Blog
see also: Making Data Mistakes consulting blog on data science, software, and management.
All blog posts (newest first)
(or navigate by tags)
-
Human readers deserve human writers: a policy on AI and human craft
( - 579 words)
The magic hammer of AI makes everything look wonderfully nail-like, especially for the solo creative who finds in it both trampoline and anchor. But I've decided that human readers deserve the many tiny choices that make creative work personal, reserving my automation for the machines who won't notice the difference.
-
Product idea - an episode-focused podcast player
( - 301 words)
Podcast apps still think in terms of shows, but what if they recommended the best episodes instead—using AI to parse transcripts and behavioural data to surface hidden gems across the entire medium?
-
I refuse to use this technology, even if it would make me money
( - 317 words)
When AI systems produce hate speech, technologists dismiss it as mere 'bugs' - what would it actually take for us to refuse profitable technology on moral grounds?
-
Psychological effects and cognitive illusions in AI-assisted programming
( - 1,231 words)
We judge AI's coding mistakes as permanent character flaws whilst excusing our own as situational—and find a precarious flow. What can psychology experiments tell us about how it feels to program with AI assistance?
-
How can we measure the effectiveness of an AI tutor?
( - 578 words)
Perhaps our future grades will depend not on our exam performance, but on how well our digital avatars handle a battery of simulated tests.
-
What new developments will we need for personalised, relevant one-to-one AI tutoring?
( - 604 words)
Future AI tutoring may require more than prompt engineering—perhaps custom teacher models that learn alongside students, plus explicit student models to predict and optimize learning outcomes.
-
We should de-stigmatise suicide, our innermost freedom and right
( - 244 words)
My greatest fear has never been dying—it's not being able to die if I want to.
-
Metaphors for depression, clean language, and the DSM
( - 671 words)
Could we diagnose depression by the metaphors people use to describe it? Black veil, gravity, suffocation - each might reveal different underlying patterns requiring different treatments.
-
A field guide to AI-first development
( - 5,817 words)
It's a bazooka strapped to a chainsaw, and the default is that you cut off your arm rather than build a house with it.
-
Mindstone AI London 2025-Feb-18
I'll be performing a live dance recital with Cursor, my favourite AI-assisted programming editor—because the best way to show AI partnership is through the improvised choreography of real-time coding.
-
Your team needs a default, glue system for knowledge - use Notion
There's a tension between wanting sharp, specialised tools and the powerful network effects of having everything in one central place—Notion threads this needle surprisingly well.
-
Podcast episode on the neuroscience of workplace learning
Greg Detre discovered that cramming teams into intensive workshops is scientifically designed to fail—after a year of monthly sessions focusing on 'sharpening blunt axes,' his Channel 4 data team transformed completely.
-
Making Tea While AI Codes: A Practical Guide to AI-assisted programming (with Cursor Composer Agent in YOLO mode)
We've progressed from razor-sharp manual tools to chainsaws, and now someone's strapped a bazooka to the chainsaw—powerful, sometimes scary, but with the right approach, transformative and addictive.
-
Ease/value ranking
Sometimes we're better at breaking down an overall, intuitive judgment into narrower, structured criteria first—otherwise we're trying to weigh all these facets against one another simultaneously.
-
What will AI do to jobs?
( - 381 words)
Even if AI-assisted programmers can do the work of ten, we'll build orders of magnitude more software—but technological unemployment will outpace this Jevons Paradox by the 2030s.
-
Pronouns & plosive sounds - can you change society with a single word?
( - 1,157 words)
Could adopting a single pronoun for everyone subtly reshape how we think about gender? Language creates perceptual boundaries in our minds—and changing those boundaries might change society.
-
What is wisdom?
( - 274 words)
Wisdom is often about truths whose opposite are truths - balancing seemingly contradictory principles like self-love and self-honesty, accountability and forgiveness.
-
Tensions between Engineering, Commercial, and Product?
Just as in a suspension bridge, the structural tensions between Engineering, Commercial, and Product teams are the key to holding everything together and building a successful company.
-
Advice to a younger self
( - 228 words)
Honestly doing the exercise is far more interesting than reading other people's answers. What would you tell your younger self about procrastination, balance, and learning?
-
The real reason we can’t define 'Artificial General Intelligence'
( - 271 words)
Could it be that any clean, broad definition of general intelligence would exclude us? Our intelligence is only somewhat general, existing on a continuum rather than being binary.
-
Buy Microsoft stock as they add AI to every one of their products
( - 629 words)
Microsoft makes AI integration just about possible across their product suite—frustratingly clunky today, but enterprise customers will choose familiar pain over third-party complexity.
