Ben Fry @benfry.com · 16h“Post hoc ergo fucking propter hoc.” letterror.com/ailetterror.comOn generative AI at LettError.com 220
Reposted by Ben Frymattie lubchansky @mattielubchansky.com · 29/09/2026awoke this morning with a Vision 4990051654
Ben Fry @benfry.com · 26/09/2026Loved this story from a woman talking about having a functional relationship with her husband, and how they're supporting each other and their kid. 030
Reposted by Ben Fryryan cooper @ryanlcooper.com · 26/09/2026great piece. man social media is a cancer themomoftheyear.substack.com/p/im-the-mom...themomoftheyear.substack.comI’m the Mom in That Viral Giants Clip. Let Me Tell You About My Husband.You saw me carrying our baby and food. There was more happening in our family that night. 12352831359
Ben Fry @benfry.com · 24/09/2026…while Paul and I plotted about what a modern type specimen might look like when your input is the trove of information in a modern font file. 000
Ben Fry @benfry.com · 24/09/2026…Katherine figures out the color changer on the knitting machine and makes a scarf… 110
Ben Fry @benfry.com · 24/09/2026…Karlie donning headphones to make weird sounds with a new AMYboard… (www.amyboard.com) 100
Ben Fry @benfry.com · 24/09/2026To celebrate the launch, we spent the afternoon on projects around the studio… 100
Ben Fry @benfry.com · 18/09/2026“Microsoft recorded 83–93% drops in click-through for some news plaintiffs & 51–94% for others. Suddenly, it becomes easier to see how declining news revenue could rob chatbots of the abundant streams of reliable information that makes them such groundbreaking tools.” arstechnica.com/tech-policy/...arstechnica.comMicrosoft exec called AI scraping the “largest theft of labor in human history”Microsoft, OpenAI emails reveal fear of AI “doom loop” killing news orgs. 010
Ben Fry @benfry.com · 17/09/2026Having passed this hurdle (a paper! a usable tool!) our next steps are all about digging further into that last question. How can we better focus on the *meaning* of the data, rather than getting distracted by the data itself? Lots of ideas about what that entails, but I'd hate to ruin the surprise. 010
Ben Fry @benfry.com · 17/09/2026How can they poke at it, look at it from different sides, play with it? It's not about single answers, it's about building a mental model of what the data *means*, whether for a research paper about how an outbreak has spread, or for a policymaker to make decisions about outbreak response. 100
Ben Fry @benfry.com · 17/09/2026More effective is when they can actually engage with it: building, interacting with other students, etc. The same thing applies for a dataset: instead of users sitting back and passively “reading” data, or visualization of it, we want to build tools to help people interact with it. 100
Ben Fry @benfry.com · 17/09/2026Part of understanding a dataset is the idea of being able to interact with the data to learn about it. This is something I take from my days during undergrad working in child development. The basic idea in constructivism is that you can't just shovel information at kids and have them memorize it. 120
Ben Fry @benfry.com · 17/09/2026Like the map of a hurricane path, with its error bars that everyone agrees is a range yet mentally perceives it as an actual path, a tree can be a distraction. So we've been trying to lean into that, and showing the haziness everywhere we can. 100
Ben Fry @benfry.com · 17/09/2026Another thing we enjoyed about this project is the ability to visualize the uncertainty, ranges, and error behind its calculations. The output of this analysis is a phylogenetic tree, but like all such trees, it's inherently a lie. It's the least worst lie we can come up with, but it's still a lie. 100
Ben Fry @benfry.com · 17/09/2026(That last part is a post for another time. This is already too long.) 100
Ben Fry @benfry.com · 17/09/2026Instead of spending all your time cleaning data first, there's essentially no cost to taking a quick look and proceeding with your research. That's really powerful, because *all data is bad* and the pursuit of what passes for “good” is both time consuming and has its own kinds of badness. 100
Ben Fry @benfry.com · 17/09/2026While it's always nice to run faster, what's *actually* exciting about that is the way that it can change the way you work. It means trying things early instead of waiting for perfection. You can immediately identify overall trends, bad data, outliers, etc and fix these problems before the next run. 100
Ben Fry @benfry.com · 17/09/2026The project is the brainchild of the inestimable Patrick Varilly, and the tool is 100-1000x faster(!!) than others. For our part, we worked with Patrick on how to make Delphy usable & accessible beyond the handful of PhDs who know the necessary command line incantations to run this kind of analysis. 100
Ben Fry @benfry.com · 17/09/2026We have a paper in Nature today! www.nature.com/articles/s41... Several years in the making, Delphy (www.delphy.bio) is one of the tools for outbreak response that we've been building over the last several years with Pardis Sabeti over at the Broad Institute. 281
Ben Fry @benfry.com · 15/09/2026This also seems like a terrible path for universities: if you start “scaling” the way the rest of these platforms do, they become more generalized/commodified/whatever term you want to use… and you're back to students waking up to “what the hell am I paying for?” the way they were during COVID. 110
Reposted by Ben FryDon Moynihan @donmoyn.bsky.social · 13/09/2026A recurrent theme in this excellent story about China’s progress is that people there regard Trump as essential to their success, including his ability to drive scientific talent from the US www.newyorker.com/magazine/202... 361742737
Reposted by Ben FryBen Recht @beenwrekt.bsky.social · 12/09/2026The right way to pace the frontier is to commit to a robust, open-source AI ecosystem, thereby diluting the power of these badly behaving frontier labs. 5599
Ben Fry @benfry.com · 12/09/2026News you can use: “Noctuary is to night as diary is to day.” www.wordnik.com/words/noctuary (Or maybe it just seems more interesting after several hours of not being able to sleep) 010
