis.

https://web.archive.org/web/20230227100951/https://www.seanlawson.net/2021/03/the-role-of-folgezettel-in-the-zettelkasten-method/

Sean Lawson on the rationale behind Folgezettel as used by Luhmann.

https://web.archive.org/web/20230227055845/https://www.politico.com/news/magazine/2023/02/24/russia-ukraine-war-oral-history-00083757

Fascinating read. Esp the build up in the year before the Russian full invasion of Ukraine started.

https://web.archive.org/web/20230226072321/https://blog.xot.nl/2023/02/14/analysing-the-architecture-of-the-european-digital-identity-framework/index.html

ha! psychohistoric event. Yes, I recognise some of that. I've been in recent sessions that were the 3rd of 4th larger public gathering with the same group, with the group leader still repeating the mantra that we hadn't been able to meet up for so long, where we just did the same thing several times before. A ritualised phrasing to excuse any time spent on catching up. I'd rather put the catching up on the actual program. No excuse needed, psychohistoric or not.

Indeed. 'the lockdown' in various conversations I've been in seems to be an indetermined period between 2019 and now, which serves as explanation for anything that wasn't finished in the past 3 years. As if we were all in actual stasis all that time continuously. Yes it was hard for us at times, I know it was much harder for other people I know at times in other locations, let alone what's been going on in China. But it wasn't constant and everywhere in NL or in EU. The Dutch actual lockdowns were 3 different periods and to very different degrees, with the first being the strictest, but the last one feeling the most difficult to me. I should mark the actual lockdowns and restrictions more clearly in my notes as factcheck.

Reading this I realise I'm not using Folgezettel really, only linking back to a previous notion. There's some sequencing, esp when I create little 'trains' (a notion, a link to a more abstract notion, a link to a more detailed one, a link to an example). The forward linking I generally not do, except sometimes. L always did forward linking in the sense of placing the index card.

https://web.archive.org/web//https://www.binnenlandsbestuur.nl/bestuur-en-organisatie/io-research/morele-vragen-rijksambtenaren-vaak-onvoldoende-opgevolgd

#nieuw reden:: Verschillende soorten morele dilemma's volgens I&O bij (rijks)ambtenaren. Let wel, alleen rondom beleidsproces, niet om digitale / data dingen. #openvraag is er een kwalitatief verschil tussen die 2 soorten vragen, of zijn het verschillende verschijningsvormen van dezelfde vragen? Verhouden zich de vragen tot de 7 normen van openbaar bestuur? Waarom is dat niet de indeling dan? #2022/06/30

Singer, b1946, Australian @ Princeton, applied ethics, utilitarian. EA / LT as utilitarianism ad absurdum

Lofty goals can serve as 'moral shield', excusing immoral behaviour in other situations, because the higher ends 'prove' the ultimate morality of the actor.

Colin Walker added a start here page to his site to present posts in a more curated way than stream to new / incidental visitors.

A comment about the diff between desktop and mobile browsing experience: one might miss the multiple columns. For my site that is true too: all the right column stuff comes at the bottom on mobile. Noone scrolls that far.

individually yes. Feedback loops is the response. It's just that we allowd socmed to base feedback almost entirely on outrage.

I've always assumed the point of the stream is that you can't drink it all. My [[Infostrat Filtering 20050928171301]] is based on the stream being overwhelming. Never twice into the same river etc. You don't make sense of the stream or catch what's important. Social filtering is the bit you 'drink' from the stream, and what you reshare is feedback into it. Given enough feedback what is important will always resurface.

Metaphor of algo-timeline as a canalised river, vs the free flowing stream that is e.g. a socmed stream like Mastodon.

True, not sure I did stop, but a rethink is definitely useful, and extending it.

blogroll is that too, no?

I don't follow, RSS is the stream I'd say, it's entire design is the reverse chronological order? Or does Amit specifically mean the representation of the stream on the blog front page?