-
Worst scientific experiment ever?
( - 587 words)
I changed almost every variable in human life simultaneously—diet, community, silence, meditation, environment—then felt frustrated as a scientist trying to identify the active ingredient.
-
Abstractions are distractions to an AI-assisted programmer
( - 439 words)
Might AI assistants, who can type faster and memorise vast codebases, actually perform better with less abstract programming languages and frameworks?
-
How to kick off a successful project
We'd nearly finished the algorithmic work when IT delivered the crushing news: no room on their roadmap this year—a painful reminder that brilliant code means nothing without early stakeholder buy-in.
-
How to make your software team much more effective
These factors are multiplicative, so if any of them are completely broken, overall effectiveness can drop towards zero—the hidden mathematics of why brilliant teams sometimes accomplish nothing.
-
Postmortems: The single most important keystone habit for a team (step-by-step guide)
My habit of reading books about habit-formation is stronger than any of the habits they've helped me form—perhaps why I'm drawn to keystone habits that catalyse other positive changes.
-
Make it work, make it right, make it fast
I can't count the number of times I've spotted people trying to correct code that doesn't run, or speed up code that isn't right yet.
-
How Humans and Machine Learning Deliver the Best Result (Podcast)
The best chess player twenty years after Deep Blue wasn't human or machine, but a Centaur team—proving that even in domains as black-and-white as chess, hybrid intelligence exceeds the sum of its parts.
-
How to reproduce your analysis
I had to become the world's most boring archaeologist, poring over version control changelogs and filesystem backups to understand why my analysis suddenly gave different results.
-
How to run an Idea Stampede to make a group gallop
You'd think everyone will trample over each other when typing, but after some initial confusion and giggling, people make space for one another.
-
What is precision vs recall, and why should I care?
Machine learning algorithms are lazy and opportunistic—faced with imbalanced data, they might just guess 'no' every time, achieve 90% accuracy, and go home early for a beer.
-
How to Cheat at Data Science (with help from Centaurs and the Wizard of Oz)
Wizard-of-oz UI testing and AI-assisted centaur teams—two deceptively simple techniques that help data science teams sidestep the usual trap of building elegant solutions to problems nobody actually has.
-
Why we're bad at estimating (and what to do about it)
People can't reliably estimate how long things will take beyond a few hours—so break solutions into smaller chunks, ask what could go wrong, and make estimates in isolation before discussing together.
-
A little functional programming goes a long way
It's better to have 100 functions operate on one data structure than 10 functions on 10 data structures—a principle that transforms how we think about code modularity.
-
What does it take to transition from Senior to Chief Data Scientist?
As you get more senior, your area of responsibility grows—from implementing well-defined tasks to picking the right projects and ensuring stakeholders actually see their value.
-
What makes a great data scientist and data science team? (Podcast)
The most profound approach to building excellent data science teams isn't about hiring brilliant individuals—it's about creating teams the rest of the business actually wants to support.
-
What is Machine Learning?
Machine learning algorithms face the same challenge you had as a school pupil: there's going to be an exam, and you only have past papers to study from.
-
What is Overfitting?
Solomon Shereshevsky could memorise endless playing cards but struggled to recognise familiar faces—the same trap that snares AI models when they overlearn training data.
-
How to write good software faster (we spend 90% of our time debugging)
If we spend 90% of our time debugging, learning to type twice as fast only speeds up development by a few percent—focus on writing fewer bugs instead.
-
How to know if your recommendations algorithm is actually doing a good job
You can get a pretty good 'average hit rate' score by just recommending the most popular programs to everyone—but that's a failure of the metrics, rather than a success of the algorithm.
-
Necessary vs Sufficient
The Turing Test reveals intelligence's slippery nature—passing it proves you're intelligent, but failing might just mean you're French or think like an alien.
-
How to make your data science team faster (and speed up progress)
That complaint about your team being 'too slow' might not be about throughput at all—listen carefully and you'll often discover it's really about latency, flexibility, or predictability instead.
-
Why are my emails being sent to spam - a guide for GSuite
Like car brakes, we can all agree that email deliverability is important, but we don't want to have to learn how it works.
-
The Difference: Throughput vs Latency
Speeding up a conveyor belt reduces latency; widening it increases throughput—but a wider, slow belt still takes ages for any single widget to traverse.
-
Will data scientists become obsolete in the next 10 years?
Just as cloud servers replaced physical hardware with three clicks, machine learning services promise to commoditise data science—yet the real world stubbornly refuses to cooperate with our Dorothy fantasies.
-
How to find a job through a recruiter
I've watched big companies systematically ignore direct applications whilst favouring candidates forwarded by recruiters—they trust that good recruiters have already done the careful screening work.
