The Job Description Is a Fictional Document

A job description is written in the past tense, about a person who already left, for a role that will look different in six months. The requirements list isn't a hiring spec. It's a wish list assembled by committee, run through filter logic, and dressed up as a standard. When you read it as a contract and count yourself out at 70%, you're treating fiction as fact.

That's not a motivational point. It's a structural one. The document and the hiring decision are two different things, made by different people, at different times, against different criteria. Most candidates never learn that, so they let the posting do their rejecting for them.

Why is the job description written backward?

A posting almost never starts from "what does this role actually need." It starts from the last person who held it. Someone exports the old req, the hiring manager adds two things that annoyed them about the previous hire, HR pastes in the standard degree line, and three stakeholders each tack on their pet requirement. Nobody zeroes it out and rebuilds from the work.

So the posting says "5 years Python" because the last person had five years of Python. It says "bachelor's required" because the template said so. The list is an archaeological record of past hires and committee anxieties. It is not a competency model.

In Praxy's own job-postings data, we keep seeing early-career roles carrying multi-year experience asks: "entry-level" listings that want three years, "associate" roles that want five. The posted requirements and the people who actually get hired are not the same population. That gap is the whole story.

What's the best-documented version of this problem?

Degrees. In 2015, 67% of production-supervisor job postings required a college degree, while only 16% of people actually doing that job had one. That's a 51-point gap between what the posting demanded and what the role needed. The other 84% were running production lines fine without the piece of paper the posting would have screened them out for.

The mechanism is mundane. In slack hiring markets, when there are more applicants than openings, a degree gets added as a cheap proxy filter. It cuts the pile faster. Then the market tightens and the requirement never comes back off, because nobody owns removing it. That same Harvard research, drawn from 26 million job postings, projected 6.2 million middle-skill jobs at risk of being fenced off this way, with 3 in 5 employers admitting they reject qualified experienced candidates in favor of fresh graduates who hold the credential.

The reset, when it comes, is slow but real: 46% of middle-skill and 31% of high-skill occupations saw material degree resets between 2017 and 2020, 63% of it structural rather than a pandemic blip. The requirement was never load-bearing. It was a habit.

The honest caveat: this is not universal. In medicine, law, structural engineering, and nuclear safety, credential requirements carry legal and safety weight. A degree gate on a surgeon is not lazy. But most postings aren't surgery. The default assumption should be "this requirement is a proxy until proven otherwise," not the reverse.

How many of the "nos" were never real openings?

Here's the part that should change how you read silence. A meaningful share of live postings aren't real openings at all. On the Greenhouse platform, 18 to 22% of listings in any quarter are classified as ghost jobs, and 3 in 5 candidates say they suspect they've hit one. In a separate survey of hiring managers, 40% admitted to posting at least one ghost listing in the past year, with the most common reason being "to appear open to external talent."

So when you apply, hear nothing for six weeks, and conclude your resume failed: there's roughly a one-in-five chance there was no one on the other end to impress. The role was filled, paused, or posted to look busy. The silence wasn't a verdict on you.

The honest caveat: survey numbers run hot. Self-reported employer surveys capture intentions and loose definitions; platform hiring analytics, when they're shared, tend to show that most live postings do resolve in a real hire. The truth sits between them. But "between them" still means a real fraction of your rejections were structural, not personal. Stop scoring yourself on them.

Doesn't the "apply anyway" advice come from a debunked stat?

Probably you've heard it: men apply to jobs when they meet 60% of the requirements, women only when they hit 100%. It gets cited everywhere as a confidence story about women. It traces back to a speculative comment by one HP executive in a McKinsey interview, with no study and no methodology behind it. The author later confirmed no original document exists.

When someone actually surveyed 1,000+ professionals, the real reason people don't apply when underqualified was rarely lack of confidence (only 10% of women, 12% of men), and was overwhelmingly rational self-filtering: 41% of women and 46% of men said they "didn't think they'd get hired and didn't want to waste the time". Note the near-even split. This isn't a women's confidence problem. It's a document-design problem that hits everyone.