As is the blogroll.

https://web.archive.org/web/20221214055312/https://wildrye.com/roundup-of-67-tools-for-thought-to-build-your-second-brain/

Glad to notice that: - I've heard of / know many of these tools, so have an ok overview of the current space. No surprises in the list. - I have not cycled through all these tools.

Also interesting that The Brain still exists. Used to be my desktop interface in the late 90s/early 00s.

This sounds like what I call soft-focusing. Some years ago I let go of being strict with myself, and stopped having defined goals in favor of course/directions and a vaguer sense of the destination. I also started soft-focusing my inputs (if there's a connection to my running list of interests connected to my sense of direction it qualifies), and am now trying to soft-focus my outputs. Not blogpost / project A or deliverable B as I would earlier, but more emergent. Then when I have task / creative thing to do, I use it to formulate questions to my notes and see what comes up. This evolved from doing the same in conversations with clients and colleagues where the value of that and resulting associations was clearly visible.

are there compelling stories about what comes after?

oversimplification abounds: this isn't the single cure, so it's not helpful. While we're in a truly complex environment and per def this means a whole collectiojn of simultaneous interventions is needed. Complexity never has a single answer.

stories about not having the solutions / being too late.

stories lack context discouraging proper comparison (imo often by design)

green washing is a category of stories

social media the last refuge of climate change denial

we are all storytellers or should be, and we're in a power negotiation situation.

It might be we have all the pieces, just not the connective and compelling narrative.

who's the we here? Just the few (rich countries) or also the many. Then lists aspects systems / ethics and what method of change you think to deploy.

The Ponzi scheme that is western society is centered as cause and its end the remedy. Vgl [[De externe input cheat 20091015070231]] where I formulate the same. We've run out of being able to ignore [[Geexternaliseerde effecten 20200914204533]]. Is this a [[Ethics of Agency 20201003161155]]?

This sounds like crap detection and [[Infostrat Filtering 20050928171301]] What do you amplify, how do you judge sources, how do you shape your info-diet, and are you aware that how/what you share is a feedback loop to those who shared the stuff you're reacting to? Solnit focuses here on the narrative/shape of what you share in response to your intake (in contrast to resharing other people's narratives)

progressive populism? Vgl [[countering the populist narrative 20221103141532]] and https://jarche.com/2022/09/better-stories-for-a-better-world/

This is how I've read SF for years, both near future and space opera. As mood board and thinking input.

adrienne maree brown https://en.wikipedia.org/wiki/Adrienne_Maree_Brown in turn inspired by SF author Octavia Butler (have I read her xenogenesis trilogy?)

https://web.archive.org/web/20230112060303/https://www.theguardian.com/news/2023/jan/12/rebecca-solnit-climate-crisis-popular-imagination-why-we-need-new-stories Vgl [[We Need New Stories by Nesrine Malik]] wrt argumentation and types of narratives.

All this, and in the end a statement that in absolute terms there's stability? Wow, did they miss the demographic factrs at play in scientific community as possible explanation of the relative effect?

At the same time this is precisely the argument above wrt 19th century. Right when you think you know all, everything gets turned over.

Kind of like an inverse singularity: a brick wall.

Pointing to [[Evolutionair vlak van mogelijkheden 20200826185412]] again and that new disruptions probably have higher thresholds to cross (resources, crossdiscipl teams)

Book to find.

Again this reads (second hand) more as a quantity-dynamic. Many confirming the status quo btw is also K. (Vgl Edison's '999 ways I established that don't work'.)

Rate of disruption, need to check how this rate is determined. Absolute number of big breaks over time, or relative to scientific production in general (which is when it would be expected to slow with rising production).