-
How to build a model that performs at chance
I shuffled the labels and ran my model on scrambled data—it still performed above chance, revealing that my canary had died and I was peeking.
-
Running a Premortem
We told a story together out loud about a potential worst-case future that hadn't happened yet, picturing the Apple launch event happening in slow motion with nary a mention of our product.
-
Technical debt isn't like financial debt
It's helpful to imagine I've borrowed money from a capricious loan shark, who might show up at your door one day with a baseball bat and demand all the money right now.
-
How to write proposals that lead to good, quick, collective decisions
A good decision today is better than a perfect decision tomorrow—especially when you can reverse it if you're wrong.
-
How to remember names at a party
( - 636 words)
The biggest reason we forget names isn't poor memory—it's that we weren't paying attention in the first place while worrying about spinach in our teeth.
- LTSF 2019 ( - 29 words)
- How to Cheat at Data Science (With Help from Centaurs and the Wizard of Oz) - presentation at Unbound 2019 (VIDEO) ( - 32 words)
-
Fun things to do for free in London (list)
( - 103 words)
From roller skating through royal parks to staging impromptu street dramas, London offers countless free adventures for the creatively curious.
-
Small wins (list)
( - 247 words)
Ten simple actions that can brighten your day, from learning song lyrics to waterproofing your favorite shoes. Small changes that create disproportionate satisfaction in daily life.
-
Friendships and memory effects
( - 408 words)
The longer you've known something, the deeper its roots in your mind—it reassures me to think that friendships work the same way.
-
Soft and hard skills
( - 217 words)
After years focused on management over coding, wrestling with TensorFlow raises a familiar question: was this always this hard, or have I gone soft?
-
Chatbots - from Rogerian psychotherapists to cognitive behavioral therapists
( - 1,623 words)
Building a stateful Slack chatbot in two days taught us that representing conversational logic as data rather than code opens exciting possibilities—despite our demo day disaster.
-
Why has Google open sourced TensorFlow?
( - 519 words)
Google's open sourcing of TensorFlow isn't altruism—it's strategy. The real goal may be "pay by the parameter" cloud computing that disrupts AWS.
-
Entrepreneur vs Manager vs Technician
( - 315 words)
Gerber identifies three business personalities: the Entrepreneur who dreams, the Manager who orders, and the Technician who crafts. Which one dominates your work life?
-
Todo Zero
( - 881 words)
What if you finished each day with nothing left on your todo list? Like a juggler with six balls but only two in hand, focus on just one or two tasks at a time.
-
Presentations at BarCamp Tampa 2015
( - 27 words)
Two presentations from BarCamp Tampa 2015: mastering Python unit testing for better code, and killing your crusty old PHP system through seamless replatforming.
-
Mindfulness in scuba diving
( - 122 words)
Beneath the surface, diving transforms into a meditation where breath controls buoyancy and the alien underwater world demands complete presence in the moment.
-
Listing things to look forward to
( - 157 words)
Making a list of things to anticipate felt like mental physiotherapy - unfurling a cramped limb that had been injured, atrophied but finally free.
-
"Oh, that should be easy - maybe a few minutes..."
( - 707 words)
We humans are walking sacks of blood, bile and bias, and estimating how long things will take brings out the worst in us.
-
Sanity checks as data sidekicks
( - 286 words)
When complex data analysis fails, test something dead simple first—like distinguishing a blank screen from visual stimuli in brain data.
-
Two-level tagging
( - 262 words)
What if you could combine the flexibility of tags with the organization of folders? Two-level tagging lets items belong to multiple categories and creates hierarchies between those categories.
-
Blogging with Wordpress and Emacs
( - 192 words)
A simple workflow combining Emacs, Python, and WordPress lets you write posts in your favorite editor and publish with a single command:
M-x wordpress-publish-this-file. - Final dissertation ( - 16 words)
-
Pluviocabulary (rain words), Memrise-style
( - 33 words)
From petrichor (the smell of first rain) to tirl (rain's sound on rooftops), exploring the rich vocabulary of precipitation reveals language as nuanced as weather itself.
-
The iPhone apps my cold, dead hands would cling most rigidly to
( - 179 words)
From Instapaper for reading queues to Snaptell's book-scanning wizardry, these essential iPhone apps prove indispensable for modern digital life.
-
Startup thinking - the people who have most influenced my thinking on startups.
( - 147 words)
From Steve Blank's lean methodology to Paul Graham's 'release early, iterate often' philosophy—exploring the foundational thinkers who shaped modern startup wisdom.
