Excellent description of configuring OBS et al for streaming code/development.

"That demo got the attention of venture capitalists. And when a cool-looking magical thing gets the attention of venture capitalists, discourse tends to spiral out of control." Good, even-handed look at GPT3. It's both impressive and unexciting for me – there are so many underlying issues besides the 'magic', not to mention the relative failure rate, the complexity of any real-world deployment, and as ever, a lack of nuance in a lot of media about discussing text-generation. This lays out some of the points with the latter well.

"In this way, Dynamic Yield is part of a generation of companies whose core technology, while extremely useful, is powered by artificial intelligence that is roughly as good as a 24-year-old analyst at Goldman Sachs with a big dataset and a few lines of Adderall."

This is good – and largely well written, bar an unnecessary cheap shot at one point. It overlaps with lots of what I have to teach students about AI: namely, those letter have become this huge suitcase concept for anything from gnarly machine learning problems and recurrent neural networks down to applied statistics and a splash of arithmetic. And meanwhile, everyone just keeps adding to this cyclone of nonsense as they try to out-claim one another. It's exhausting, and it pollutes the public sphere, such that inexperts – politicians, policymakers – get themselves tangled up about all the wrong things. Sigh.

Highly recommended by SimonW. Possibly something to prod.

This is very good stuff from Kars: from the challenges of designing with machine learning through to Value Sensitive Design and the complexity of good work.

On the problems of machine-learning and medical data.

"From 1950 to 1990, Tinsley had been the world champion of checkers whenever he wanted to be. He’d occasionally retire to work on mathematics or devote himself to religious study, but he’d eventually return, beat everyone and become champion again. In that 40-year span, he lost five total games and never once dropped a match." Brilliant article from Alexis Madrigal on the race to solve draughts/checkers, one man and his computer, and another man and his faith.

Only read part one so far, but is proving useful for at least wrapping my head around a few concepts.

As used by fullest.house – perhaps something to play with in due course.

"A Big Hairy Audacious Goal (BHAG) is a strategic business statement similar to a vision statement which is created to focus an organization on a single medium-long term organization-wide goal which is audacious, likely to be externally questionable, but not internally regarded as impossible." As discovered in this week's crit.

Some useful reference points in here – bookmarking for when I actually have time to reutrn to it.

"As serious intellectuals often do, we spent hours discussing these questions, what data we would want to collect to answer them, and even how we might go about collecting it. It sounded like a fun project, so I wrote a program that takes video captures of our Mario Kart 64 sessions and picks out when each race starts, which character is in each box on the screen, the rank of each player as the race progresses, and finally when the race finishes. Then I built a web client that lets us upload videos, record who played which character in each race, and browse the aggregated stats. The result is called Kartlytics, and now contains videos of over 230 races from over the last year and change." Yes, it's a plug for manta, but it's also a nifty piece of engineering.

"Termshows are purely text based. This makes them ideal for demoing instructions (as the user can copy-paste), making fail-safe "live-coding" sessions (plain text is very scalable), and sharing all your l33t terminal hacks." Really lovely: record terminal activity, upload it to a URL, share it with others, dead simple. And the client playback is all javascript. Lovely.