Suggesting it isn't in scientific practice. So what changed very much? Volume of scientists -> volume of publications.

the narrowing maybe a relative one? Or is it really being on a branching but ending path in [[Evolutionair vlak van mogelijkheden 20200826185412]]

Is this caused by anything in science, or a symptom of the growing scientific community globally, and rising average edu levels? Quantitative phase shifts have qualitative effects.

switching here from publications to patents which are a very different beast. Patents are transactions (publishing ideas for temporary market exclusivity) There isn't a necessary path from paper to patent, and def not in all scientific fields. By def patents are more engineering (in the primary meaning) oriented imo, they're about how to (potentially) make things (work). Why's aren't patentable.

I can see how the volume of publications rising is also result of broader access to scientific disciplines. By def high impact is rare, so the average impact will decline with volume. If only because by def a lower percentage of eyes will ever see a paper.

https://web.archive.org/web/20230116221448/https://www.dailykos.com/stories/2023/1/16/2147067/-Are-we-living-in-the-last-days-of-the-Scientific-Age Discusses a recent Nature article looking at how increasing numbers of new patents (a rightly criticized indicator) deal with ideas of decreasing impact. Conclusion is though that the number of disruptive patents remains high, just that the overall number of patents rises. Meaning perhaps more the democratisation of patenting, or perhaps the end of the utility of patenting, than stalling scientific progress.

Some points from a 1996 book mentioned vgl [[Evolutionair vlak van mogelijkheden 20200826185412]] wrt scientific progress / increasing niche-specification in the evolutionary plane of possibilities. The book suggests skating to a different place has prohibitive costs and maybe out of reach. Vgl local optimisation in complexity, and what breaking loose from a local optimum takes. Is the loss of the Scientific Age here discussed a needed path into chaos to be able to reach other peaks? Check comments on the Nature article to see if this type of aspect gets discussed.

Call to action: curb the hype.

Does that ever work? (Vgl when I was recently seen as old and negative simply because of listing a range of (pos and neg) real experiences wrt metaverse from earlier waves of hype for VR, and asking questions that are a lithmus test to determine contextual value to a user.

One can not participate in hype, and ensuring the hypers are never able to be seen at the same table / level of disucssion as you (it legitimises the entity of lesser status if a key figure debates a more trivial figure) But can one pour cold water on it when others and other outlets do join in? And without being seen as 'just negative' Hype shifts perception of the neg-pos spectrum [[Overton window 20201024155353]] and Trevino scale, which also applies I think in ethics / phil of tech discussions (which in part means circling back to monstertheory).

speed and mainstreaming are points of contention here, in light of unripe tech, and unprincipled company behind it.

Vgl w 'disinfo innoculation' a la Finland.

When hype is used to influence public opinion, there's a obligation to resist. (Vgl [[Crap detection is civic duty 2018010073052]], en [[Progress is civic duty of reflection 20190912114244]]) Also with which realm of [[Monstertheorie 20030725114320]] are we dealing here with this type of response? In the comments on Masto it's partly positioned as monster slaying, but that certainly isn't it. It's warning against monster embracing. I think the responses fall more into monster adaptation than assimiliation, as it aims to retain existing cultural categories although recognising the challenges issued against it. Not even sure the actual LLM is the monster perceived, but its origins and the intentions and values of the company behind it. Placing it outside the Monster realm entirely.

Additionally the ecological footprint of the tech is problematic (Vgl the blockchain mining activities footprint discussion).

Claims: Doing free labour for OpenAI is problematic. (not expressed: every usage feeds back into the machine and is more free labout put in) Reasons: * OpenAI founders are on the utilitarianism followed ad absurdum end of Effective Altruism. The 'Open' bit is open-washing. * The provenance of training data is ethically suspect (internet scraping), and not controlled for quality. * Externalities aren't taken into account * Social impact of use (including based on faulty output) not considered. * LLMs are still bad at understanding (and routinely fail test that contain imprecise references to other words in the sentence, which humans find easy to solve based on real world knowledge outside the text)

buying into the hype equated to doing PR work for OpenAI

Iris van Rooij means to counteract edu sector buying into the AI hype

https://web.archive.org/web/20230114235338/https://irisvanrooijcogsci.com/2023/01/14/stop-feeding-the-hype-and-start-resisting/

https://web.archive.org/web/20230116150040/https://cacm.acm.org/blogs/blog-cacm/267674-ais-jurassic-park-moment/fulltext

Dave positions free will as a 1 or 0 thing, and then tends to 0. Is it that binary though? A spectrum (matters of degree across different contexts) would also help explain things.

