[{"data":1,"prerenderedAt":572},["ShallowReactive",2],{"nav-categories":3,"category-rambles":9},{"pinned":4,"overflow":8},[5,6,7],"Technology","Security","Rambles",[],[10,374],{"id":11,"title":12,"body":13,"category":7,"date":358,"description":359,"extension":360,"featured":325,"image":361,"imageCredit":362,"imageCreditUrl":363,"meta":364,"navigation":325,"path":366,"public":325,"seo":367,"stem":368,"tags":369,"__hash__":373},"posts\u002Fblog\u002Fthe-producer-that-couldnt.md","The Producer That Couldn't",{"type":14,"value":15,"toc":341},"minimark",[16,49,55,58,64,72,89,94,97,100,119,123,126,132,136,139,142,153,156,161,164,173,176,180,187,190,193,196,200,206,209,212,223,226,234,237,249,252,255,258,262,269,276,280,283,287,290,293,297,300,303,306,309,312,315,326,332,335],[17,18,19,20,27,28,33,34,39,40,44,45,48],"p",{},"A while back, I came upon a few\n",[21,22,26],"a",{"href":23,"rel":24},"https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=jg1WUOxY6Cg",[25],"nofollow","clips"," from\n",[21,29,32],{"href":30,"rel":31},"https:\u002F\u002Fwww.cbsnews.com\u002F60-minutes\u002F",[25],"\"60 Minutes\""," where American record\nproducer Rick Rubin was being interviewed by\n",[21,35,38],{"href":36,"rel":37},"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FAnderson_Cooper",[25],"Anderson Cooper",". In the\ninterview, Rubin self-attests to having ",[41,42,43],"em",{},"no"," technical abilities ",[41,46,47],{},"whatsoever",".\nHe states himself that he knows nothing about music!",[50,51,52],"blockquote",{},[17,53,54],{},"I know what I like and what I don't like. I am decisive about what I like and\nwhat I don't like.",[17,56,57],{},"Cooper, quite reasonably, pressed him on it:",[17,59,60],{},[61,62,63],"strong",{},"So what are you being paid for?",[50,65,66,69],{},[17,67,68],{},"The confidence that I have in my taste and my ability to express what I feel\nhave proven helpful for artists.",[17,70,71],{},"Rick Rubin",[73,74,76],"collapsible-aside",{"title":75},"An aside, if you are unfamiliar with Rick's work",[17,77,78,79,82,83,88],{},"Slayer, Tom Petty, System of a Down, Johnny Cash, Slipknot, Linkin Park, the\nRed Hot Chili Peppers. The range ",[41,80,81],{},"is"," the point. I grew up listening to a good\nnumber of those records, all shaped by the same pair of ears, unbeknownst to me\nat the time of course.\n",[21,84,87],{"href":85,"rel":86},"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FRick_Rubin_production_discography",[25],"The full discography","\nis a rabbit hole worth falling into.",[90,91,93],"h2",{"id":92},"paid-for-taste","Paid For Taste",[17,95,96],{},"So here is a man who, by his own cheerful admission, cannot play the\ninstruments, cannot work the mixing desk and holds not a scrap of music theory,\nand who has nonetheless spent four decades handsomely paid to sit in a room and\npronounce on what is good. Four decades of tasting other people's cooking, in\neffect, at the most rarefied table the industry can lay.",[17,98,99],{},"Watching him work is telling in itself: eyes closed, swaying, letting the melody\nand the rhythm move through him, and somewhere in all of that a verdict quietly\nforms, long before there are any words to hang on it.",[17,101,102,103,108,109,114,115,118],{},"Something measurable is happening in that stillness. The pleasure of music, it\nturns out, lives less in the sound than in the\n",[21,104,107],{"href":105,"rel":106},"https:\u002F\u002Fwww.science.org\u002Fdoi\u002F10.1126\u002Fscience.1231059",[25],"prediction",". Your reward\nsystem quietly wagers on where a piece is headed and pays out when it lands\nsomewhere better than you guessed, and the\n",[21,110,113],{"href":111,"rel":112},"https:\u002F\u002Fwww.nature.com\u002Farticles\u002Fnn.2726",[25],"dopamine even arrives"," in the beat\n",[41,116,117],{},"before"," the peak rather than on it. Rubin's ear is just decades of those\nwagers, settled so many times that the verdict shows up before the words do.",[120,121],"related-post",{"path":105,"title":122},"Interactions Between the Nucleus Accumbens and Auditory Cortices Predict Music Reward Value",[120,124],{"path":111,"title":125},"Anatomically Distinct Dopamine Release During Anticipation and Experience of Peak Emotion to Music",[17,127,128,129,131],{},"Parts of the interview struck a chord with me, and I found myself drawing\nparallels