The "separate compartments" model doesn't survive contact with the work
Lee's father used to argue that work and home life belong in watertight compartments: nothing from the office comes home, nothing personal goes to work. The obvious part of that advice — don't neglect your family for overtime — doesn't need a metaphor. What the metaphor implied was stranger: that a developer should keep some kind of split personality, not being a developer at home and not pouring too much of himself into the work. There is a grain of truth in it, since even a beloved job can't love you back, but what he describes being hooked on isn't a single job — it's the power of code and language.
The compartment model fits poorly with web development because the discipline is so multidisciplinary that it resists being boxed up and left at the office. Usability comments from someone you love can surface later as a competitor to a company you join, like a plant paid off much later in the narrative. And existence is online enough that you can't unsee the matrix you helped build well after 5 p.m.; the interface you're constructing demands the skills of a psychologist, an artist and a scientist at once, and the elegant implementation of a complex user flow may only arrive in a dream.
Douglas Crockford's position gives some comfort here: he considers it a miracle that we can program at all, and argues that the absence of any evolutionary basis for it is why we haven't hit the singularity. If we understood how the brain produces software, we could build an AI that programs well enough to improve itself, recursively, until it surpasses us. What we have instead is something like GitHub Copilot — branding that captures a duality rather than a singularity. The hope is that we harmonize with what Noam Chomsky calls a "kind of super-autocomplete," the way autotune used well can flatter a good singer and used badly makes everyone sound like the same robot.
Is the missing off-switch a bug or a feature?
One line of inquiry treats programming aptitude as its own category of intelligence, closer to neurodivergence with its particular strengths and weaknesses, than to anything an IQ test captures. Lee cites a study whose researchers designed a test that appeared to predict which first-year computer science students would learn to program. Their conclusion: aptitude correlates with a "comfort with meaninglessness." Writing a program means accepting that whatever you intend it to mean, the machine will blindly follow its meaningless rules to a meaningless conclusion, and the students who learned showed a pre-acceptance of that fact.
That realisation cuts both ways, as Orwell's Newspeak and Philip K. Dick both warned: whoever controls what words mean controls people, not just machines. Swipe on Tinder for a while and sit with the feelings the phrases "swipe right" and "swipe left" produce — the app's visual language has taught you good and bad. It's the Humpty Dumpty move from Through the Looking-Glass, where words mean what he chooses, delivered in a scornful tone as though Alice should already know it: a form of gaslighting, if a more transparent one than most modern examples. An xkcd strip has Alice turn the same logic back on him and take his possessions. The ability to reshape reality by shifting consensus meaning may be neither good nor bad but simply true, which is perhaps why the person who coined "the map is not the territory" and "the word is not the thing" was an engineer. What we do with it depends on a moral compass — the same instinct that could make someone a surgeon could make them a serial killer.
Boundary-pushers share a motivation with the rest of us
Abuse, in Lee's account of seven years of psychological and physical abuse, reduces to violating boundaries to gain control — but he doesn't consider the person irrational. Pushing boundaries to get what you want is a natural human appetite. Children do it, and so does anyone who makes CSS do what it doesn't want to, which is the premise of his CSS-Tricks articles. The productive move is to channel that impulse toward something illuminating, while accepting that people who harm us are meeting a relatable need in unacceptable ways.
The clearest illustration is a hack that hit CSS-Tricks' former hosting provider over a decade ago. Chris Coyier got a reactivation notice for his domain showing that the account's primary email had been changed to someone else's. After it was resolved, he interviewed the hacker to understand the social engineering behind the attack and the motivation. "Earl Drudge" — an anagram of "drug dealer" — said it was nothing personal; Earl does things for "money and attention," and Chris noted that he does things for money and attention too, which makes them "not entirely different at our core."
Cops and criminals are said to share personality traits, and everyone in technology shares the mindset that lets a person bend a system's assumptions to their will — hence the blackhat/whitehat distinction, two sides of the same coin. Applied to life, the same logic is what we now call a life hack: something like emergent gameplay, a logical if unexpected consequence of what nature does, and a conscious form of human evolution. On a popular website, a surprisingly high percentage of people follow the rules when you explain them properly, a large percentage ignore them from laziness or ignorance rather than malice; hackers and developers want to know how the rules can be used, or what happens when they aren't followed. A seven-year-old entering a wrong answer to trigger an animation is the benign version of that. The blackhat lifehacker version is someone experimenting with your deepest insecurities because exploiting them gets results.