Free will is not free of consequences, which is where conditioning kicks in. (and society demanding responsibility or not)

How about animals? Conditioned by their ecosystem (like us, by ecosystem and culture), yet free to roam which coalesces into patterns (migration, foraging, trails)

Is emergence then disproving free will? Emergence is the only possible definition of 'they' here. Which lacks intention and planning. Still means you can perceive it as forcing you / hostility compared to you individually.

Evolution. Your starting point on the evolutionary field of possibilities is determined outside of you, and limits choices/paths, as does every choice made along the way cut off certain paths and brings others in reach. One has limited control there by def, and no control for the first phase of life (what little control as child is there is while ignorant of consequences down the line wrt options).

This juxtaposes invidivual (having free will or not), and society (imposing full conditioning). Crux of complexity is group level, groups one is part of by conditioning/birth and by choice/seeking out within available pre-conditioned options. There is no influence free place, but doubt it's a prerequisite for free will. Is agency more useful term, as it does apply to groups, and by extension organisations and countries. Bit weird their mention in context of free will as if anthromorphisation is something actual.

Actually 'delivering targeted ads' based on protected data is inconsistent with the GDPR entirely.

so if I annotate and highlight something at the same time?

I wonder if it uses the W3C recommendation for highlighting and annotation though? Which would allow it to interact with other highlighting/annotation results.

To me highlighting is annotation, though a leightweight form, as the decision to highlight is interacting with the text in a meaningful way. And the pop up box actually says Annotation right there in the screenshot, so I don't fully grasp what distinction James is making here.

Kev is equating integration with any service to attempts to increase conversation around a post. That is often true but not always. E.g. I'm looking at AP to increase what own words I am sharing. E.g. AP for limited audience postings, and e.g. RSS for a subset of posting that are unlisted for the general public on my site.

Valid point Kev makes here. A site should do only what its author needs it to do. I want interaction visible on my site, though I probably will cut down on the facepiles.

I wonder if that would increase responses to my blog as Kev indicates. There might be those who will respond in e-mail, but not in a public comment. Worth a try.

https://de.wikipedia.org/wiki/Soziale_Systeme_(1984)

https://web.archive.org/web/20221020022908/https://www.spiegel.de/kultur/soziologie-flug-ueber-den-wolken-a-1ef2fd09-0002-0001-0000-000013511556

A disc by Dirk Käsler of Niklas Luhmann's 'Soziale Systeme', both from 1984. Käsler notes something I hadn't realised yet. Luhmann and Habermas are almost the same age, and Käsler in 1984 dubs them Western-Germany's best sociological and philosophical contemporary export products respectively. E did her master thesis on Habermas in comparison with blogs (H never said anything about internet). The Western-Germany context is also something to take note of I think. Reading the name made me realise how much context / cultural background there is in it. How different I experience Germany now from the 1980s when even a year before the fall of the wall that happening seemed a generation away. The summers I spent in Germany, all in the west, and the visit to East Germany in 1987. A different world compared to later visits on both sides of the former divide.

Tarkowski has only seen a few more participatory server governance set-ups, most are 'benevolent dictator' style. Same. I have seen several growing instances that have changed from 'dictator' to a group of moderators making the decisions. You could democratise those roles by having users propose / vote on moderators. It's what we do offline in communal structures.

