Reposted by michael 📈 👨💻ponder @ponder.ooo · 07/10/2026heartbreaking: this measure just became a target 3877
Reposted by michael 📈 👨💻Matt Henderson @matthen.com · 24/09/2026every (convex) quadrilateral is a square- if you view it from the right perspective 1231432
michael 📈 👨💻 @mikedecr.computer · 18/09/2026static.klipy.comStar Wars Revenge Of The SithALT: Star Wars Revenge Of The Sith 150
michael 📈 👨💻 @mikedecr.computer · 03/09/2026crazy how the fence turned green in the specific area behind your safety goggle lenses 010
michael 📈 👨💻 @mikedecr.computer · 28/08/2026I have no idea whether this DNA -> RNA transcription detail is retained in people's heads enough for this to land 010
Reposted by michael 📈 👨💻dad (aka miguel) @aka.dad · 27/08/2026the logs refer to me as "the event." 071
Reposted by michael 📈 👨💻value plus discounted @akhilrao.bsky.social · 26/08/2026become unpriceable 0133
michael 📈 👨💻 @mikedecr.computer · 25/08/2026you can do: thing.method_name( x=y ).method_name_2( is_z ) 010
michael 📈 👨💻 @mikedecr.computer · 24/08/2026Now available as a blog post: mikedecr.computer/blog/sugar-f...mikedecr.computermikedecr . computer | Sugar free: Porting the exact `ggplot2` syntax to Python is not important 120
michael 📈 👨💻 @mikedecr.computer · 21/08/2026tbf I am goofing around about c[.]name. I know it's inconvenient compared to R but there's some structural clarity at least in saying "this is a field of some struct" rather than hand wavey "this will be interpreted in some way, trust me". I like Julia's use of :symbols which kinda splits the diff 020
michael 📈 👨💻 @mikedecr.computer · 14/08/2026plenty of weird and tough things about getting older but one thing I really do love is not knowing what the hell people are talking about. I love not being bothered w/ stuff. "Alex Cooper's beverage business is failing." I don't know who that is or whether that's good or bad. you kids have fun 180
michael 📈 👨💻 @mikedecr.computer · 12/08/2026I read stuff like this and it's like, okay I just can't pretend to be nonchalant about it in situations like this: I. AM. BAYESIAN. 040
michael 📈 👨💻 @mikedecr.computer · 11/08/2026It's just tedious. I learned this stuff once so I could be wise forever. The deal was I become a freak about some things so I could then make good choices automatically because the payoff of caring is that your philosophies and perspectives just ooze out of you whenever you work 060
michael 📈 👨💻 @mikedecr.computer · 11/08/2026Anyway, my irk is that I'm not sure how to prompt around stuff like this. This is clearly just like a Stance you can have about your view of math x data or not. Am I really going to prompt about every possible thing I might encounter w/ this? Do I really need to tell models to "remember" this 140
michael 📈 👨💻 @mikedecr.computer · 11/08/2026that the pkg we're using to generate the basis fns won't set knots outside the range of X, because there are no data out there to fit. You... foolish... insolent... swine! Basis fns as an abstraction should not be concerned w/ model fitting. Just give me the fns & we'll penalize like adults. Bah! 120
michael 📈 👨💻 @mikedecr.computer · 11/08/2026there's a lot of what I would just call "expertise" or "experience" that lead me to get very frustrated with LLMs sometimes. Today's example: generating basis spline fit for me (great use case bc I hate spline boilerplate). But I want to set knots that go outside the range of X. I get an error... 140
michael 📈 👨💻 @mikedecr.computer · 09/08/2026it looks cool but red flags to me: anonymous author, no posts except to promote the software. worth an inspection before yolo install 120
michael 📈 👨💻 @mikedecr.computer · 07/08/2026how much value is there in the answer to this question for you 100
michael 📈 👨💻 @mikedecr.computer · 02/08/2026do I know any data engineers or software devs who work on research tech? There will be some hires on my old team working with our research platform / financial market data pipelines / distributed compute etc. Crazy data and very cool tech where Python meets high-perf C++ 055
michael 📈 👨💻 @mikedecr.computer · 01/08/2026went to a skatepark today, did not die, no broken bones 010
michael 📈 👨💻 @mikedecr.computer · 24/07/2026This is very Intuitive in the academic high praise sense 010
michael 📈 👨💻 @mikedecr.computer · 23/07/2026gf and I have been going to a nearby parking lot to skateboarding. she's learning fast. and I'm gonna hurt myself for sure 130
