It’s probably the only one with the balls to.
Apple doesn’t appear to have an unusually bad hiring system. I’m singling Apple out because it’s one of the very few companies on Earth with the brand, capital, engineering talent, recruiting scale, and gravitational pull to simply stop doing this stupid thing everybody does.
And yet Apple still mostly does what everybody else does.
Say a manager needs an engineer. A requisition gets opened. A job gets posted. Engineers are expected to find the right listing, decide it describes them, raise their hands, feed themselves into the machine, and wait to see if the machine likes them.
Apple’s own careers site currently returns 600+ results in just one Software and Services subcategory. Individual jobs still terminate in that familiar command:
Submit Resume.
Why?
For a company like Apple, the job application is obsolete technology.
The one thing you could usually count on Apple for was getting an Apple experience out of an Apple product. Sometimes that meant a beautiful interface. Sometimes it meant a shitty laptop keyboard. But it was recognizably Apple.
Seeking employment with Apple is weirdly un-Apple-like.
Apple obsesses over the experience of touching almost everything it makes, then hands prospective employees over to the same generic rituals everybody else uses.
I use the word rituals purposefully. At least a séance or a rain dance gives you a reason to keep cool clothes in your wardrobe. A job application is a sad yank by comparison.
Humans instinctively fear and hate Human Resources.
Rather ironic, really.
Every prestige employer says some version of this.
We hire the best. We’ve got an incredibly high bar. Our people are exceptional.
Great.
Best at what?
Best at LeetCode?
Best at writing software?
Best at making Apple money?
Best at solving the problem the team actually has?
Best at having exactly the right sequence of job titles?
Best at triggering the interviewer’s “this person reminds me of me” reflex?
Best at having nothing better to do that afternoon than fill out another application?
These aren’t the same thing.
If the status quo is “hiring only the best,” then I submit that a lot of very expensive companies are measuring the wrong things, optimizing the wrong things, and doing so hilariously badly.
Here’s the question I’d like to hear asked in the boardroom:
How do you know your selection process ranks the thing you actually care about?
(!(how selective it is))
(!(whether your interview loop is "rigorous"))
Does the thing actually predict who’ll create the most value after you hire them?
If you can’t answer that, “we have a high bar” just means you reject a lot of people. It doesn’t mean you reject the right ones.
Steve Jobs used to say:
A players hire A players. B players hire C players.
I’d update it for the ATS era:
A players hire A players. B players hire C players. F players feed A through C into an ATS and report a talent shortage.
Suppose the best person for an Apple role already has a good job. They’ve got kids. They’ve got a side project. They maintain an open-source library. They’re doing research. They’re generally busy being the sort of person Apple supposedly wants.
They see the job, see the ritual, and think:
Nope. Not doing this again.
Nothing bad happens in Apple’s metrics. There’s no rejected candidate, no recruiter loss, no dashboard flashing red. That person simply never appears.
And the more valuable someone’s alternatives are, the more rational that decision becomes.
The process may not just shrink the applicant pool. It may shrink it in the wrong direction. The people with the least tolerance for wasted time are often the people with the most valuable time.
Then HR looks at what’s left and adds another filter. Another assessment. Another interview. Another little proof-of-seriousness ritual.
Congratulations. You’ve created adverse selection and named it rigor.
Every technology company says it wants “rock stars.”
Fine.
Imagine a nightclub that wants the best rock band in town and tells the bouncer to pick the best musicians from whoever gets in line.
The bouncer can be spectacular at his job. He can reject 99% of the line. None of that answers the obvious question:
Why do you think the best musicians are standing in your line?
Mötley Crüe doesn’t wait in line.
If you actually want Mötley Crüe, somebody from the club goes and gets Mötley Crüe.
That’s what makes “rock star engineer” rhetoric so unhinged. Companies claim to want scarce, exceptional people and then make voluntarily joining the same cattle chute as everybody else the first qualification.
Management surveys the queue and announces:
It’s getting harder to find rock stars.
No. You’re looking for Mötley Crüe in the velvet-rope line.
Go easy on management, though. They probably know this is a stupid system.
Managers suffer from the same unfortunate human compulsion as the rest of us: they like remaining employed. Stupid systems have one enormous institutional advantage over untested better ones:
their paychecks always direct-deposit on time.
The recruiting director who says “maybe our entire model of hiring is backwards” creates a new problem for themselves. The one who says “we should improve funnel conversion by 7%” gets invited to the next planning meeting.
So put some of the blame where corporate governance says it belongs: with the ivory-tower board of directors asleep at the switch.
The bouncer shouldn’t have to redesign the nightclub while he’s working the door. The owners should eventually realize they may as well offer Kid Rock a residency, because the top acts are finding—and packing—their competitors’ clubs.
Not “apply to Apple once.”
Zero.
When Apple authorizes a difficult technical role, its recruiting system should answer:
Who, out of the set of all humans currently alive on Earth, are the three strongest people we could realistically recruit for this job?
Now. Not after 4,000 people apply. Not six months from now.