We have that proof, once you let go of the notion that everyone should be on the same tech platform to count as social network. Offline social networks are by def decentralised and operate by consensus (or they split, change, move on), we build governance mushrooms on top of it (countries, democracies) that are the scaling layers, but not the building block. Mimick human networks digitally (which internet fundamentally already is) and it will be fine, messy as humans are, but fine in terms of utility and value derived by people using their tools. Tarkowski hits on key points, but seemingly takes the notions of scale and the centrality of a tech platform (and calling that the community) as a given, whereas I see them as the narrative that bigger tech platforms created. For them that narrative was necessary to be able to monetise, to get above the Cosean Floor with funding. There is no presumptive need for any social networks to be at scale for the entirety of human interaction to be at scale as a result. Community is not a synonym for platform or for the combination of a platform and its users. Approaching the fediverse from this 'big tech' framing more likely permeates the Paradox of Open it starts with than the resulting suggestions solve it. Escape it by working with the actual human meaning and level of social networks and the tools they deploy to interact, and ensure they can interconnect. E-mail, snail mail, SSB, IFPS make more sense here than Twitter, FB et al.

This again is where it derails: 'a new bigger Mastodon'. There is no Mastodon as such. That's the point to/of leverage.

Yes. It requires however letting go of starting from the tech as a single plaform or the presumption of scale, and starting from community and network. The group once formed shapes/defines the tech tools used. Then link those tools used up into a federated whole, that in turn may need its own governance structures. Vgl [[Waardevol op zich, waardevoller verbonden 20190513163855]] (valuable on its own, more valuable connected)

This makes sense. Not just for public institutions, also for other types of organisations. Public institutions have a leading by example role here.

Why is the apparant assumption here that it is needed to start from a perspective of global scale? Why one 'shared mission' e.g.?

I'm surprised that the underlying assumption (and tone) not just here but in most tech discussion of this type, is still that 'everything' around a tech tool should be done through that tech tool. Of course you need to organise around it, and professionalise that in the face of growth or becoming more central to some group's functioning. Obviously you need to leverage other types of governance and decision making than what went into creating a tech at first. Institutionalising is a time proven way to sustain an effort. Technology = politics. You need to be a politician in your own technology space. A politican in the artisanal and as a practice / behaviour sense, not in the occupation sense. Vgl [[Mijn werk is politiek 20190921114750]] which mostly implies thinking at different levels of abstraction about your situation simultaneously (Vgl [[Triz denken in systeemniveaus 20200826114731]] but then socially as well as tech)

PublicSpaces overtones here. They're not wrong. Presumably Paul Keller is involved in it?

This comes back to my 'tech smaller than us', governance and technological control over a tool should reside within the context-specific group of users using it. Which does not mean every single person within the group needs all the skills to control the tool, just that the group has it within itself, and decides on how that control is used a group. Vgl [[Networked Agency 20160818213155]] and [[Technologie kleiner dan ons 20160818122905]] n:: This would e.g. imply in Tarkowski's text that a community run instance would ensure having at least 1 member contribute code to Mastodon, and strategically operate to ensure that. Is this also an element wrt the above about Paradox of Open, as the non-monetary benefits of contributing may be well be enumerated as part of the operating costs of a community instance?

unless as I said servers and individuals are often the same. Not per se to run a Mastodon instance as discussed here, but to have a website that also supports #ActivityPub interaction

Indeed, and this is what many don't seem to take into account. E.g. calls for various centralisation efforts. The way 'out' of being under 'benevolent dictators' is to fragment more to the level where you are your own dictator and moderation decisions at individual level and server level are identical, or where you have a cohesive circle of trust in which those decisions take place (i.e. small servers with small groups)

