1. How This Term Blew Up
Over the past two years, the term DRI has shown up in job descriptions and in bosses’ slide decks at an absurdly high frequency.
It isn’t new. Directly Responsible Individual—an old piece of machinery from Apple’s internal project culture: from something as small as a single action item to something as big as the One More Thing on a Keynote stage, everything gets a name attached to it, and that person owns whether it succeeds or fails. It traveled from Silicon Valley into the big tech companies, got written into their OKR templates, and until a few years ago was just one unremarkable entry in the corporate-jargon dictionary alongside “let’s align on the granularity.”
The change started with AI.
In the old days, “wear many hats” and “own it end to end” were HR-speak; you listened and let it wash over you. This year it became a number on a headcount spreadsheet: one position doing the work of three people, because “AI supercharges them.” Before, if you couldn’t finish the work, that was on you; now that excuse no longer holds—the tools are this powerful, so how can you still not finish? And so the memo arrives about “transforming into an AI-native organization,” bundled with the people who got cut and the raise that never did.
For the first time, “cut costs, boost efficiency” actually worked out in the boss’s favor: the cost is one person’s salary, the efficiency is three people’s output, and the gap in the middle is credited to AI.
They copied six of the traits straight into the requirements. I’ll fill in the remaining three.
2. The Six Traits of a Super DRI
There are many circulating versions of the Apple doctrine. The common first three go like this:
First, absolute ownership. A DRI never passes the buck. No matter how many departments are involved or how weird the situation gets, the attitude is “the responsibility stops with me.” That sense of ownership isn’t imposed from outside; it’s an internal instinct: once this lands in my hands, the outcome is mine.
Second, influence without authority. A DRI doesn’t necessarily hold a high rank, but can drive a project sideways across teams. Not by leaning on people with title and power, but through professional judgment, communication, and results-orientation—making people who don’t report to him willing to work with him.
Third, the whole-picture view. A DRI sees the entire thing, not just the segment in his own hands. He knows what customer support will face after this feature ships; he knows whose next quarter this delay will bite. His judgment isn’t “this code is fine,” it’s “will this thing work.”
The remaining items usually go:
Fourth, information transparency. Bad news gets said first, without sugarcoating, without stalling. When the project is about to blow up, the DRI is the first person in the meeting room to say “it’s about to blow up,” not the last to admit it.
Fifth, closing the decision loop. Gather input, then make the call, then carry the consequences. Don’t drag a decision until it becomes “consensus,” don’t use “let’s align once more” to offload risk. In his hands a thing is either decided or dead—there is no third state.
Sixth, retrospective skill. When something goes wrong, fix the system first, then talk about individual blame. Ask “where did the process let this happen,” not “whose fault is it.”
Taken together, these six really do make a good portrait. The problem isn’t the portrait. It’s how the portrait gets used.
3. The Price Tag on This Portrait
Translated into the language of the hiring market, those six items mean:
- Absolute ownership → overtime with no boundary, on call whenever something breaks, take the blame without appeal;
- Influence without authority → design solutions for other departments in meetings, grind for resources yourself because no budget gets approved;
- Whole-picture view → do the product, QA, and ops work along the way too, since it’s all “being accountable for the outcome”;
- Information transparency → carry the bad news yourself, write the briefing materials yourself, and be the one explaining it to the outside world when it blows up;
- Closing the decision loop → once you decide, you must deliver; if you can’t deliver, that’s your problem—even though resources and schedules were never within your authority;
- Retrospective skill → the conclusion of every retro somehow always lands on “individual capability improvement.”
See it now? Every one of these six traits maps to responsibility. Not a single one maps to power or money.
This isn’t just my reading. Go back to Apple: the DRI mechanism never stood alone. Right next to it stood another set of things: high enough compensation, stock, rank, and the right to say “I’m pushing this project back.” Responsibility and reward were issued as a hedge—make you own the outcome, but also give you a share of that outcome, and give you room to say no.
When the term traveled elsewhere, only the first half got copied down. The responsibility is global; the reward stays inside the department that keeps it.
So demanding this portrait from someone earning three to five thousand a month is, in essence, trying to use one position’s money to buy six positions’ worth of accountability. And accountability is precisely the thing that can’t be had for free.
4. What a Salary Actually Buys
A lot of people assume a salary buys working hours: eight hours, billed by the second, and since AI makes one hour produce what used to take three, what a steal.