Now.
Apple already has access to a ridiculous amount of professionally useful information: employment histories, patents, research papers, conference talks, GitHub, open-source work, shipped products, professional websites, previous interactions with Apple, employee referrals—the ordinary professional internet.
That doesn’t mean scraping people’s private lives or building creepy dossiers. It means doing what executive recruiters have done forever:
Look for the people.
Approve the role, figure out the actual problem, find the strongest plausible people, let experienced humans challenge the result, then call them.
We’ve got a real role. We think you might be unusually good at it. The hiring manager wants to talk.
Number one says no? Call number two. Then three. Keep going.
Headhunters call that retained search. It’s not new. What’s new is using modern software to make it cheap enough to do for far more jobs than CEO.
Instead, the industry seems determined to use more and more AI to process larger and larger piles of applications containing more and more AI slop.
It’s a battle of the bands and it’s Norteño-Banda night.
Apple knows its own organization, which teams actually have headcount, what those managers really need, and which qualifications are boilerplate.
But the outsider gets hundreds of internal job records and is expected to guess which one Apple meant for them.
Why is the human applicant performing the database join?
Apple’s job isn’t to collect the best pile of applications. It’s to find the right three people before they’ve even considered applying.
A giant applicant pile isn’t evidence of successful recruiting. Quite the contrary. Like using lines of code to model who your best coders are, it’s evidence that the matching problem got dumped onto the candidates.
Imagine Apple’s head of procurement reporting this to the board:
We only use the world’s best capacitors.
Wonderful. How do you know?
We accept bids from suppliers that happen to find our requests for quotation.
How much of the global supplier base do you actually search?
Maybe ten percent.
Do your screening criteria predict field reliability?
We’ve never really tested that.
Do you conduct extensive sample evaluation and destructive testing on the capacitors you do manage to find?
Of course.
And have you established that those tests identify the capacitors that’ll perform best in the actual product?
Not really. But they’re very rigorous tests.
What happens to potentially superior suppliers who never enter your procurement process?
We don’t know they exist.
So you destructively test the samples you happen to receive, using a demanding qualification process you haven’t shown predicts the thing you actually care about, while most of the supplier universe never gets examined?
Yes, but we only use the world’s best capacitors.
You’d fire somebody.
Capacitors have datasheets, but engineers are vastly more consequential than capacitors.
And also vastly more leaky.
A capacitor is fungible. A commodity with specifications. An engineer has a manager, a team, a project, timing, incentives, chemistry, autonomy, judgment, curiosity, technical taste, and luck.
The engineer who’s brilliant on one team may be merely good on another.
So this isn’t even really a ranking problem. It’s a matching problem.
Which means your talent-search system should be better than your parts-sourcing system, not worse.
Nobody at Apple is sitting around saying:
Let’s hire middling engineers.
Companies don’t usually become mediocre by explicitly voting for mediocrity. The stated goal stays “find extraordinary people.” The daily work becomes controlling applicant volume, standardizing interviews, hitting time-to-fill, keeping the pipeline moving and maintaining the bar.
Eventually you can wind up selecting for people who are good at getting hired by Apple rather than people who’ll be unusually valuable once they’re there.
Apple doesn’t have to decide to hire worse people. The system they decided to use makes that decision for them.
And somewhere inside Apple are probably people who know the mythology and machinery aren’t quite the same thing. Some may even suspect Apple wouldn’t hire them today.
Pointing that out too aggressively inside a giant company isn’t always an excellent career move, so people learn the vocabulary.
Maintain the bar. Follow the process. Don’t become the problem.
This is how great companies become OK companies.
Jobs had a reputation for being perfectly willing to tell you your work was a pile of shit if that’s how he saw it. Even and especially while smelling like it himself.
Whatever else you think of that management style, Apple could occasionally use that degree of institutional candor.
Steve’s not here.
But Apple, I’m perfectly willing to tell you your HR system is a pile of shit.
The difference is, I’m ready, willing, and able to help you fix it. Well, that and roll-on deodorant.
Leave Apple for a moment and look at the broader software market. Greenhouse currently estimates that about 19% of advertised jobs are ghost jobs.
As someone coming from the trenches, that sounds hopelessly optimistic.
You don’t need 90% of employers manufacturing fake jobs to get a job market that feels 90% fake. The real jobs just have to disappear faster than the advertising does.
Imagine 1,000 ads in a healthy market, 850 of them real. Genuine hiring falls 90%, leaving 85 actual opportunities. But frozen requisitions linger, evergreen jobs refresh themselves, geographic duplicates survive, agencies copy them, aggregators scrape them, and aggregators scrape the aggregators.
You can still have hundreds of ads floating around while most of the transactional demand underneath them has vanished.
Craigslist might be the accidental control experiment. It’s primitive.
Good.
It still charges actual money for job postings. If you need a $200,000 engineer, a small posting fee is noise. If you want 200 speculative, evergreen, stale or geographically duplicated jobs floating around indefinitely, somebody eventually gets a bill.