to what I see in the tech industry today. But before we get there, it\nis worth asking a more basic question. What actually ",[41,130,81],{}," taste?",[90,133,135],{"id":134},"sweet-wine-and-other-contrasts","Sweet Wine And Other Contrasts",[17,137,138],{},"Begin with the literal sort. Ask most people what wine they favour and a fair\nnumber may admit to a preference for something sweet. Put that to a sommelier\nand you will naturally get a different response. Dry reds are meant to be the\n\"grown-up\" answer (or so they say), all tannin and structure and hard-won\nappreciation, and sweetness is where one is presumed to have loitered before one\nknew better.",[17,140,141],{},"Except no one nursing the sweet glass is having a lesser evening for it. Their\nwine is every bit as vivid to them as the dry red is to the connoisseur doing\nthe wincing.",[17,143,144,145,152],{},"The wince has a rather human explanation, mind you. Nobody is born liking\ntannin; bitterness only turns pleasant after enough exposure, so a taste for dry\nreds is mostly proof that the training happened. And preferences that take years\nto acquire have a way of becoming badges for the people who acquired them. The\nFrench sociologist Pierre Bourdieu wrote a whole book on exactly this,\n",[21,146,149],{"href":147,"rel":148},"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FDistinction_%28book%29",[25],[41,150,151],{},"Distinction",", taste\noperating as a marker of class. Quoting him here is, I am aware, precisely the\nsort of badge-polishing he was describing, but the line is too good to leave\nout: taste classifies, and it classifies the classifier.",[120,154],{"path":147,"title":155},"Distinction: A Social Critique of the Judgement of Taste",[157,158,160],"h3",{"id":159},"trained-by-contrast","Trained By Contrast",[17,162,163],{},"I believe taste, in the literal sense, is built out of contrast. Savoury means\nalmost nothing until there is sweetness to lean against; salt to sharpen the\ncaramel, acid to cut through the fat. And we are the luckiest generation ever to\nsit down to dinner, with chillies from one continent, citrus from another and\nsome fermented oddity from a third all crowded onto the same supermarket shelf.\nOur palates are schooled by more of that collision than anyone before us, which\nis why \"good taste\" refuses to hold still long enough to be pinned to a wall.\nEach new contrast quietly re-teaches the palate that will judge the next one.",[17,165,166,167,172],{},"Some of it is just hardware. Around a quarter of people are \"super tasters\"\nwhose bitter-taste receptor gene\n",[21,168,171],{"href":169,"rel":170},"https:\u002F\u002Fwww.nature.com\u002Farticles\u002Fs41598-019-54604-1",[25],"TAS2R38"," makes coffee and\ntonic water far harsher than the rest of us feel them, and you will not talk\nthem out of it, so the wine drinker reaching for something sweet might not be\nlazy or untrained at all. They could just be running different hardware. The\nrest of it gets written by exposure: pair a flavour with enough repetition and\nthe machinery that decides whether you like it quietly rewrites itself, and it\nwill rewrite again the moment the bargain changes. Nobody issued you a palate.\nYou have been cultivating one your whole life, every time you sit down to eat.",[17,174,175],{},"And the palate you school at dinner does not have the courtesy to stay at the\ntable. The same faculty, forever appraising each new thing against everything it\nhas already met, turns its attention to whatever you spend your days among.",[90,177,179],{"id":178},"taste-in-software-systems","Taste In Software & Systems",[17,181,182,183,186],{},"For a good many of us, that turns out to be software. Now and then you cross\npaths with someone who simply ",[41,184,185],{},"has it",": taste in design, or its near relation, a\nfeel for standards and process and the unshowy excellence of a system built with\ncare.",[17,188,189],{},"Apple is the obvious example, and obvious is exactly why it works. You can level\nall sorts of criticism at them, and plenty is fair, but pick up one of their\ndevices and you feel it anyway. The weight sits right in the hand. The box opens\nwith that slow, deliberate exhale of air, like it was engineered to make you\nwait half a second longer. All of that is the residue of thousands of small\nmoments