mastodon
4.5.7
Running https://fathom.info since 2010 / spent 22 years building https://processing.org and supporting its community / sometimes seen teaching at https://mit.edu
Running https:// fathom.info since 2010 / spent 22 years building https:// processing.org and supporting its community / sometimes seen teaching at https:// mit.edu
Did OpenAI talk to Rosalind Franklin’s family about co-opting her good name this way?
Is that a dumb question when their CEO uses an AI slop avatar of himself in the “style” of a Miyazaki character in spite of Miyazaki calling AI imagery “an insult to life itself”?
https://arstechnica.com/science/2026/04/openai-starts-offering-a-biology-tuned-llm/
Running https:// fathom.info since 2010 / spent 22 years building https:// processing.org and supporting its community / sometimes seen teaching at https:// mit.edu
“I am, thank God, curiously detached from all this. Four and a half years ago I was minding my own business, cutting metal in my machine shop, when I received a text message from John Gaeta, a former colleague at Magic Leap, reading simply “Sorry for your loss.” At first I thought that he’d sent it to me mistakenly, but after a bit of Googling I became aware that Facebook had changed its name and announced that it was now going to build the Metaverse.”
“In retrospect, John’s message was prescient, since it marked the moment when the Metaverse really did break free and become my alienated, prodigal brainchild.”
https://nealstephenson.substack.com/p/my-prodigal-brainchild
Running https:// fathom.info since 2010 / spent 22 years building https:// processing.org and supporting its community / sometimes seen teaching at https:// mit.edu
Spoke at a panel at Northeastern last week about careers in information design & data visualization.
With 10 minutes allotted, it’s not a comprehensive guide to finding your path in life, but I wanted to pose a few questions that might help students focus their search:
Running https:// fathom.info since 2010 / spent 22 years building https:// processing.org and supporting its community / sometimes seen teaching at https:// mit.edu
RE: @justvanrossum@typo.social
Ah, the Alien film titles regex…
Running https:// fathom.info since 2010 / spent 22 years building https:// processing.org and supporting its community / sometimes seen teaching at https:// mit.edu
Which is even worse, right? Acknowledging that there's a problem but missing the actual problem and instead doubling down on it?
LLMs are not a sampling technique, it's non-deterministic autocomplete. So the answer that comes back isn't within some understood range, you can ask the same question and be 5%, 50% or 100% incorrect each time it's asked.
(And yes, you can do lots of things to mitigate all these issues, but again, wtf are we even doing at that point? These are solved problems! We have well-defined and efficient methods, along with supercomputers for laptops and instead we're blowing all that power on petascale ELIZA.)
Running https:// fathom.info since 2010 / spent 22 years building https:// processing.org and supporting its community / sometimes seen teaching at https:// mit.edu
Running https:// fathom.info since 2010 / spent 22 years building https:// processing.org and supporting its community / sometimes seen teaching at https:// mit.edu
Running https:// fathom.info since 2010 / spent 22 years building https:// processing.org and supporting its community / sometimes seen teaching at https:// mit.edu
Running https:// fathom.info since 2010 / spent 22 years building https:// processing.org and supporting its community / sometimes seen teaching at https:// mit.edu
I simply cannot get my head around why anyone would let AI anywhere near their data analysis work.
Why would you add something to your work that can drop its accuracy by half? by 10%? by 1%? What would be an acceptable amount? What is the possible upside that would make this worth it?
I'm floored by the number of people in this field who take themselves all too seriously but are out there starting their whatever *dot AI* companies to get in on the grift, or say things like “and of course, AI” like it's both obvious and inevitable. And how is this so shiny that even educators have taken leave of their senses?
What the f*k is the point of analyzing a dataset if you're ok with answers simply being incorrect? What are we even doing here?
Running https:// fathom.info since 2010 / spent 22 years building https:// processing.org and supporting its community / sometimes seen teaching at https:// mit.edu
“Porting Mac OS X to the Nintento Wii”
This made my day—first, that there are people out there who do things like this, but second (and more to the point), that he also took the time to document the process in detail.
So much to be learned here: https://bryankeller.github.io/2026/04/08/porting-mac-os-x-nintendo-wii.html
Running https:// fathom.info since 2010 / spent 22 years building https:// processing.org and supporting its community / sometimes seen teaching at https:// mit.edu
RE: @benfry@information.garden
Just finished; wow, this was *fantastic*
(for someone who is very curious about China, tech history, business, supply chain, how things are made, materials and manufacturing, design process and scale, the intersection of politics/policy/business/history, etc etc…)
Running https:// fathom.info since 2010 / spent 22 years building https:// processing.org and supporting its community / sometimes seen teaching at https:// mit.edu
Fall is the best, but spring in Boston is a close second. Animals re-emerging, plants are blooming, and massive flakes of heavy, wet snow are falling on this fine morn, the seventh of April.
Running https:// fathom.info since 2010 / spent 22 years building https:// processing.org and supporting its community / sometimes seen teaching at https:// mit.edu
Unintentionally celebrating Apple's 50th anniversary today having just started “Apple in China” which so far is the best retelling of their history I've read in years.
Running https:// fathom.info since 2010 / spent 22 years building https:// processing.org and supporting its community / sometimes seen teaching at https:// mit.edu
Incredible thread about the source for Claude Code’s CLI being leaked due to a misconfiguration.
Includes analysis of the many ways that the code is a mess, and how the mess is so extreme that it's practically its own coding pattern.
Much hilarity, starting on April Fools' Eve:
Running https:// fathom.info since 2010 / spent 22 years building https:// processing.org and supporting its community / sometimes seen teaching at https:// mit.edu
Was finishing Andy Grove's “Only the Paranoid Survive” while on my walk to work this morning, and was about to post about how older tech business books are often better as unintentionally hilarious history, with pronouncements like:
“But as my knowledge deepens, my conviction is growing that the triad of software coming from personal sources, from telephone and network sources and from the Internet will together be what will drive our industry in the years ahead.”
And the next sentence that follows?
“My conviction is also growing that the media and the advertising industries represent a growing opportunity for us.”
…and I arrive to work to find this on my timeline:
Running https:// fathom.info since 2010 / spent 22 years building https:// processing.org and supporting its community / sometimes seen teaching at https:// mit.edu
Still working to get the right marketing message for Rowboat, but in the meantime, lots of updates this month: improved cell editing, map improvements, plus a fresh desktop release: https://rowboat.net
And for the superfans, all the gory details on the release page: https://docs.rowboat.net/releases/ 💪
Running https:// fathom.info since 2010 / spent 22 years building https:// processing.org and supporting its community / sometimes seen teaching at https:// mit.edu
Nearly every 7 digit number can be found in the first 100 million digits of pi.
Search for your favorites, or simply bask in the wonders of math and probability: https://www.benfry.com/pi/
Happy π day!
