Socio-technical trailblazer. I help civil society organisations use the internet for good: connecting people, processes, data, and technology.
- About me and my work
-
Glad to close off Friday night with an actually working setup of #Ruby, #Rails, #Hotwire, and #Selenium in a #docker desktop environment. Thanks to #ChatGPT, #duckduckgo and #StackOverflow for a quick-ish way into all of this 😉 It feels very declarative, let's hope it still works next week!
-
Github published a first video in a series on software development and accessibility, which also demonstrates the value of open source code. #a11y #opensource "GitHub celebrates the ingenuity of developers with disabilities in new video series"https://github.blog/2023-03-14-github-celebrates-the-ingenuity-of-developers-with-disabilities-in-new-video-series/
-
From TL;DR to TL;DW, thanks to Large Language Models… #LLM #ChatGPT #AI "AI Written, AI Read"https://marketoonist.com/2023/03/ai-written-ai-read.html
-
A talk about how twitter came to be, by @rabble, as historical (and technical) context for the current state of social media. Every time I hear it, I'm surprised by the twists in twitter's story -) #fediversehttps://youtu.be/B6YQQC5Q_5g?t=10141
-
#RFP for a #facilitation team for the Online Progressive Engagement Network [OPEN] Global Summit in Warsaw, May 29-June 1. https://www.the-open.net/work-with-us
-
@BrianJohnson Yes, "chronological" is an algorithm too… When social media was about social connections, we tried to organise around places, groups, meetings, and such. With current Mastodon, we sort of only have hash tags and "drinking from the hose". Cal Newport's thought experiment: what if current platform algorithms are deemed not "neutral/selection", but "editorial". I…
-
Nice thought experiment: What if the US Supreme Court would say: a platform provider is not the publisher of information, but if you use technology to select and push part of the content to users, you do take on responsibility. Mastodon/the #Fediverse doesn't use selection algorithms, and figuring out again how to organise and follow…
-
one thing less to worry about…
-
Ok, #graph databases at #fosdem, self-assigned homework: study "embeddings" in general, my "hand wavey" understanding isn't cutting it anymore 🙂 but working with HashGNN node embeddings in #Neo4J looks interesting.
-
First #fosdem session: transforming EU tender data into #OCDS open contracting standard data, and adding it to a #Neo4j graph interface. Same old same old: entity recognition and deduplication on messy input data. #IATI memories… https://tedective.org/