where somebody who trusted their own palate looked at a nearly-finished\nthing and said no, not yet. I still cannot bring myself to throw the packaging\naway, which is a slightly embarrassing thing to admit about a cardboard box.",[17,191,192],{},"Software perhaps sometimes admits the same flair. A codebase whose naming is\ncoherent, an API that reads as though a single mind presided over it, an\ninterface with no discordant seams (although granted more mathematical in nature\nin this sense than food, who says maths can't be tasteful ☺️).",[17,194,195],{},"Taste in a system is almost never perceived head-on, any more than you\nconsciously admire the equilibrium of a well-composed dish. You simply register\nthe absence of friction, the merciful sense that nothing is clamouring for your\nattention.",[90,197,199],{"id":198},"the-producers-chair","The Producer's Chair",[17,201,202,203],{},"Chatting with a colleague of mine recently, he phrased something rather\neloquently: ",[41,204,205],{},"\"LLMs can only look backwards, never forwards.\"",[17,207,208],{},"Well, me being myself, I thought \"huh\", that is a rather profound statement...\nNeurological fireworks commence!",[17,210,211],{},"The models are trained on everything already made; they interpolate the past.\nThere is a half truth to this, and the interesting part is the half that's\nmissing. Because Rubin looks backwards too, thousands of records deep, and the\nwhole first half of this essay is backward training of one sort or another. A\npalate is only the food already eaten. So the gaze cannot be the thing that\nseparates us from the machine, since we are all of us looking the same way. The\nhalf it misses is what the looking produces.",[17,213,214,215,218,219,222],{},"Rubin's thousands of records forged one particular pair of ears that had made\ntheir mind up about things, while the model swallowed every palate at once and\ncommitted to none. You could almost say it has too ",[41,216,217],{},"much"," taste, everyone's at\nthe same time, and a taste that excludes nothing has nothing to say, since on\nthis essay's own terms taste is exclusion. He told us as much: decisive about\nwhat he likes ",[41,220,221],{},"and"," what he does not. And the other half is that Rubin's\nbackward look was felt, the verdict forming in the body before the words\narrived, which is where a commitment comes from at all.",[17,224,225],{},"Suno is where this stops being abstract for me. In the right hands it can\nreverberate ideas and expressions into something audible far sooner than the\ncrafted way ever could, and fair play to anyone using it like that. My own\ninterest dies about ten seconds in, because what comes out never quite carries a\ngroove or any reason to be, which is about what you would expect from every\npalate at once.",[17,227,228,229,233],{},"The model has looked at more than he could hear in a thousand lifetimes, and not\none second of it ever landed as liking, no wager placed and none paid out. What\nthey furnish instead is speed, and speed unaccompanied is not the same as taste.\nGiven time, it is your judgement that decides whether the output was worth\nanything. I dwell on this at greater length in\n",[21,230,232],{"href":231},"\u002Fblog\u002Fan-antidote-to-speed","this article",".",[17,235,236],{},"And here is where it landed for me. With \"teams\" of agents now doing the making,\nthe engineer's hands are coming off the instruments. We are shifting, whether we\nasked for it or not, into the \"producer's\" chair. Call it the \"Rubin role\". You\nsit in the room while the band plays, you decide what is good, decisively, and\nyou express it clearly enough that the band can act on it.",[17,238,239,240,244,245,248],{},"Now just imagine, as a fun mental exercise, if real-life Rick were to employ\n",[241,242,243],"code",{},"--dangerously-skip-permissions"," in his processes. With Claude Code, we can pass\nthat flag and let the agent do whatever it likes without ever stopping to ask us\nfirst. Perhaps, in some strange organic way, Rubin already ",[61,246,247],{},"does","?",[17,250,251],{},"Software engineers need good taste. Arguably we always did, but AI has just\npromoted it from a nice-to-have to the differentiating skill, because when any\nfella and his laptop can generate the notes (or 20k lines of code in a day), all\nthat is left to