-
DaisyDisk - where has all my disk space gone? (Mac app)
( - 70 words)
I have a theory that inability to manage one's finances and one's hard disk space have the same psychological factors at their root.
-
Get your users to tell you what they really want
( - 405 words)
A simple feedback box at the end of learning sessions generated 5 times more user suggestions than a prominent red UserVoice button that users had learned to ignore.
-
Storyville and Jenga iPhone apps
( - 90 words)
Two friends have released iPhone apps just in time for Christmas: a fiendishly realistic Jenga with beautiful physics, and Storyville delivering weekly stories to your pocket.
-
Setting up your own domain name and website
( - 485 words)
Setting up your own domain and website doesn't have to be complicated—there are three main approaches, from DIY hosting to Google Apps to all-in-one platforms like Weebly.
-
Breaking the seal
( - 135 words)
After years of procrastination, a simple context shift finally unlocked the floodgates: "Writing's like peeing - once you break the seal, the words just spill forth all evening."
-
A letter to a prospective grad student
( - 600 words)
A PhD can be wonderful when you're excited about it, but don't expect the idiosyncrasies of my experience to apply closely to anyone else.
-
When I am famous, I will decline interviews
( - 334 words)
Famous actors' glossy interviews fill the author with existential dread—conversations stripped of context, distilled by strangers into caricature.
-
But where does the wisdom come from?
( - 53 words)
With courage and serenity within reach, the most elusive element of the Serenity Prayer remains: where does the wisdom come from?
-
The muses are deaf, so speak up
( - 243 words)
Want to unlock better thinking while walking? Talk out loud—loudly, proudly, with gestures—because the muses are deaf and your inner monologue isn't entertaining enough.
-
A wiki for spaces. A town anyone can edit. School architecture founded on mnemonic principles
( - 373 words)
What if the collaborative editing power of Wikipedia could be applied to spaces instead of text—from virtual towns to schools designed as living timelines?
-
How can your iPhone make you even more entertaining and interesting than you already are?
( - 376 words)
What if your iPhone could listen to conversations and automatically surface answers to disputed facts in real-time, keeping dialogue flowing instead of grinding to a halt?
-
Master Turkers
( - 401 words)
What if Amazon's Mechanical Turk could identify Master Turkers capable of higher-skilled work like research, programming, or creative tasks—transforming micro-work into meaningful portfolio building?
-
Brain orchestras and fMRI analyses
( - 623 words)
Think of the brain as an orchestra with instruments playing simultaneously—Multi-variate Pattern Analysis helps decode whether you're hearing Mozart or Beethoven from the neural symphony.
-
Auto-links
( - 342 words)
What if wikis could automatically create links as you type, without CamelCase or brackets? Auto-links detect existing page names and transform them into links instantly, creating lazy serendipity.
-
Eroding our minds
( - 255 words)
Advertising works by exploiting cognitive fluency—the more familiar something feels, the more we like it. Like the banks of a river worn smooth by ceaseless flow, advertising erodes our minds.
-
I refuse to consider an advertising-based business model
( - 91 words)
The idea of monetizing people's attention makes me feel the same way I'd feel about burning books to stay warm.
- Nietzsche ( - 22 words)
-
Communal interactive jukebox
( - 190 words)
What if wireless technology could transform any stereo into a communal interactive jukebox where party guests queue songs and vote on tracks, building reputation like social media karma?
-
House MD, gibberish and clown school
( - 95 words)
How can House MD remain compelling when a third of it is medical gibberish? A friend's clown school experience reveals the surprising power of pure nonsense.
-
What's blowin' in the wind?
( - 197 words)
What if we could identify musical early adopters through data, then track what they're listening to now to predict next year's hits?
-
Dropbocumentation
( - 500 words)
Every time a programmer goes away, infrastructure they know best breaks—that's Murphy's Algorithm. How do you optimize wiki 'edit' conversion rates to capture knowledge while it's fresh?
- Excretation over ( - 24 words)
-
A hivemind with a sense of humor
( - 116 words)
Unlike the Bible's collective wisdom, Google's search autocomplete reveals a hivemind with humor—proving that our digital consciousness can both giggle and think.
-
My two favorite music videos of all time
( - 54 words)
A manic messiah and his merry band versus a bullfrog with a bullhorn - two wildly different artists showcase what makes live performance truly transcendent.
-
The pain of forgetting
( - 600 words)
A memory researcher's personal struggle with forgetting led him from obsessive note-taking to building sophisticated software systems, and ultimately to co-founding Memrise.
-
How to beat an fMRI lie detector
( - 1,211 words)
fMRI lie detectors might seem foolproof, but they rely on baseline comparisons just like polygraphs—meaning simple mental tricks during control questions could disrupt their accuracy.