Ben Fry @benfry.com · 11/09/2026There's a 10% chance that Rowboat will kill us all before the end of the decade. Until then, I hope you'll find it's drastically faster, more accurate, less expensive, and more fun than using an LLM to play with large datasets. Our latest: buttondown.com/rowboat/arch... Give it a try: rowboat.net 0134
Ben Fry @benfry.com · 11/09/2026As someone else said: “wake me up when an ‘agent escapes containment’ and starts paying off medical debt.” Until then, it's all human error. 061
Ben Fry @benfry.com · 10/09/2026amazing; also impressed that 1) someone had the prescience to run a separate network for artists, and 2) someone else bothered to decode the messages and identify them as JPEG data 110
Ben Fry @benfry.com · 10/09/2026Haha, straightforward enough that I almost didn't mention it (and don't want to be that guy), but shared since I thought it would be especially funny if it actually closed the loop. 110
Ben Fry @benfry.com · 10/09/2026We did an installation early 2001 that was extremely prone to crashing due to the sensor hardware: acg.media.mit.edu/projects/atm... Late in the install, purchased a machine fast enough that it could crash & even restart in the split second between using the sensor & glancing up to see the result.acg.media.mit.edu a t m o s p h e r e 010
Ben Fry @benfry.com · 10/09/2026Ha, yes—that may even be a method he picked up from Golan Levin who picked it up from me: installations are always prone to crashing, so you'd deploy on a Windows machine with a run.bat file that read @echo off .\installation.exe run.bat 220
Ben Fry @benfry.com · 10/09/2026Neat idea: instead of relying on attempts to perfect systems that will never be perfect, focus on warning when they're about to break down. Used here for autonomous vehicles, but a broadly applicable approach to making complex systems more tractable. news.mit.edu/2026/system-...news.mit.eduSystem helps humans predict when self-driving cars will make mistakesThe CW-Net technique explains the behavior of an autonomous vehicle, using concepts a human can easily understand. Researchers found these explanations helped drivers predict how a vehicle would behav... 171
Ben Fry @benfry.com · 09/09/2026The new Duo rumored to be announced by Apple today is looking sweeeet 051
Ben Fry @benfry.com · 06/09/2026Ha, they've deleted the post, but the kids in the picture are still bored 020
Ben Fry @benfry.com · 06/09/2026They're just trolling me for not donating last year, right? Of all the institutions I would not have expected to lazily jump on this destructive LLM bandwagon… 110
Reposted by Ben FryPaul Rohr @pevohr.bsky.social · 13/01/2026From the folks behind rowboat.net: "We want to make thoughtful products, where it’s clear that someone has spent time thinking about the hard parts of a task and has invested the time and effort into making that a smooth process." 011
Ben Fry @benfry.com · 04/09/2026“After 2029 they increasingly turned their attention to much more difficult problems, like making an AI companion who could convincingly fake empathy and concern.” www.math.brown.edu/reschwar/Sto...math.brown.edu 010
Ben Fry @benfry.com · 04/09/2026Ooh, and they have an apps solution too, since I'd also like to get us off Google Apps… (We started using it since it was the only way to get our email actually delivered, since running your own Postgres server for a, uh, dot info domain is not the path to righteousness.) 110
Ben Fry @benfry.com · 03/09/2026For all the effort I make to quiet the “just build that yourself!” voice in my head, the rent seekers and nazis are making it nearly impossible these days. Let's see, what else do we depend on… Notion? No way that ends well, right? 061
Ben Fry @benfry.com · 03/09/2026We *just* finished migrating off of Harvest after its acquisition by Bending Spoons. And now it's time to move off 1Password… www.theverge.com/tech/988536/...theverge.com1Password wades into a right-wing mess after funding a Linux projectThe controversy surrounds David Heinemeier Hansson, who has posted strident anti-vax and anti-immigrant views. 240
Reposted by Ben FryFathom Information Design @fathom.info · 02/09/2026This summer, we welcomed Rekha as the latest member of our team! Rekha is a developer and designer with a background in theoretical linguistics and interactive journalism. 141
Reposted by Ben FryDavid_j_roth @davidjroth.bsky.social · 02/09/2026I love to live in the neighborhood of Autechre Heights. 2747161
Ben Fry @benfry.com · 31/08/2026Returning from vacation and happy to see that the squirrels were continuing to improve Rowboat while I was out. “Inline representations” break up cells by category & color, making it easy to scan down a column and see what's similar & different. Now even easier w/ multi-column selection: 030
Ben Fry @benfry.com · 21/08/2026Difficult to read this story of how the insane cuts to science funding affect the best people, and harm some of the best work being done: www.statnews.com/2026/04/07/b... Archive link: archive.is/zsAF3statnews.comA star scientist showed that better genetics lessons could reduce racism. It was the death knell for his careerBrian Donovan had persuaded high school teachers and education researchers that prejudice might be ended by changing how genetics is taught. 130
Ben Fry @benfry.com · 20/08/2026Felt obligated to pick this up just as a thumbs up to the author for its title, but I'm half an hour in and we already have this banger from Paulina Borsook: “As many political schools of thought do, these techno-libertarians make a philosophy out of a personality defect.” 0122
Ben Fry @benfry.com · 19/08/2026Thanks! and as it turns out, the piece I was preparing for the show was the original for the Nature cover: 010
Ben Fry @benfry.com · 19/08/2026(This was a sketch on the way to creating this piece: www.benfry.com/isometricblo... where I was thinking through ways of showing interconnections, getting proportions to work, and getting a feel for variation in the data.)benfry.com isometricblocks | ben fry 010