be paid for, really, is knowing which ones to keep.",[17,253,254],{},"A producer does more than pick the good takes though. He tells the band what to\nplay in the first place. These days, I spend most of my time writing specs and\ndocuments that get turned, near enough directly, into code, and I have written\nplenty of bad ones. When a spec is vague about the hard case, the code is vague\nabout it too, and you find out much later than you would like. So the taste has\nnot gone anywhere. It just has to show up earlier now, in the thing you write\nbefore any code exists.",[17,256,257],{},"Which is really only to say that a document is a piece of writing, and writing\nhas been a matter of taste since long before any of us turned up to fret over\nit. A spec can be lucid or muddled, generous to the poor soul who inherits it or\nquietly hostile, in precisely the way a novel can. We seldom think of a design\ndoc as prose, yet prose is all it is, and the very instincts that make a novel\nsing or fall flat are the ones presiding over whether the thing you handed the\nmachine was any good.",[90,259,261],{"id":260},"but-is-your-taste-any-good","But Is Your Taste Any Good?",[17,263,264,265,268],{},"There is, of course, an obvious objection loitering here. How do you know you\n",[41,266,267],{},"possess"," good taste? What even is good taste? The easy answer is that it is\nsubjective the whole way down, with no bedrock to strike. Only that is not quite\nright, and the rest of this piece keeps quietly contradicting it, because in\nsoftware taste turns out to be judgement on a very long feedback loop. The bill\ndoes come, just years late, and it tends to land on whoever inherited the code\nrather than whoever wrote it.",[17,270,271,272,275],{},"Culture and creed aside, a human is a human. A jarring system or product leaves\na sourness in the mouth, and you know the flavour at once: the settings screen\nthat behaves unlike every other screen, the API that reports its errors in three\ncontradictory dialects, the app that contrives to make you feel stupid for\nholding it wrong. And yet there will always be those who take to it regardless,\nand sincerely so. Some people simply prefer the sweet wine, and, as with the\ndrinkers from earlier, they are having ",[41,273,274],{},"their"," experience.",[157,277,279],{"id":278},"hard-to-teach","Hard To Teach",[17,281,282],{},"So taste cannot be graded, and it cannot really be taught either. Rubin did not\ntake a course. He listened to thousands of records until the contrast between\nthem trained his ear, and then, the part most people skip, he became decisive\nabout what he felt. It is the same for us: read a lot of code, use a lot of\nproducts, feel enough jarring seams and enough seamless ones that the difference\nbecomes something you sense long before you can explain it. Nobody hands that to\nyou in a curriculum, which is why so much of it fails to transfer in one. You\nhave to go and get it, one contrast at a time.",[90,284,286],{"id":285},"the-awkward-thing-about-software","The Awkward Thing About Software",[17,288,289],{},"Software is an awkward place to talk of taste, mind you. The thing is\nmathematical underneath. Code compiles or it throws, the tests turn green or\nred, and beneath it all lies a hard floor of correctness that no amount of\nelegance will excuse you from. Set against something so binary, taste can look\nlike ornament brushed on after the fact.",[17,291,292],{},"Except anyone who has lived in a codebase long enough knows the poetry is real.\nI have deleted a working function and rewritten it, changing nothing a test\ncould see, purely because the first version made me tired to look at. Two\nsolutions pass every check and one of them still makes you exhale while the\nother makes you brace for the day you have to change it. The good one names\nthings well enough that you stop carrying them around in your head, and it\nleaves a shape the next edit drops into. A green test suite has nothing to say\nabout any of that, which is how it gets waved off as fussiness.",[157,294,296],{"id":295},"the-taste-of-another","The Taste Of Another",[17,298,299],{},"The trouble is that \"it works\" is only true on the day you measured it, and code\nhas to keep being true for years afterwards, usually in the hands of someone who\nwas not in the room when you wrote