David Arnoux recently shared something that caught my attention: "This headline isn’t actually about jobs. It’s about a Nobel prize you haven’t heard of… It’s a story about what happens when signals become free." That framing is doing a lot of work—and it nails the weird mood many hiring teams and candidates feel right now.

What David is pointing at isn’t a minor tooling change. It’s an economic shift in how trust gets built when strangers try to transact—whether that transaction is a date, a job interview, or a business partnership.

"Signals only work when they cost something. When they’re free, economists call it 'cheap talk.' Cheap talk is worthless." —David Arnoux

Below, I want to expand on David’s core idea (signals become meaningless when AI makes them cheap), connect it to Michael Spence’s signaling theory, and then get practical: what signals still work, and how candidates and employers can rebuild trust without reverting to needless bureaucracy.

The Nobel Prize idea behind your resume

David references economist Michael Spence’s 1973 work on signaling, which later earned Spence a Nobel Prize. The premise is simple: when two parties have asymmetric information (an employer can’t truly know a candidate’s ability; a buyer can’t truly know a seller’s quality), they look for signals.

But not all signals are equal. A signal only separates “real” from “fake” if it’s costly in a way that’s hard to mimic.

A peacock’s tail is costly to grow and maintain; that cost is what makes it believable.
A degree is partly about learning, but it also signals endurance, conformity to standards, and the ability to complete a long, structured commitment.
A well-crafted portfolio used to signal time, taste, and repeated iteration.

The key: the cost has to be harder for low-quality actors to pay than high-quality ones. If everyone can send the same signal at near-zero cost, the market stops trusting the signal.

What AI changed: the cost of “looking qualified” collapsed

David’s most provocative claim is also the most observable one:

"Here’s what AI did. It drove the cost of nearly every signal to zero." —David Arnoux

In hiring, a lot of the process has always been made of proxies:

A resume as a proxy for competence
A cover letter as a proxy for communication and motivation
A take-home assignment as a proxy for practical skill
A polished LinkedIn profile as a proxy for professionalism

AI didn’t just help people communicate faster. It made it cheap to produce the appearance of competence. Resumes that once required effort now require a prompt. Cover letters that once exposed thinking now often reveal prompt literacy and model selection.

And crucially, employers also started automating judgment. If an AI writes the application and another AI screens it, the system becomes a conversation between generators and filters—neither necessarily tethered to reality.

The result isn’t “more efficient hiring.” It’s more throughput with less trust.

Why the Tinder comparison is uncomfortable—and accurate

David notes that The Atlantic compared this dynamic to Tinder, and I think that analogy holds because both systems drift toward optimizing what’s measurable (activity) rather than what’s valuable (connection).

When sending a message costs you something—time, vulnerability, reputational risk—you think harder before you send it. When swiping and templated messaging are frictionless, volume replaces intent. The platform can show “engagement,” but participants feel a growing distrust that the other side is real, serious, or safe.

Hiring marketplaces can fall into the same trap:

Candidates spam applications because it’s easy.
Recruiters bulk-reject because it’s easy.
Everyone complains the process is broken, even while activity metrics look “healthy.”

David’s punchline is the one that matters for leaders designing systems:

"The strange implication: the solution isn’t efficiency. It’s friction." —David Arnoux

Friction isn’t a bug; it’s the trust mechanism

We usually treat friction as waste: steps to remove, time to shorten, clicks to reduce. But in signaling environments, some friction is the point. It forces prioritization. It makes intent legible.

Think about it like this:

If it costs nothing to apply, applying doesn’t mean much.
If it costs nothing to say you’re passionate about the role, those words don’t carry information.
If it costs nothing to claim expertise, the claim becomes cheap talk.

The uncomfortable reality is that trust often requires “burning something”: time, effort, reputation, or money. That’s why David highlights signals that still work.

Which signals still work when content is cheap

David gives three examples that share a common structure: they require commitment.

1) Referrals (reputational risk)

A referral works because someone puts their name on the line. Even a light referral (“I’ve worked with them; they’re solid”) has a cost: if the hire fails badly, the referrer’s credibility takes a hit.

That reputational collateral makes the signal hard to fake at scale.

2) Handwritten notes (time you can’t recover)

A handwritten note is not “better writing.” It’s a cost signal. It shows someone invested non-reusable time. In a world of infinite generated words, the effort becomes the meaning.

3) Long-form work samples (difficulty to fake)

A real work sample—especially one with depth, decisions, tradeoffs, and artifacts—has texture. It’s harder to fabricate convincingly because the details have to cohere.

This is why “show your work” beats “tell your story” in AI-saturated markets.

Practical ways to rebuild credible signals (for employers)

If you’re a hiring manager or founder, the goal isn’t to add hoops. It’s to design signals that correlate with job success and are costly in the right way.

Use structured, job-relevant evaluations

Instead of open-ended prompts that AI can ace, use structured questions tied to real decisions the role makes. Ask for reasoning, constraints, and second-order effects.

Prefer “portfolio plus walkthrough” over “portfolio link”

A link can be curated or generated. A live walkthrough where a candidate explains tradeoffs, alternatives, and lessons learned is much harder to fake.

Add calibrated human judgment back into the loop

Automation can assist, but it can’t be the entire judge. Train interviewers, use scorecards, and run calibration sessions so human decisions become more consistent—not replaced.

Consider paid work trials (carefully)

A short, paid trial (with clear scope and compensation) is a powerful signal because it’s costly for both sides. The employer invests budget; the candidate invests effort; both reveal seriousness.

Practical ways to rebuild credible signals (for candidates)

If you’re applying in 2026, “more applications” is rarely the winning strategy. Credibility comes from signals that are costly for you in a way that also helps the other side make a decision.

Turn your experience into a real artifact

Write a teardown, a postmortem, or a case study. Include the messy middle: constraints, failures, what you’d do differently. The specificity is the moat.

Get introduced with context, not hype

A warm intro that includes concrete examples (“Here’s what they shipped, here’s the measurable impact, here’s how they work”) is far more credible than generic praise.

Create a “role-specific” micro-sample

Instead of a generic cover letter, produce something the role would actually need: a 1-page GTM plan, a support triage rubric, a dashboard redesign, a security threat model—whatever fits. Make it small but real.

The paradox David names: efficiency can destroy trust

David ends with a line that feels like the thesis for modern hiring:

"The paradox of this moment is that efficiency has become the enemy of trust." —David Arnoux

When communication is free, words inflate. When signals are cheap, claims flood the market. And when markets flood, everyone raises their defenses—leading to more filters, more automation, and even less trust.

So the north star isn’t “remove every step.” It’s “make the right steps meaningful.” The things that don’t scale—time, risk, human judgment—become valuable precisely because they resist mass production.

This blog post expands on a viral LinkedIn post by David Arnoux, Helping GTM Leaders & Founders Grow With GTM x AI | Fractional CxO | Building Linkedin Tools @ humanoidz.ai. View the original LinkedIn post →