That’s not how it works.
The first thing a salary buys is judgment. The same bug: a rookie sees an error message, a veteran sees an incident. The same requirement: one person sees a feature, another sees what kind of debt it will become in three months. Judgment can’t be crash-coursed; it’s fed by the pits you’ve fallen into—and back then, nobody paid overtime for those pits either.
The second is experience. The value of experience is that it only matters on the day something breaks; the other three hundred and sixty-four days, the person looks idle. The one you cut is never the person who did the least today—it’s the person who fished the system out of the water at three in the morning last year. It’s just that when he fished it out, nobody wrote it down as a merit.
The third, and the real foundation of those six DRI traits: a sense of belonging. “Once this lands in my hands, the outcome is mine”—that sentence only holds if he feels the thing has something to do with him.
And belonging is the one item on this list that can’t be given by mouth. It’s fed, piece by piece, by contract length, salary level, year-end bonus, options, and whether his name is on the layoff list.
None of these three are written into the JD, but all three are inside the salary.
How much a person invests in his own desk depends on how long he thinks that desk will hold him. He does this calculation more sharply than anyone, and he never needs to be taught it.
5. Why the Three-to-Five-Thousand Person “Doesn’t Care”
I’ve worked as a contractor. I know the feeling.
Same building, same desk, except the full-time employee next to me has a different-colored badge lanyard—the kind that gets a seat at the annual party. Mine didn’t. The night before a system launch, the two people staying behind to watch it were me and him. The next day, the congratulatory email didn’t include me; the incident retro did.
So I learned a few things:
- Don’t cross the line. The extra work—don’t do it. Not because I can’t, but because I won’t. That boundary wasn’t drawn by me; it was drawn by the company—the purchase order says “man-month,” so a man-month does exactly the few lines written in the SOW.
- Don’t leave a name. Make what you build look, as much as possible, like nobody in particular built it. If the promotion isn’t mine and the bonus isn’t mine, then why would the credit line be mine?
- Don’t be the safety net. When the ball drops, watch it drop. Picking it up means unpaid overtime, and if it lands on my foot, that’s on me.
These three, in any manager’s report, get written up as the same disease: lacks ownership, weak execution, attitude problem.
Same behavior, two narratives. Where’s the difference? The difference is who holds the performance review sheet.
So-called “not caring” is not a moral failing; it’s a pricing fact. Three to five thousand buys execution—“point and shoot.” You pay the execution price, you get execution, and the trade is fair. If you want ownership, that lives on a different price list.
And plenty of contractors aren’t lacking in ability. It’s that the money you pay can’t support his sense of responsibility. No full-time salary, no full-time benefits, yet you want full-time output—how is that possible? The uglier side of the ledger: the contract price is one hundred twenty a day, and sixty reaches his hands; the other sixty goes to the staffing firm, which doesn’t write a single line of code. A deal like that is unfair to begin with, and in an unfair deal, people instinctively stop honoring the rules the other side sets. Rules belong to both parties; a rule imposed unilaterally is treated by the other side as, at most, a reference.
There’s one more layer of incentive, harder than attitude: for someone on low pay, the goal is to not screw up, to finish the task, and that’s where it ends. What security hazards or latent risks are buried in the system—not his damn business. This may be a job where his position is gone tomorrow, so spending today’s time and energy maintaining the system means the output ultimately replaces him and costs him his employment. For him personally, there is nothing in it. Why would he pour energy into something with nothing in it?
The reverse holds too: there are people who get paid well and still don’t go all in—but that means what you want was never just money, and that’s a different article. At least at this company, at this desk, within this budget, accountability is a premium add-on purchased on demand; it is not included in the base package.
6. Whose Money Did AI Save
Back to AI.
There’s nothing wrong with using AI to write code; I use it every day. The real problem is that sentence: “one person doing the work of three.”
What AI saves is hours, not responsibility.
The code gets written faster, but when production is on fire, AI doesn’t get paged to put it out. When a requirement is misunderstood, AI doesn’t wake up at midnight replaying “I thought you meant A.” When the architecture is chosen wrong, the person carrying that debt three years from now is still the one who signed the tripartite agreement.
The efficiency part gets cashed in immediately; the incident part gets deferred indefinitely. And the boss’s spreadsheet only contains the former.