it. I learned this the hard way at Amazon,\nand not from my own code. Most of the pain came from systems I inherited, things\nbuilt by people long gone, for which I carried the pager without ever having the\ncapacity to change them. You get woken at three in the morning by a decision\nsomeone made years ago, in a service you barely understand, and there is no\nfixing it tonight. You are just holding it upright until sunrise. That is the\npart nobody warns you about.",[17,301,302],{},"That is where the value of this subject actually hides. It is the outage that\nnever happened because the failure mode was sitting right there where nobody\ncould miss it, the page that never came. Nobody thanks you for it, and that is\nrather the point, the same way nobody compliments a dish for not being too\nsalty, or the way I still cannot bring myself to bin the box the phone came in.",[17,304,305],{},"And the absence compounds in a way a single bad decision never hints at. Each\nshortcut works, in the narrow sense the tests care about. Stack a few thousand\nof them and you get a system nobody dares touch, where a one-line change takes a\nweek because it could be anywhere and the code refuses to tell you. That is what\ntaste quietly prevents.",[17,307,308],{},"Of course, in reality, many of those prior decisions were never about taste at\nall. They were made under pressure, against time constraints that were often\narbitrary. You could argue that undoes the whole premise. Or you could argue it\nis the better case for the machines: let them absorb the pressure and weed out\nthe rushed decisions earlier, so we can \"Rubin\" our systems up to a standard we\nactually chose.",[17,310,311],{},"So no, taste is not decoration. It is most of the reason software you shipped a\nyear ago is still standing, and still bearable to work on, for whoever inherited\nit. It is a hard thing to teach, yet one must learn. Experience is the learning,\nand education, more often than we would like to admit, just gets in the way. And\nif you don't fancy good wine or music, then at least have good taste in your own\nwork.",[17,313,314],{},"But hey, that's just my opinion after all.",[316,317],"iframe",{"src":318,"width":319,"height":320,"style":321,"frameBorder":322,"className":323,"allowFullScreen":325},"https:\u002F\u002Fgiphy.com\u002Fembed\u002FfqtyYcXoDV0X6ss8Mf","100%",365,"","0",[324],"giphy-embed",true,[17,327,328],{},[21,329,331],{"href":330},"https:\u002F\u002Fgiphy.com\u002Fgifs\u002Fmoodman-fqtyYcXoDV0X6ss8Mf","via GIPHY",[17,333,334],{},"P.S. pun was intended.",[336,337,338],"ai-disclaimer",{},[17,339,340],{},"Since it feels only fair given the subject: I used AI\u002FLLMs to help research the studies linked above and to review drafts for clarity. Every thought, argument and narrative element here is my own. The taste, for better or worse, is human.",{"title":321,"searchDepth":342,"depth":342,"links":343},5,[344,346,350,351,352,355],{"id":92,"depth":345,"text":93},2,{"id":134,"depth":345,"text":135,"children":347},[348],{"id":159,"depth":349,"text":160},3,{"id":178,"depth":345,"text":179},{"id":198,"depth":345,"text":199},{"id":260,"depth":345,"text":261,"children":353},[354],{"id":278,"depth":349,"text":279},{"id":285,"depth":345,"text":286,"children":356},[357],{"id":295,"depth":349,"text":296},"2026-07-12","Nobody can teach you good taste, yet somehow you are expected to have it. A look at where it actually comes from, exploring the parallels in taste across audio, food and software, for anyone who makes human things.","md","\u002Fimages\u002Fblog\u002F2026\u002Fwater-droplet.jpg","Natalia Gutnik","https:\u002F\u002Fwww.instagram.com\u002F__dust__in__the__wind__?igsh=MTEwd28zZ2Nta3IwNQ==",{"edited":365},"2026-07-23","\u002Fblog\u002Fthe-producer-that-couldnt",{"title":12,"description":359},"blog\u002Fthe-producer-that-couldnt",[370,371,372],"ai","learning","reflection","CBUPcby-iYvkeEGXu1NGFyzUJzQmnQGtJ1OOKpuJf2E",{"id":375,"title":376,"body":377,"category":7,"date":563,"description":564,"extension":360,"featured":325,"image":565,"imageCredit":362,"imageCreditUrl":363,"meta":566,"navigation":325,"path":231,"public":325,"seo":567,"stem":568,"tags":569,"__hash__":571},"posts\u002Fblog\u002Fan-antidote-to-speed.md","An Antidote To