- Losing weight ( - 27 words)
-
Our one true inalienable right
( - 90 words)
A death of our own choosing, at our own time and on our own terms, may be the most noble way to end one's life.
-
Why is it that chocolate and whisky help when I want a cigarette?
( - 58 words)
What does it mean that chocolate or whisky can sometimes quench that belly-fire-for-fire when craving cigarettes? Are these just interchangeable sources of endorphins drowning out absent pleasures?
-
Word pools for psychology experiments
( - 150 words)
Need carefully curated word lists for your psychology research? From imageable nouns to frequency-filtered datasets, these specialized word pools offer researchers precisely controlled stimuli for memory and cognition experiments.
-
I don't believe in luck
( - 215 words)
Some people are luckier than others, but this is a function of attitude rather than fortuity—Lady Luck wants to be serenaded, not strip-mined.
-
On patents as a business tool
( - 194 words)
Patents aren't just legal protection—they're deterrents that make large companies more likely to partner with or acquire you rather than compete directly.
-
Self Control through software
( - 139 words)
When asked whether he'd rather be smarter or have more willpower, the choice was clear—but can software like Self Control actually help break our addiction to digital distractions?
-
Mobile phones should be felt but not heard
( - 91 words)
What if your phone could tap you on the wrist instead of screaming from the bottom of your bag?
-
Sssssssshhhhhhhhhhh... for now (OBSOLETE)
( - 143 words)
What if phones could automatically silence themselves for an hour, then restore the ringer without relying on our notoriously unreliable prospective memory?
-
Emacs backup files
( - 158 words)
Emacs creates billions of little ugly-twin backup files scattered across your hard disk—here's how to tame them with better configuration options.
-
Google, here's some context
( - 179 words)
What if Google let you provide context rather than exact keywords? A simple
context:(python web database programming)could revolutionize how we search for related topics. -
Picking scientific reviewers
( - 415 words)
A good reviewer's recommendations will be predictive of whether a paper went on to be published—imagine if every reviewer had a public scorecard tracking their efficacy.
-
iThimble
( - 94 words)
What if a simple thimble could transform your clumsy smartphone interactions into precise, pen-like control over those tiny touchscreen buttons?
-
Assortative mixing for peer review
( - 282 words)
Being a good reviewer is like wetting yourself in a dark suit—you get a warm feeling, but no one ever notices.
-
The wallet washkit for the person about town
( - 107 words)
No thicker than an ordinary wallet, but packed with NASA-designed micro-toothbrush, potent chewing gum, and roll-on deodorant gel for the modern gentleman's nocturnal adventures.
-
Open source the Drobo data format
( - 334 words)
Open sourcing Drobo's proprietary data format could spark innovation, reduce customer lock-in fears, and paradoxically make more people willing to trust their data to the platform.
-
Mobile phone ringing
( - 271 words)
What if phones could automatically silence themselves by detecting nearby devices? A simple algorithm might solve cinema embarrassment, but predictable dumbness often beats surprising smartness.
-
Dice Man iPhone app
( - 42 words)
An iPhone app that lets dice decide your fate—from calling your ex at 5am to throwing your phone in the nearest river.
-
Spacetime alarms
( - 255 words)
What if alarm clocks could wake you based on location instead of time? From train naps to meeting reminders, spatial alarms could revolutionize how we navigate daily life.
- Did you know that the word 'gullible' isn't in the dictionary? ( - 28 words)
-
Alleviating tinnitus, and the shape of the auditory phenomenological landscape
( - 255 words)
Could psychological interference from carefully tuned external sounds disrupt the brain's perception of tinnitus, creating a kind of neural "noise cancellation" effect?
-
Emacs Freex gets a new home
( - 100 words)
Emacs Freex mode transforms note-taking into a massively-hyperlinked database of ideas—a personal wiki on steroids that's now found its new home on GitHub.
-
If I were to write a novel
( - 146 words)
What if you could transform London's subway platforms into the pages of your novel, turning daily commuters into captivated readers progressing billboard by billboard?
-
Adam Detre's recommendations on music in late 2008
( - 380 words)
A charming, typo-filled email from 2008 captures one brother's passionate music discoveries, from Fleet Foxes' Pet Sounds-inspired harmonies to Crystal Castles' "lovely Blip Blop noises."
-
We only use 10% of our brains?
( - 306 words)
The myth that we only use 10% of our brains is like saying we only use 10% of a piano—it's the pattern that matters, not the percentage.
-
PyEPL blog
( - 116 words)
PyEPL is a versatile Python library for coding psychology experiments, but it needs more tutorials and community resources to reach its full potential.