michael 📈 👨💻 @mikedecr.computer · 21/07/2026It's perfectly fine to have a principled commitment to empiricism and scientific discipline, actually. And when political actors try to bully you into being quiet or changing your story, it shows your integrity and bravery to resist that pressure. Right or left. 380
michael 📈 👨💻 @mikedecr.computer · 21/07/2026Even taking an average is a model that extrapolates from data to the thing you want to study 100
michael 📈 👨💻 @mikedecr.computer · 21/07/2026it's fun to philosophize about science but we can also be specific and pragmatic. At least a hardcore scientific nihilist has principles. But dissembling about the fallibility of data and empirical inference over here but not over there is motivated reasoning 001
michael 📈 👨💻 @mikedecr.computer · 21/07/2026is "average global temperature" an ontologically real, self-evident thing or an extrapolation from local measurements using a model? Is "climate change" something we can see by fixing CO2 levels and replaying counterfactual history or a modeled inference that combines observations and assumptions 300
Reposted by michael 📈 👨💻Chris Steller @chrissteller.bsky.social · 16/07/2026Someone just sent me this photo from Chicago, unbelieveable 35183711145
michael 📈 👨💻 @mikedecr.computer · 13/07/2026it also costs you nothing for senators to be civil when their coworkers die 010
michael 📈 👨💻 @mikedecr.computer · 05/07/2026Classic non-researcher orientation toward the world to think that their questions could possibly be answered. Real researchers know that nothing has ever been, or could possibly be, explained 010
Reposted by michael 📈 👨💻Kuba Suder 🇵🇱🇺🇦 @mackuba.eu · 03/07/2026RIP to everyone techdebted by the LLMs, but I'm different. And better. 4563
Reposted by michael 📈 👨💻amos @fasterthanli.me · 30/06/2026therapist: and... would you say this flips the story? claude: you're right *sobs* therapist: I'm right tooooo? claude: you're right to push back on this. *wails* 1874
michael 📈 👨💻 @mikedecr.computer · 30/06/2026Safari will exit full screen if you hit the Esc key. it's infantilizing 000
Reposted by michael 📈 👨💻The Great Pumpkin Papers 🎃🎃🎃 @professormusgrave.bsky.social · 10/06/2026Kalshi now allows you to place Pascal’s Wager 1034671
michael 📈 👨💻 @mikedecr.computer · 10/06/2026nah. the world is high dim and there's no perfect work. Tradeoffs abound. My wish for academia is that they recognize that collab between ideas ppl and computer/stats ppl would be good but ego or "incentives" won't allow it and the generational crisis of science will continue. What more is there? 130
michael 📈 👨💻 @mikedecr.computer · 09/06/2026Something in the water.. academics keep posting about all the ill-advised things they are doing in the computer. This post is a breath of fresh air 110
Reposted by michael 📈 👨💻Thomas Lumley @tslumley.bsky.social · 09/06/2026Also, people now need to use as.factor() a lot more often, and they don't necessarily realise. notstatschat.rbind.io/2024/03/22/f...notstatschat.rbind.ioFactors as factors - Biased and Inefficient 1134
michael 📈 👨💻 @mikedecr.computer · 09/06/2026This isn't "complex" for people but there is no expressive range for the user so no point to do it manually. It's just data entry. You might enjoy it or be fast at it but it's understandable that this stuff is not fun for many people and not what they want to/should be focusing on. 020
michael 📈 👨💻 @mikedecr.computer · 09/06/2026- you have one database of sources that all projects can use so you never re-write refs across papers - software type checks your work so if you ever need to do something to ALL refs it is impossible to miss one - author name spelling etc handled in one place so random typos are not possible (...) 110
michael 📈 👨💻 @mikedecr.computer · 09/06/2026content away from the formatting style, so the author can focus on the content. Formatting is programmatically handled bc there are no(!) degrees of freedom. Journal A rejects your paper and Journal B has a different style. Should this be hours of work or one config change? Other benefits include: 130
michael 📈 👨💻 @mikedecr.computer · 09/06/2026Some people are gloating a bit too much so I will answer in earnest. There are many styles for formatting inline cites, foot/endnote cites, and references in the back. From the author's perspective you only need an ID for the source and maybe a page number. Citation software abstracts the (1/n) 110
michael 📈 👨💻 @mikedecr.computer · 07/06/2026i bought a skateboard so you know I'm early to hit m*d l*fe cris*s 020