Speed",{"type":14,"value":378,"toc":553},[379,382,385,388,391,394,397,400,404,407,410,413,417,424,427,430,433,438,442,445,448,451,455,458,461,464,467,470,475,478,482,485,488,496,503,506,509,513,516,519,522,525,528,531,535,544,547,550],[17,380,381],{},"There is a particular seduction to speed. With tools like Claude Code, you can\nmove from idea to working implementation in minutes: scaffolding an API, wiring\nup a frontend, resolving a class of bugs that would have taken hours before. It\nfeels like flying. The feedback loop tightens, the backlog shrinks, and you find\nyourself shipping things you would have deferred indefinitely only a year or two\nago.",[17,383,384],{},"(Trust me on that 😉 I am something of an expert procrastinator)",[17,386,387],{},"As of late, I have been fortunate to contribute in environments where there is a\ngeneral top-down appetite for the adoption of frontier tooling.",[17,389,390],{},"The problems are varied, sometimes interesting, sometimes rote mental skirmishes\nbut often come with real scale and opportunity to optimise how one works day to\nday. Not everyone gets that, and I do not take it for granted.",[17,392,393],{},"However, speed has a shadow. The faster you move through a problem, the thinner\nyour understanding of it can become. You can find yourself holding a codebase\ntogether with solutions you didn't fully author (more often than not these\ndays), patterns you haven't internalised, and abstractions you haven't\nstress-tested mentally.",[17,395,396],{},"For me, a certain dissonance occurs when you tell your digital co-worker that\nthey are an industry expert with 450 years of experience and voilà... a codebase\nappears.",[17,398,399],{},"The code works. How much is understood? Well, that is the question.",[90,401,403],{"id":402},"the-problem-with-always-moving-fast","The Problem With Always Moving Fast",[17,405,406],{},"When a tool produces a working result almost immediately, the temptation is to\naccept it and move on. Why dwell on the implementation when the tests are green\nand the feature is deployed? The incentive to slow down disappears because there\nis no visible cost to going fast.",[17,408,409],{},"The cost shows up later. It arrives when you need to extend something and\nrealise you do not actually understand the model it was built on. It surfaces\nwhen a subtle bug appears in production, and you stare at the code as though\nreading someone else's handwriting, and it becomes visible in a code review\nwhere you cannot explain a decision you nominally made.",[17,411,412],{},"This is not unique to AI-assisted development. Senior engineers have always\ncautioned against copy-pasting Stack Overflow answers without understanding\nthem. What AI tools do is raise the ceiling on how fast you can accumulate\nsurface-level understanding. The gap between what you have shipped and what you\ntruly know can grow quickly and quietly.",[90,414,416],{"id":415},"what-slowing-down-actually-means","What Slowing Down Actually Means",[17,418,419,420,423],{},"Slowing down does not mean going back to writing every line by hand, nor does it\nmean distrusting tools that make you ",[41,421,422],{},"more"," productive. It means being\ndeliberate about when you engage deeply versus when you delegate freely. Breadth\nand depth are not inherently mutually exclusive. To consider otherwise is a\npolarised philosophy.",[17,425,426],{},"There are moments in a working day that are worth protecting from speed. When\nyou encounter a new concept, be it a design pattern, a protocol, or a language\nfeature, resist the urge to generate an example and move on. Spend time with the\nprimary source, read the RFC, work through the algorithm on paper if it helps.\nBuild the toy implementation that no one will ever ship, just to feel how the\nthing actually works.",[17,428,429],{},"There are also moments when the right response to a generated solution is not\nacceptance but interrogation. Ask why the code is structured the way it is.