-
WriteRoom? DarkRoom? MushRoom!
( - 329 words)
Why build yet another "distraction-free" text editor when Emacs already has undo/redo and full-screen capabilities? Here's a simple function that outperforms WriteRoom's expensive alternatives.
-
Switching from Linux to Mac
( - 579 words)
A longtime Linux user discovers the joys and frustrations of switching to macOS, finding wonderful hardware but some fundamentally broken behaviors that may never be fixed.
-
Losing an old friend - a paean to a laptop
( - 571 words)
A dying ThinkPad forces a choice between the MacBook Air's sleek reliability and the Lenovo X300's Linux freedom—beauty versus ideology in laptop form.
-
Email is the mind-killer
( - 61 words)
A Dune-inspired meditation on conquering email overwhelm transforms Frank Herbert's iconic litany against fear into a mantra for digital mindfulness.
-
Running a psychology experiment
( - 1,769 words)
The most important advice I can give is to run yourself in your own experiment at least a couple of times before you run anyone else.
-
Evangelizing Emacs (in terms of economics)
( - 397 words)
Emacs has a huge barrier to entry, but its keyboard shortcuts and customizations provide enormous economies of scale once mastered.
-
Sensing the buzz at a conference
( - 276 words)
What if conference badges tracked attendee movements to recommend sessions based on where like-minded people gather? A simple RFID system could transform how we navigate overwhelming conference schedules.
-
Create something shareable
( - 393 words)
Create something shareable - add to the sum total of human knowledge by building something useful, trustworthy, and worth the effort for others to discover rather than recreate themselves.
-
Matlab cookbook
( - 60 words)
A collection of Matlab tips and scripts inspired by O'Reilly's Python Cookbook, offering practical solutions for common programming challenges.
-
Emacs freex 0.3.1
( - 47 words)
The first public release of Emacs freex minor mode is now available, complete with starter tutorial and installation instructions.
-
Using fancy methods to estimate shipping dates (Joel Spolsky's FogBugz)
( - 613 words)
Joel Spolsky's FogBugz uses non-parametric statistics to predict shipping dates by analyzing developers' historical estimation errors and generating thousands of synthetic futures.
-
Scroogled
( - 75 words)
Cory Doctorow's dystopian fiction explores 'What if Google were evil?' — a compelling thought experiment about privacy erosion and surveillance power.
-
Entangling the ground and cloud
( - 1,088 words)
What if your local filesystem and the cloud were quantum entangled, automatically syncing every file edit between your laptop and web services without you ever noticing?
-
My take on emacsclient
( - 322 words)
Emacsclient promises instant file opening but fails basic usability tests. A Python script fixes the frustrating gaps between emacsclient's potential and reality.
-
Testing the legs of the Sansa Express
( - 296 words)
The Sansa Express died after being yanked from Ubuntu without safe removal, despite working fine on Windows XP with the same treatment.
-
Collaborative filtering and how it's going to help us consume
( - 1,521 words)
Amazon's recommendation engine is just the beginning—imagine a system that knows your tastes so well it could curate your entire life, from books to concerts to travel destinations.
-
Make tags not trees - filesystem idea based on tags instead of hierarchical directories
( - 1,183 words)
What if we replaced hierarchical directories with a flat filesystem organized by tags, where files could exist in multiple categories simultaneously without the constraints of choosing a single location?
-
It has to be easy, and worth it, for you to add tags
( - 1,340 words)
Tagging lets you assign things to multiple categories, reducing cognitive cost—but only if it's effortless to add tags and reliable for finding things later.
-
Neuroscience notes (OBSOLETE)
( - 272 words)
PhD qualifying exam notes transformed into a wiki-style neuroscience resource with hyperlinked brain regions, tracts, and disorders—though some pages remain sparse in this gargantuan task.
-
John Cunningham Lily, RIP
( - 75 words)
A tribute to consciousness researcher John C. Lilly inspires a humorous reflection on alternative paths to exploring the nature of human awareness.
-
The Pittsburgh EBC competition
( - 456 words)
Researchers used fMRI brain scans of people watching Home Improvement to predict moment-by-moment experiences like amusement and arousal, placing second in a groundbreaking neuroscience competition.
- Push Singh ( - 28 words)
-
Statler pulchrifex
( - 22 words)
A compelling fiction piece by Matt Weber graces the back page of Nature, blending scientific imagination with literary craft in an unexpected venue.
- Lawrence Lessig ( - 5 words)
-
Does eating salmon lower the murder rate?