The point: people treat requirements as binary because the posting gives them no signal that they're negotiable. There's no "must-have vs nice-to-have" label. So a reasonable person reads the list as a gate and allocates their time elsewhere. The fiction caused the dropout. (Both can be true, by the way: the survey behavior is near-even, and there's still a documented confidence gap that shows up differently across groups. The structural fix and the individual one aren't in competition.)

Is "skills-based hiring" actually fixing this?

Mostly it's theater. Companies have spent three years announcing they're dropping degree requirements. The 2024 Harvard and Burning Glass analysis of 11,000+ postings is blunt about what changed: hiring of non-degree candidates rose by just 3.5 percentage points from 2014 to 2023, fewer than 1 in 700 new hires actually benefited, and almost all of the real change happened in only 37% of the firms studied, meaning roughly 63% showed no change in hiring behavior at all.

The announcement is the easy part. The work underneath it isn't.

	Weak version (most firms)	Strong version
Careers page	"We've dropped the degree requirement"	Same
ATS criteria	Still auto-filters on degree	Rebuilt around competencies
Hiring managers	Still default to degree as a tiebreaker	Retrained, given a rubric
Replacement signal	None. Degree just deleted	Skills assessment in its place
Measured outcome	Nobody tracks it	Non-degree hire rate, tracked

The weak version is a press release. The strong version is an operating change. IBM did closer to the strong version: roughly half of its US roles no longer require a four-year degree, and the degree-free hires performed equivalently with better retention. The degree had been a social filter, not a competency one. But IBM is the exception that proves how much work the rest skipped.

The honest caveat: the poor track record doesn't mean requirements are evil. A degree is a flawed but functional signal. Deleting it without putting a real assessment in its place doesn't help anyone. The fix isn't fewer standards. It's better ones.

So what should you actually apply to?

Read the posting as a ranked wish list, not a checklist. Three buckets:

• Genuine blockers. You truly lack a foundational skill the work depends on every day. A backend role needs you to write code. Respect these. This is where agency meets reality: you close the gap or you pick a different target, but you don't bluff it.
• Label mismatches. You have the skill; the posting calls it something else. You ran "stakeholder alignment"; they wrote "cross-functional leadership." Same work, different vocabulary. These are not gaps. Translate and apply.
• Arbitrary proxies. "Bachelor's required" with no licensing reason. "5 years experience" with no competency anchored to it. The five-year line usually means "we want someone who's seen the full cycle of this problem go wrong and get fixed." So show that you've seen it. Don't apologize for the year count.

Weak application: reads "5 years experience, bachelor's preferred," has four years and a diploma, closes the tab. Self-rejected. The ATS never saw them.

Strong application: same person notices the five-year ask is really about having owned an end-to-end cycle, leads with "I shipped two full launches of exactly this, start to incident to fix," and applies. They addressed the underlying need instead of the literal line. This is the same instinct that beats the ATS screening myth: you're answering what the role needs, not what the form printed.

The trade-off, named plainly: applying to roles where you hit 70% costs you time, and some of those applications will go nowhere because a real blocker existed or the posting was a ghost. For people from non-traditional backgrounds, that rejection carries a real psychological cost on top of the time. The advice is correct at the population level and can still sting at the individual one. Hold both. The fix is to spend that time on the label-mismatch and proxy roles, where the gap is fiction, and not on the genuine-blocker ones, where it isn't.

What to do now

1. Pull up the last three postings you talked yourself out of. For each requirement you "failed," label it: blocker, label mismatch, or arbitrary proxy.
2. If it's a mismatch or a proxy, rewrite your top line to address the underlying need, then apply this week.
3. Stop counting a six-week silence as feedback. A fifth of those were never real openings. Move on faster.
4. Reserve your real energy for closing one genuine blocker. That's the only gap worth grinding.

Consistency beats intensity here too: a steady stream of well-targeted applications to roles you're 70% matched on will out-perform a heroic, infrequent push at "perfect-fit" listings that mostly don't exist. And it pairs with knowing your own number before you walk in, which is the whole point of negotiating without a competing offer.

Want a read on whether a requirement is a real blocker or just the posting talking? Send me the job description on WhatsApp and I'll tell you which lines to ignore, which to translate, and the one gap actually worth closing before you apply.