Bills cost.
Modern ATS infrastructure lets a requisition cast an enormous shadow across the Internet at effectively zero marginal cost. Craigslist asks somebody to pay again.
That tiny friction might make Craigslist a more honest sensor of hiring intent than the sophisticated systems that replaced it.
Not because Craigslist is better.
Because Craigslist stayed away from the acid.
The experiment is pretty simple: compare the long-run collapse in paid Craigslist software listings with Craigslist usage overall, then compare that with apparent inventory on LinkedIn/Indeed/ATS systems. If software Craigslist collapsed vastly faster while modern job-board inventory stayed strangely fat, I’d want to know why.
And “ghost job” is too narrow anyway. A job can exist and still be effectively fake to you. Internal candidate already has it. Manager already knows who they want. Three staffing companies advertise one opening. Five cities are really one requisition. Recruiter stopped seriously examining inbound after applicant 200.
Those can all be real jobs. They aren’t necessarily real opportunities.
Naturally we respond to all this by making applicants work harder.
Upload résumé. Watch the billion-dollar ATS parse the résumé deliberately badly and make sure the wrong fields are populated and the others are blank. Now retype résumé. Create account. Confirm email. Enter dates already on résumé. Write cover letter.
Tell us why you’re passionate about Enterprise Middleware Platform 7.
Take assessment. Record asynchronous video. Send a photocopy of your ID, Social Security card, passport, birth certificate, Blockbuster Video Membership Application from 1989.
Then silence.
Or get sent to a .NET error page, or watch the billion-dollar ATS insist one of its fields needs to be populated even though it already is.
The jobosphere of the 2020s is a volunteers-OnlyFans.
Everybody’s selling, nobody’s buying, and somehow everyone’s still filling out forms.
Applicant time isn’t merely treated as free. Sometimes it’s treated as having negative value. Make people waste enough time and some leave, which reduces the screening burden.
From the local HR dashboard, that can look useful.
From the company’s perspective it’s insane.
If the people leaving first are the people with the strongest alternatives, the employer is using inconvenience to screen out exactly the people it claims to want.
And the most expensive false negative never appears in the data.
This essay isn’t called Apple HR Sucks More Than Everyone Else’s.
I don’t think it does.
Apple HR sucks because it’s ordinary.
And Apple has no business being ordinary here.
Apple doesn’t generally accept:
That’s just how our industry does it.
Ports disappear. Silicon gets redesigned. Entire product categories get rethought. Milliseconds matter. Battery cycles matter. A fraction of a millimeter matters. Supply-chain yield matters.
Then Apple goes shopping for perhaps its highest-leverage input and apparently decides the appropriate interface is:
Submit Resume.
Because that’s what employers do.
That should embarrass Apple.
Apple’s board of directors has less interest in an applicationless talent pipeline than Steve Jobs had in deodorant.
They also smell worse somehow.
This is a shareholder problem, not an HR problem. Apple doesn’t have to hire dramatically worse people for mediocre talent selection to cost shareholders billions. The payoff distribution is too lumpy.
The important miss isn’t somebody who’s 10% less productive. It’s the engineer who would’ve killed a disastrous architecture decision, shipped something six months sooner, or invented the thing everybody else copies five years later.
So here are the questions worth asking: does the hiring system actually predict what Apple cares about, what does a false negative cost, and how much realistically recruitable talent never enters the funnel in the first place?
And has Apple ever tested a system where there is no funnel?
If you’re on Apple’s board of directors, ask your favorite AI what replacing the ATS model with a Boba Fett-style direct-acquisition engine might be worth.
Maybe it says the value is enormous. Maybe it tells you I’m full of shit.
Great.
Run the experiment.
Then ask why nobody on the board has run it and, assuming nothing changes in 24 hours, riddle me this: what idiot hired you?
Don’t optimize the application.
Kill it.
Figure out the actual problem, find the strongest plausible people, have smart humans argue about the ranking, and call them. If they say no, keep going. Then see whether the people your system thought would be exceptional actually turn out that way.
Don’t tell shareholders:
We hire only the best.
That’s cope.
Say something that could actually be wrong.
We hire intelligently.
We hire well.
Better yet:
We hire profitably.
I’ve got trouble imagining a shareholder objecting to that one.
And Apple:
Hire me. I’ll help fix it.
The feds will only take off this ankle monitor if I get a job offer. No cap.
If Apple proves the job application itself is obsolete, Google copies Apple. Meta copies Apple. Microsoft copies Apple. Nvidia copies Apple. Recruiting vendors adapt. Smaller companies inherit the tooling.
Before you know it, companies have a stub for a “Careers” page. Or none at all.
And maybe the software labor market starts drifting back toward something resembling a meritocracy, if it was ever really one in the first place.
Not because engineers finally learned how to apply better.
Because the companies claiming to hire the best finally learned how to look.
Apple HR sucks. Not because it’s exceptionally bad.
Because it’s exceptionally ordinary.
And ordinary is a strange thing for Apple to aspire to.