( - 62 words)
A British prison study found that fatty acid supplements reduced antisocial behavior by over a third, raising intriguing questions about nutrition's role in violence.
-
The world is your canvas
( - 51 words)
What if massive alien flying saucers could be painted on snowy hillsides, visible only to passing drivers? Street artists are transforming everyday surfaces into mind-bending optical illusions.
-
The Turing tournament - a proposal for a reformulation of the Turing Test
( - 5,910 words)
What if we replaced the binary pass/fail Turing Test with a competitive tournament where players devise rules for each other, creating a richer, multi-dimensional measure of intelligence?
-
Paul Graham, Joel Spolsky and Steve Yegge and the Law of Increasing Returns
( - 380 words)
Three influential tech writers share a crucial insight: the Law of Increasing Returns means truly smart people, powerful tools, and quality environments deliver exponentially greater value than their mediocre counterparts.
-
Procrastination
( - 648 words)
Reading about avoiding procrastination is among the best ways to procrastinate—but what if you could harness this tendency as a force for good?
-
Google Desktop Search
( - 170 words)
Google's desktop search consistently fails to find existing files and struggles with PDFs, while browsers lack simple options to download everything for offline access.
-
Google as my permanent online user ID?
( - 63 words)
What if Google became your universal online identity, automatically logging you into every site and syncing your digital life across any device, anywhere?
-
Things that are still crap about Linux (in 2005)
( - 113 words)
From dependency hell to wireless woes, a candid look at Linux's persistent frustrations that keep users pulling their hair out.
-
Hard disk acceleromotors
( - 34 words)
Why don't all hard drives sense their own motion and protect themselves from impact—or have we been catastrophically overestimating the fragility of spinning platters?
-
Philosophy as debugging
( - 66 words)
Philosophical argument resembles debugging code: you modularize problems, test cases, and hunt for the source of error—often finding wrongful initialization or overlooked implications.
-
For a moment I thought my wisdom was spreading by word of mouth
( - 26 words)
The mixed blessing of academic citation: discovering your work is being referenced, only to realize the authors haven't actually read your papers.
-
Cringely on the RIAA
( - 75 words)
Cringely proposed an intriguing legal workaround to challenge the RIAA's grip on music distribution, but nobody ever tried it—perhaps because it seemed too easy.
- Electronic voting ( - 38 words)
-
Michael Behe, 'Darwin's Black Box'
( - 1,219 words)
Michael Behe argues that complex biochemical systems like blood clotting are irreducibly complex—requiring all components to function, making gradual Darwinian evolution seemingly impossible.
-
Fahrenheit 9-11
( - 200 words)
Moore's Fahrenheit 9/11 could swing the election, but raises questions about whether subjective journalism succeeds where objective reporting has failed to expose uncomfortable truths.
-
Google goes public
( - 128 words)
Beyond reasoned analysis lies faith in Google's potential—a company whose motto includes "you can make money without doing evil" and "great just isn't good enough."
- Cool interview with Linus Torvalds ( - 16 words)
-
Reading: 1984 + 20 = 2004. coincidence?
( - 121 words)
A chilling examination of America's contingency plans for postponing elections, raising urgent questions about constitutional safeguards and the balance of power in crisis situations.
-
Direct democracy
( - 1,165 words)
Switzerland holds referenda multiple times yearly on everything from village fair bunting to adopting the Euro—could this direct democracy model transform how citizens engage with government?
-
Nobody seems to want to improve the world
( - 504 words)
Designing a utopia proves surprisingly difficult—people resist radical improvements, fearing that eliminating suffering might destroy what makes us fundamentally human.
- My epitaph ( - 17 words)
- Shameless bid to get a Gmail account ( - 14 words)
-
Journal - Crashed
( - 61 words)
After all, the flipside of a personal God is that, in order to be able to understand us, he must himself be capable of tantrums.
-
Dragonlance gods and the Ancient Greeks
( - 72 words)
Dragonlance gods become far more compelling when viewed as modern fantasy versions of Ancient Greek deities—flawed personifications of cultural archetypes and the morality of the strong.
-
The big lessons of philosophising
( - 149 words)
If you've decided that one group of intelligent people are clearly wrong, then you've probably failed to understand the full complexity of the problem.
-
Definitions - O'Reilly & Munakata
( - 37 words)
What if the secret to symbol manipulation isn't in our algorithms, but in whether our neurons detect or merely activate—a distinction that could redefine intelligence itself?
-
Intrinsicality, symbols, self-organisation and gradual transformation
( - 218 words)
Self-organising symbols can transform gradually unlike rigid code, creating intrinsic environmental representations. Is it coincidental that both neural networks and genotypes speak the universal language of vectors?