Injection attacks and verbal abuse
Faulkner's line that the past is never dead and not even past fits: with children shared with the abusive ex, that relationship isn't over. You can't always control who sends you input, but recognising that the input may be malicious, and taking control of how it's interpreted, defends against both cross-site scripting and verbal abuse.
The ex would inject the word "stupid" and other names into his consciousness repeatedly. It is just a word — a set of characters with no inherent meaning, like a malicious piece of JavaScript saved into a site. The damage comes from how you let it be interpreted. When the "stupid" script ran in his head, it carried meanings and assumptions the way a high-level keyword stands in for a set of lower-level instructions:
- Intelligence was conflated with self-worth.
- He believed the hurtful words would stop after tearful promises made once she knew they caused pain — as though she hadn't known the first time.
- He felt trapped by the name-calling because he believed the relationship was something he needed.
- He took the input at face value, as if his intelligence were the issue, rather than the power she gained by eliciting the reaction one word produced.
Sanitizing the input
A psychologist pointed out that the ex likely knew he wasn't stupid; the intent was to damage his self-worth until he was easy to control. Acknowledging strengths wouldn't accomplish that. His own brand of intelligence isn't the kind she values — the strengths that make him a capable engineer were invisible to her. Whether she believed what she shouted is ultimately irrelevant, because the purpose was the effect, not the surface meaning. The vulnerability was that he treated her input as a first-class citizen, able to execute with the same privileges as the scripts he had written for himself. Once he sanitized it — through therapy and self-hypnosis — her malicious scripts no longer carried the importance of his own, because she didn't deserve that privilege. Reviewing the untruths now is like reading an inert block of JavaScript: still visible, unable to hijack his self-worth.
Turning Humpty Dumpty's logic around, if words have no inherent meaning, there's no reason he can't reengineer himself so that his meanings outrank the ones the abuser wanted. The sanitized "stupid" script reads:
- I want to hurt you.
- I want to get what I want from you.
- I want to lower your self-worth so you'll believe I'm better than you and won't leave.
Translated that way it has nothing to do with intelligence, and he's secure enough to call himself an idiot jokingly in print. That isn't collusion with his ex's ghost but permission not to be perfect — human fallibility is where our ability to achieve what a computer can't resides. A former manager's response to a bug captures it: "That's good, at least you know you're not a robot." Being an idiot makes what he's done with CSS seem more beautiful, because the work is done around both the limits of the technology and his own; some people won't like it or won't get it, and he has made peace with that. The discipline is the same in either realm: never take needless risk, but stay in your lane assuming malicious input keeps trying to reach you, and spend energy on reducing the attack surface rather than on dwelling on what malicious people shouldn't do.
Interviews, scars and leaps of faith
In a new relationship, his partner remarked that dating him was starting to feel like "a job interview that never ends," because he kept vetting her to avoid being hurt again. The analogy is apt in hiring too: some interview processes map an organisation's scars from the negative forces it once let in inadvertently. The horror trope requiring evil to be invited reflects a truth — we unknowingly open doors to mistreatment. This isn't victim blaming; abusers can only use the power they are given. An interviewer may well ask what you'd do with the authority they're considering handing over and explain, "I ask because we've seen people do [X]." They are thinking of a specific person. Taking the grilling less personally is easier with that in mind, though the first developers who got the company off the ground probably didn't face four interviews and an elaborate React challenge. Organisations, like people, evolve in what they need from a newcomer, and we can't hold that against them. A startup may exist only because of a pile of ill-considered high-risk decisions; his relationship with his kids, more treasured than anything he owns, came out of the worst mistake he ever made. At 30 his driver's licence said one thing while emotionally he was unqualified to decide for his future self — much as reviewing last year's code and wondering what kind of idiot wrote it is a healthy sign.
The point behind the interview analogy is that no amount of age or wisdom removes the leap: letting a new person in is always a leap of faith, on both sides.
Shipping as a managed risk
Launching a site into the wild is another leap of faith, but an air-gapped machine holding the best website ever built, reachable by no human, is worth less than the most primitive contact form that serves a handful of users. His gambling father may have applied his appetite for risk badly, yet calculated risk is necessary. The workable stance is to trust that we can establish boundaries limiting the damage a bad actor can do, instead of pretending risk can be preempted entirely. Surviving hard things or not is a binary. Getting security wrong can be an existential threat to a company, just as compromising psychological safety can be one to a person — which is why "being vulnerable" is a positive phrase. Public-facing sites require holding a paradox: opening yourself to the world while mitigating every risk you can. If front-end coding truly draws on every part of your psyche, let dev change your life, and let your life change how you do dev.