The Paradox of Open (Source) is as Keller and Tarkowski formulate it, that the clear socio-economic value and tech impact of open source comes from underfunded / volunteer work. Vgl [[Bootstrapping 20201007204011]] and the role of precarity in it. Makes me think about 1937 Ronald Coase's transaction costs which in [[Here Comes Everybody by Clay Shirky]] is used to derive a Cosean Floor (and Cosean Ceiling). Openness allows you to operate below the Cosean Floor, but it seems that to bootstrap beyond that first stage is harder. Are such projects incapable of finding a spot above the Floor further down the chain, or are they pushed aside (or rather 'harvested') by those already positioned and operating above that Floor? Perhaps your spot above the floor needs to be part of the design of the work done below the floor. What's the link with my [[Openheid en haar grenzen 20130131154227]] in which I position openness as necessity to operate in a networked environment, and as necessarily limited by human group dynamics and keeping those healthy? Balancing both is the sweetspot in the complex domain. Are we any good at doing that in other terms than just social group behaviour in a room? #openvraag what's the online/bootstrapping equiv of it? How did we do that for my company 11 yrs ago (in part by operating something else above the Floor alongside)?

https://hans.gerwitz.com/about/ this one?

An eye-opening metric. 'souls saved' vs eyeballs. As in, the number of people with access to a community values reinforcing platform. Vgl [[EU Digital Rights and Principles]] and [[Digital Services Act 20210319103722]] more geared to things that aren't specifically aimed at the culture of interaction, but at things that can be regulated (in the hope that it will impact culture of interaction).

An actor has id and type properties and an inbox and outbox. It should also have a number of things (which I don't all see the use of, why should I disclose followers and followed e.g.?), and may have a number of things, including token endpoints which might be useful

this is akin to how Dave Winer added the capability to add source to a feed in the RSS spec. I read this as only of interest if you use the client to potentially edit the source. If I'd use WP to originate AP msgs, this would not be needed, as any changes would be initiated in WP anyway

dereferencing = following the link to the source, and look at its content, to see if it matches the representation in the json data. Does this imply you could leave it at the id for the object, and leave out the rest as it isn't accepted anyway? Below it says that every object has at least id and type, so yes to the above.

to address a message to followers

it is needed to be aware of the distinctions between server elements related to client interaction, and server elements related to server-to-server things. A server is basically 2 servers in 1 if it is fully capable of federatiion.

a special to group, meaning publicly available (e.g. on yr profile page, or anyone looking in your outbox.

any AP message is in context of ActivityStreams

this is the server to server step

Actors in AP can be apps, groups, orgs, people, services. It's a choice that Mastodon assumes people? What If I create an actor profile that says it's a group, would Mastodon know how to deal with it?

If it's my client posting to my server then it's of less concern, for the client to server side. Federation is by def not to be trusted.

You should do some evaluation of received content, but the spec doesn't provide an authorative way for verification.

https://www.w3.org/ns/activitystreams#activitypub this seems quite a list, and rather central to AP: All terms in this section are described in the [ActivityPub] specification.

endpoints (@id)
following (@id)
followers (@id)
inbox (@id) - This is an alias of ldp:inbox.
liked (@id)
shares (@id)
likes (@id)
oauthAuthorizationEndpoint (@id)
oauthTokenEndpoint (@id)
outbox (@id)
preferredUsername
provideClientKey (@id)
proxyUrl (@id)
sharedInbox (@id)
signClientKey (@id)
source
streams (@id)
uploadMedia (@id)

Objects are the core building blocks of AP, following ActivityStreams spec fully. Objects are wrapped in Activities, and can be part of Collections (itself an Object). Also follows the saem URI conventions

It is possible to have an AP client-server implementation that does not do the federation part, but it would be little effort to support it. And vice versa. Does this mean that in such a case every other server should come and get information, iow another server should know to come get it? Federation is more about push it seems. Not doing the federation part means limiting the amount of stuff you get to receive (all the other stuff). For my site it would be good to limit what's being processed. Currently it seems to me that maybe WP AP plugin isn't doing the federation bits?

posting to your outbox is supposed to be treated as a Create type activity and wrapped as such. Other activities (e.g. Like) should be deliberately expressed, but create is assumed if no activity is present.

Federation comes from POSTing to other server inboxes, rather than GETting from server's outboxes. This is the source of M being rather traffic heavy?