-
The Cog team
( - 68 words)
Three fundamental errors plague classical AI: assuming monolithic internal models, unified control, and general-purpose processing—all based on flawed introspection rather than modern cognitive science.
-
Attractors
( - 125 words)
Lloyd Rice warns against the devaluation of the term "attractor" when applied loosely to pattern recognizers, emphasizing its proper meaning as a sequence through states.
-
Consciousness quote from Kinsbourne
( - 19 words)
We are consciously aware of thoughts according to their stability, persistence, and level of influence, according to research from Kinsbourne and colleagues.
-
Jimmy Carr
( - 10 words)
Comedian Jimmy Carr poses a provocative theological question that challenges traditional Christian beliefs about divine hierarchy and special status.
-
Love is an illusion
( - 85 words)
Love isn't transcendent—it's biological. Humanity clings to romantic notions that separate us from animals, fearing what remains if we abandon these illusions.
-
Atheists and foxholes
( - 105 words)
Religion is a decision of the viscera—if war can break atheist resistance, perhaps desperate need can will God into psychological existence.
-
Malonnaise
( - 181 words)
A poetic meditation on navigating emotional storms, where the author embraces being both ship and captain in life's turbulent waters.
- Mathematician's refrain ( - 11 words)
-
Ralph Waldo Emerson
( - 78 words)
Emerson's timeless definition of success transcends wealth and fame, focusing instead on laughter, respect, beauty, and making even one life breathe easier.
-
Why and the religious sense
( - 565 words)
The religious sense stems from a fundamental misconception—our pattern-seeking minds eventually trick us into asking unanswerable questions like 'Why are we here?' instead of the safer 'How?'
-
The wrong end of a telescope
( - 35 words)
Have you ever looked around and seen that all the people are just hairless primates, monkeys with syntax, equally pointless, equally unwitting, and equally comic?
-
Happiness
( - 25 words)
Two profound perspectives on happiness: Dennett advocates finding purpose beyond yourself, while Kennedy emphasizes pursuing excellence through your natural abilities.
- Marriage ( - 8 words)
-
File-sharing
( - 79 words)
File-sharing is stealing, plain and simple — masked by anonymity, mass participation, and spurious justifications that ignore the industry's survival needs.
- Hot tips from the Viennese recliner ( - 21 words)
-
Memories slipping through your fingers like sand...
( - 36 words)
You don't remember life as a continuous stream—you remember episodes. The secret to slowing time's passage lies in creating salient experiences that serve as vivid bookmarks of each era.
-
Mission statement
( - 42 words)
Instead of saying I want to make machines think, I'll say I want to make people smarter—they're the same path to the Singularity.
-
Semantic network hyper-reference of the brain
( - 97 words)
Why isn't there a massive, intricately hyperlinked semantic network brain encyclopedia with neuroanatomical organization, functional links, and constantly updated cross-references between brain areas?
- Snow Crash ( - 4 words)
-
No more ringtones
( - 51 words)
Why don't we have coin-sized vibrating accessories that attach to watches, necklaces, or earrings to alert us when our phone rings?
-
Trophic levels, trade and the Terminator
( - 251 words)
Why would superintelligent machines want to dominate humans when they'll occupy completely different ecological niches? The key may lie in finding something uniquely human to trade.
-
Hans Moravec on our future generations as robots
( - 118 words)
Hans Moravec argues that robots will be our true descendants, shaped by human consciousness rather than unconscious biochemistry—"it is the tune that matters, not the platter."
- Gradual transformability ( - 18 words)
- Lyrics from Dead Milkmen ( - 20 words)
-
Douglas Adams' favourite pieces of information
( - 147 words)
Douglas Adams reveals his favourite absurd facts: sloths grabbing their own limbs and falling from trees, and a ruler's nick creating a kink in the trans-Siberian railway.
-
McEwan
( - 21 words)
McEwan's Rule of Relative Importance offers a vivid metaphor for survival in group dynamics: **"When traveling with a herd of elephants, don't be the first to lie down and rest."
-
Live each day...
( - 37 words)
A humorous take on the classic advice to live each day as if it's your last — with the logical conclusion that laundry becomes completely unnecessary.
- Rolls on representations ( - 9 words)
-
Steve Larson on Minsky's model 6 framework
( - 140 words)
Minsky's six-level model of mind progresses from instinctive reflexes to self-conscious thinking that incorporates others' perceived opinions into our decision-making processes.
-
Bruce Blumberg on curiosity
( - 27 words)
Is curiosity a single adaptive specialization or a collection of mechanisms working together? Bruce Blumberg poses this fundamental question about the nature of human curiosity.