\nRewrite it yourself without looking. Try to break it. These habits feel\ninefficient in any given moment, but they are how understanding compounds.",[17,431,432],{},"Not all ideas are meant to survive, and that is fine. One underrated quality of\nmodern tooling is how quickly it can help you stress-test a concept and discard\nit. The goal is not to preserve every generated solution but to understand why\nyou are keeping the ones you do.",[50,434,435],{},[17,436,437],{},"Einstein called compound interest the eighth wonder of the world. He was\ntalking about money. The same logic applies to understanding. The person who\ninvests a little comprehension every day ends up, quietly and without drama,\nin a completely different place to the one who just keeps moving.",[90,439,441],{"id":440},"the-distinction-between-speed-and-progress","The Distinction Between Speed and Progress",[17,443,444],{},"Speed and progress are not the same thing, even though they feel identical when\nyou are deep in a flow. You can move very fast and accumulate very little\nunderstanding. You can also move slowly and build a mental model that pays\ndividends for years.",[17,446,447],{},"The best engineers learn to read the difference in themselves. There is a\nparticular quality of engagement when you are genuinely learning: a slight\nfriction, a need to pause, a sense of things clicking into place. When this\nfeeling is absent for too long, it is usually a sign that you have been\nexecuting rather than thinking.",[17,449,450],{},"This is not an argument against execution. Execution is how things get built.\nBut most engineers who do this work long enough find that the periods of deep\nlearning, the weeks spent really understanding a distributed systems concept, or\na type system, or a network protocol, are the periods that most change what they\nare capable of building.",[90,452,454],{"id":453},"carving-space-in-a-world-that-rewards-output","Carving Space In A World That Rewards Output",[17,456,457],{},"Practically speaking, how do you protect time for depth when the ambient\npressure is always toward shipping? This is partly a personal discipline and\npartly a structural question about how you organise your work.",[17,459,460],{},"On the personal side, it helps to name what you are doing and be specific about\nit. There is a difference between \"I am going to build this feature\" and \"I am\ngoing to understand this concept well enough to build this class of feature\nconfidently.\" The second goal is harder to measure, which is exactly why it\ntends to get crowded out.",[17,462,463],{},"Some engineers keep a running list of things they have used but not truly\nunderstood: concepts, libraries, patterns, language features. Periodically\nrevisiting that list and picking one thing to go deep on turns a vague sense of\naccumulated debt into something manageable and intentional.",[17,465,466],{},"It also helps to notice when the tools are doing cognitive work you should\nprobably be doing yourself. If you are generating code and not reading it\ncarefully, that is a signal. If you are accepting explanations without\nquestioning them, that is another.",[17,468,469],{},"This doesn't imply line by line examination but at least understanding how any\nsystem operates is crucial for longevity of said system. This applies well\nbeyond the realm of software engineering.",[17,471,472],{},[61,473,474],{},"The tools are most valuable when they accelerate work you already understand,\nnot when they substitute for understanding you have not yet built.",[17,476,477],{},"This is why I believe it is a hard time for the up and coming younger generation\nof engineers and knowledge workers in general. And why wouldn't it be? This\nlandscape is completely uncharted. No human generation alive has been here\nbefore.",[90,479,481],{"id":480},"checkpoints-supply-lines","Checkpoints & Supply Lines",[17,483,484],{},"I have a great fondness for military history, so allow me to indulge in some\nanalogies here.",[17,486,487],{},"Abraham Wald identified a relevant trap during the Second World War. The US Army\nAir Forces analysed bullet holes on returning B-17 Flying Fortresses and B-24\nLiberators and proposed reinforcing the damaged areas. Wald pointed out the\nflaw: those were the planes that made it back. The ones hit where the holes were\nabsent never returned to be counted. Reinforce the parts that look fine, because\nthat is where the fatal hits land.",[17,489,490],{},[491,492],"img",{"alt":493,"src":494,"title":495},"Illustration of hit points","\u002Fimages\u002Fblog\u002Fbomber.png","Martin Grandjean, McGeddon, Cameron Moll — CC BY-SA 4.0",[17,497,498,499,233],{},"A detailed write-up of this counterintuitive thinking is available