? So at core, AP would require cycling through people's outboxes to get the stuff that is available there. Such pulling may be enough for just specific circles of people. Would it be enough for dAPplr? Come and get it?

get and post behave differently on the client and server side. Get from inbox client side, is reading ones stream. Post to outbox client side is sending messages out. Post to inbox server side is incoming stuff from the fediverse. Get from outbox serverside is getting the msgs you're authorised to see. This last one would be how I'd approach e.g. dAPplr (I think I found my working title)

Every actor has an inbox and outbox. These are endpoints URLs listed in the ActivityPub actor's ActivityStreams description

a user is an actor is an account on server. Actors can be of different type (org, group, person, app and service) I suppose here person is usually meant, but I can see group and org useful too, as well as service (different content streams by me) and even app (source of content creation e.g.)

It seems all current implementations are both, but I wonder if separating them out creates lower threshold agency, much like how microsub separates the feed server and the reading client, and micropub separates the posting server and the writing client. I can see having multiple client side thing to be able to post or see content, even if the base case is my site being my client and my server. Keeping a clear distinction of what's what is always useful.

In sum, I'd need to: understand activities use my site to create those activities, and display those of others separate the client part on my site and the server part clearly in my mind. figure out how the server part processes incoming notifications and content.

The servers between eachother also have an API to deliver notifications and content.

the client creates/updates/deletes content, and uses an API to send that to server.

Activitystreams is a fundament for AP. This is why I'm interested, as ActviityStreams contains a much wider range of verbs than just for basic social interaction. Specifically covering things that used to have a social software service since gone (e.g. Dopplr, original Foursquare, Jaiku's way of combining streams into 1) Should read and mine the ActivityStreams spec for understanding too.

The various specs by the Social Web Working Group are listed here together. AP with the purpose of interaction is mentioned alongside MicroPub Sub and Webmention. The 'related specs' mentions IndieAuth and MF2. Useful to see it presented here as a 'family' rather than as alternatives, and underscores the gap that AP could fill for my otherwise IndieWeb enabled site.

Four elements to make 'score voting' more a cooperative effort. Status quo is one of the options to choose, not the default if no decision is made, adding options is always possible (meaning no limitative list of options, which would be giving a certain power to the maker of the list), everyone marks support/opposition to all options, not just favourites (score voting) and totals are tabulated (#openvraag how does this avoid 'brainless squid' results?), acceptance (meaning no or low opposition) rather than faving is main factor in decision making. That last one reads as pointing to a balanced dual indicator: the strongest attractor wins given the lowest barrier. So first determine lowest barrier options, than the biggest attractor amongst those.

Yoyogi is a tool that taps into your mastodon account (in your browser, locally) and shows messages by author / thread not as timeline. If you'd sort that like Fraidycat that would be a pretty interesting interface.

Reading is suggested to be potentially waning, maybe becoming more elite or even obsolete. It seems to disregard its counterpart: writing. For every thing that can be read, writing has preceded it. Writing, other than direct transcription, is not just creating text it is a practice, that also creates effects/affordances for the writer. Also thinking of Rheingold's definition of literacy as a skill plus community in which that skill is widely present. Writing/reading started out as bookkeeping, and I assume professional classes will remain text focused (although AR is an 'oral' path here too)

Tracy Durnell comments on a quote that divides internet users in 'word people' and 'image people' by by position the entire internet as a modern form of oral culture. The only reason in that perspective for the abundance of text is early bandwidth and technology limitations. Nowadays presentations, streaming, videos and podcasts make a much direct version of distributing oral expressions. When Durnell talks about oral culture is that because of the style more than the format? Blogs, IRC chats, microblogging and messaging are more oral in tone. Whereas ;'serious' texts are still in document shape. Reminds me of annotation as conversation and as social interaction.

Which is why DSA and DMA target larger entities beyond that start-up scale.