There’s something subtler still: after AI chops down the barrier to “can write code,” the things that remain scarce actually got more expensive—judgment, domain knowledge, memory of the historical baggage, the ability to slice a fuzzy thing into deliverable boundaries. None of these come from AI; they come from a person sitting at this desk for many years.
Which produces a very absurd combination: the company uses AI to cut away years of accumulated experience, while the JD demands “five or more years of relevant experience.”
It wants to save the money it takes to grow that experience, and it wants that experience to cover exactly the things AI can’t cover.
This is what worries me most about the “AI-native organization.” The term sounds like the future, but the shape it actually takes is usually this: first cut three experienced people down to one, then require that one person to have all six DRI traits, and preferably to use AI to finish the work of three as well.
Structurally, this is the same company compressing cost and quality at the same time. And the bill for the quality column doesn’t come due until the next outage.
In my last post I wrote one line, and I’ll reuse it here: costs came down, speed went up, and the quality column is empty.
7. So What Should Actually Be Done
I don’t want this article to boil down to “pay people more.” That’s not a solution; that’s a wish.
A few things can be calculated.
First, tally the trust cost. The combination of low pay and high demands is not free. People who don’t cross the line, don’t leave a name, and don’t catch the falling ball—every one of those behaviors carries a price: the bad news that should be said doesn’t get said, until it can no longer be hidden; requirements that shouldn’t be accepted get accepted, because once accepted they aren’t your fault anyway; documentation doesn’t get left behind, and handovers run on luck. Those costs all come back eventually—they just don’t come back labeled “labor cost.” They come back as incidents, as turnover rate, as postmortem reports.
Second, say it out loud. If what you want to buy is execution, then say so: this position does execution, the KPI is on time and on volume, no extra responsibility carried. That’s a clean deal and both sides are comfortable. The worst is talking ownership out loud while settling by piece-rate behind the scenes—paying contractor money while expecting a partner’s heart. Plenty of companies do exactly this right now: hire a contractor, settle day by day, unpaid leave, social insurance and housing fund contributions at the minimum tier; then demand 7×24 on-call and instant replies when something breaks on the weekend. The settlement treats you like a temp; the response requirement treats you like a partner. That’s not a contradiction, it’s pricing: it wants to buy “on call at any moment” while paying “settled by the day.” The gap in the middle gets filled, free of charge, by your sense of responsibility.
Third, if you genuinely want a DRI, pay the DRI price. At least three things: complete ownership of the thing (including the right to say “this isn’t mine”), a share of the outcome win or lose, and a defined boundary of exemption when it fails. None of these three cost much money; what they cost is a little power handed downward. What most bosses can’t afford isn’t the salary—it’s the latter.
Fourth, accept what the market prices. Some positions are simply unsuited to demanding DRI behavior: high churn, short growth paths, low salary ceilings. That’s not discrimination; it’s honesty. For the positions where accountability genuinely fits, you have to buy it up to that price; if you can’t, change the organizational structure instead—for example, outsource by outcome, settle on delivery, and let nobody talk to anybody about passion.
As for the workers, I have two lines for them as well.
Don’t let other people’s vocabulary price you. You should know in your bones what “thrives under pressure” in a JD translates to. When you see words like DRI, AI-native, end-to-end, ask one question first: what’s the money?
And don’t sell yourself cheap in reaction, either. The last section said “three to five thousand buys execution”—that was a description, not a prescription. Where you draw your own line doesn’t have to match what the contract says—provided you know what you’re working extra for, and that the thing you’re building can be carried away with you.
In Closing
The phrase “amateur theater troupe” has been trending these past two years, meaning: the world is one giant makeshift stage—don’t imagine anyone is as professional as they look.
I’d like to add one line to it. The troupe keeps performing not because the people in it are stupid, but because every layer is taking the layer below’s money and betting that the layer below won’t blow up.
The boss bets that the three-to-five-thousand person has a DRI’s sense of responsibility, just like this morning I bet that a platform still running a 2008-era system would protect my national ID number—both sides saved money that shouldn’t have been saved, then staked the difference on someone else.
Once AI arrives, the bet only gets bigger: one person does the work of three, three people get “optimized out,” and the experience, judgment, and “this has something to do with me” hidden inside those three jobs get optimized out along with them.
The one who stays behind draws the same salary and is held to the same accountability.
The only difference is that he now has an AI—and one less set of safety nets underneath him.
评论
0 条评论