I just read Gian Segato’s “Building AI Products In The Probabilistic Era”. Good read. Good take. But, it makes sense. He’s a data scientist and we (as a community) have had to think this way for 10-15 years when working with narrow probabilistic models. But, the scope has changed. Inputs and outputs are open-ended. His examples around replit are good, e.g. constraining the use case to code gen for websites would prevent other use cases like code gen for games. ...
It would be nice to have a tiny store online that could just sit there for decades. Focus on one tiny and useful need and do it well, world class. Sell one or a few tiny products or services for that one thing. Not a full time thing, free content, self-serve product sales with on-demand support. I guess SuperFastPython fits the bill, but even it’s scope is too large. Ideally one thing. One algorithm, or one data structure, or one library/module. ...
I was listening to an Acquired episode and the guest mentioned the role of computers (people) being replaced by electronic computers, and how programming/coding will/is seeing the same fate. Here’s the episode: How is AI Different Than Other Technology Waves? (With Bret Taylor and Clay Bavor) And here on YouTube where I got the transcript. Here’s the quote: I self-identify as a computer programmer. It’s like the thing I like to do the most. And I’m like, man, that’s sort of like saying I’m—do remember like the calculators, people who like calculated things before we had calculators? I’m like, am I that? ...
Among Ronald Fisher’s many contributions to statistics, genetics, and evolutionary theory at large was his “fundamental theorem of natural selection”: “The rate of increase in fitness of any organism at any time is equal to its genetic variance in fitness at that time.” See: Fisher’s fundamental theorem of natural selection, Wikipedia. In plain language (with llm help): “The rate at which a population becomes better adapted (its average fitness increases) is directly proportional to the amount of genetic variation for fitness it already has.” ...
Machine Learning Street Talk is a YouTube channel and podcast by Tim Scarfe and sometimes Keith Duggar. I’m a fan of the podcast, and have been for some time. Follow them on X here: Machine Learning Street Talk Recently, like this northern hemisphere summer, they have been releasing fantastic episodes. Both the quality of the guests and the quality of the questions/discussion. Riveting stuff. Specifically: The Secret Ingredient for AI Creativity A chat with Kenneth Stanley. Did a 7 Year Old Really Solve the Universe? A chat with Guillaume Verdon. Mutually Assured AI Malfunction (Superintelligence Strategy) A chat with Dan Hendrycks. All three had be sitting and staring into space for a long time, thinking through things. Not sure I’m smart enough to keep up at double speed, a few listens are required. ...
I’ve created a new Quake archive, this time for the Navy Seals modification by Minh Le aka “Gooseman”. This was originally an ad hoc archive as a gist here: Navy Seals Quake I decided to make it a standalone archive as I came across rare/hard-to-find versions of the mod that I believed required independent hosting. You can see the full archive here: Quake Navy Seals
I listened to part 1 of the Acquired Podcast on the history of Google. Outstanding and nostalgic. I was a super early google user, thanks to a mention on slashdot on the late 90s. It got me thinking: what if… What if I travelled back to 2005 with my current knowledge, what would I do? What online business would I build? I was thinking about SEO and authority sites, based on the Acquired podcast episode on google. ...
After a long holiday where you eat like a pig and exercise not at all, how do you get back on the path? You must cut calories and get the body/brain used to the new regimen as fast as possible, e.g. avoid backsliding. Firstly cut all junk food and processed foods. Next, cut carbs. Then cut calories to below maintenance. All while exercising daily, as close to normal as possible. Don’t feel like going to the gym? Who the hell does? Fine, walk 5km. Lift at home. Get that body moving. Strain those muscles. ...
I created an archive of all official releases for Quake II (Quake2): Quake II Official Archive It complements my Quake Official Archive and Quake III Arena Official Archive. It seems that the Quake2 era circa (1998) was perfect for the wayback machine and almost all Quake2 files were archived. This Quake2 archive is the most complete of the three archives. The main files missing are a few solaris dedicated server releases, and these may or may not be “official”, more research is required.
I found and acquired a copy of the monograph: Public Knowledge, Private Ignorance, Patrick Wilson, 1977. From the first chapter: Scholars and scientists engage in attempts to make contributions to a public body of knowledge about the world. They do not work simply to increase their own private understanding of the world, nor simply to increase the understanding of their co-workers in a specialized branch of inquiry. Their work is incomplete until they have made their results public, available to anyone, now and in the future, who can understand and make use of them. ...