here\n",[21,500,501],{"href":501,"rel":502},"https:\u002F\u002Fwww.warhistoryonline.com\u002Fwar-articles\u002Fabraham-wald.html",[25],[17,504,505],{},"I like to think that the same bias operates in fast-moving engineering. The\nreturning planes are your shipped features and green tests. Everything looks\nfine.",[17,507,508],{},"The vulnerabilities are in what you cannot see: the understanding you never\nbuilt, the decisions you accepted without interrogating. You do not see that\ndamage until it matters.",[157,510,512],{"id":511},"choking-an-advance","Choking An Advance",[17,514,515],{},"One of the most consistent lessons in military history is that armies which\nadvance too quickly outrun their supply lines. The front moves forward, but the\nlogistics cannot keep pace. The advance that looked like strength becomes a\nliability the moment it is tested. Choking off the supply lines leads to\nultimate failure.",[17,517,518],{},"If a concept has grown so complex due to not taking stock from time to time, the\nnatural part of us may potentially see it as a source of anxiety. The concept is\nstarving.",[17,520,521],{},"The same dynamic applies to how understanding follows velocity in software work.\nYou can push the front line faster than at any point in history, shipping\nfeatures, integrating new libraries, building on unfamiliar abstractions. But if\nyour comprehension of the ground you have covered does not keep up, you have no\nsolid base to fall back on when something breaks.",[17,523,524],{},"The answer is a blend of sane strategy and flexible tactics. Strategically, you\nhave to accept that comprehension must keep pace with output, or the whole\nadvance becomes fragile. Tactically, the mechanism is the checkpoint: a\ndeliberate pause after meaningful work to consolidate understanding before\npushing further. It is not a retreat as much as it is the thing that makes the\nnext advance sustainable.",[17,526,527],{},"In practice, this means treating the completion of a meaningful piece of work as\na trigger for reflection, not just a cue to start the next thing. Feature\nshipped, module built, system integrated: stop and ask what you actually\nunderstand about what you just built. Where are the gaps? What did you accept on\nfaith? What would you struggle to explain or reproduce without the tool's help?",[17,529,530],{},"This kind of checkpoint does not need to be long. Even ten or fifteen minutes\nspent genuinely reviewing what you built and why it is structured the way it is\ncan meaningfully close the gap between execution and understanding.",[90,532,534],{"id":533},"conclusion","Conclusion",[17,536,537,538,543],{},"The speed that AI tools make available is genuinely valuable, and there is\nnothing virtuous about going slow for its own sake. The\n",[21,539,542],{"href":540,"rel":541},"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FLuddite",[25],"Luddites"," didn't really win their battle\nafter all.",[17,545,546],{},"The goal is not to resist these tools but to use them well, which means staying\nhonest about what you understand and what you are merely executing.",[17,548,549],{},"The engineers who will do the most interesting work with these tools are not\nnecessarily the ones who use them most intensively. They are the ones who know\nwhen to put them down, sit with a hard problem, and let understanding catch up\nwith output.",[17,551,552],{},"I may not be the wisest Homo sapien on the go, but in my view, that balance is\nworth protecting in a world that increasingly rewards the illusion of velocity.\nThe willingness to slow down is not a weakness so much as an intentional harness\naround one's cognitive discipline.",{"title":321,"searchDepth":342,"depth":342,"links":554},[555,556,557,558,559,562],{"id":402,"depth":345,"text":403},{"id":415,"depth":345,"text":416},{"id":440,"depth":345,"text":441},{"id":453,"depth":345,"text":454},{"id":480,"depth":345,"text":481,"children":560},[561],{"id":511,"depth":349,"text":512},{"id":533,"depth":345,"text":534},"2026-05-14","The case for occasionally slowing down in a world of AI-accelerated development and why your cognitive supply lines matter as much as your perceived velocity.","\u002Fimages\u002Fblog\u002Fdeer-eating-trees.jpg",{},{"title":376,"description":564},"blog\u002Fan-antidote-to-speed",[370,570,371,372],"productivity","jl6dT2An1v-ZtXferJUKj7OVNrLwwDwfU0a_iGWLKcE",1785083883694]