The Professional Matching Problem No One Can Solve Yet
We know LinkedIn doesn't work. Here's why we're stuck with it anyway.
LinkedIn has your résumé. Your résumé lists things you did in the past. The job you want requires capabilities you’ll demonstrate in the future. Professional matching bridges this gap by asking, “did you do similar things before?”
Companies match using credentials — degrees, job titles, years of experience. Recruiters match using keywords. Networks match using proximity — who knows who, which company. All of these proxy for “can this person solve the problems this role creates?”
When you need someone who can navigate ambiguous technical tradeoffs in a regulated industry, “5+ years software engineering” doesn’t tell you much. Some engineers with three years can do this. Some with ten can’t. The credential measures time, not capability.
Even when people list skills, they’re self-reporting without evidence. Most people don’t know how to explain what they’re actually good at. They lack systems view of themselves. They misjudge their strengths, overstate capabilities, miss what makes them effective.
A résumé captures none of the interpersonal dynamics that determine whether someone will succeed in a role — how they handle ambiguity, how they collaborate under pressure, how they learn from failure. The full picture doesn’t fit in bullet points.
Good matching requires ongoing behavioral data — how someone approaches problems, how they think.
Your Slack messages show how you communicate under pressure. Your GitHub commits reveal code quality. Your meeting patterns demonstrate collaboration style. This data exists but stays siloed. Companies guard their platforms as competitive moats — Salesforce locked down Slack’s API, making it nearly impossible to access the behavioral data that lives there. LinkedIn has your job history but not your thinking process. GitHub shows your code but not your stakeholder management.
The complete picture requires combining streams that live in incompatible systems controlled by competing companies. Credentials stay current through inertia — your degree doesn’t expire. But cognitive profiles require active maintenance or real-time data streams. Neither scales.
Companies like Eightfold and Gloat solve internal talent matching because they operate inside work infrastructure. They ingest data from HRIS, performance reviews, project assignments, internal communications. They see who you collaborate with, what you’re working on, how you’re growing. The matching improves because the data stream stays active.
But this only works within company boundaries. Your Eightfold profile at Company A doesn’t transfer to Company B. The intelligence dies at the organizational edge. Switch jobs and you start over with a blank résumé, rebuilding professional identity from credentials instead of capabilities.
Effective professional matching requires cognitive digital twins — persistent models that learn continuously from behavioral streams and remain portable across contexts.
Your twin would know you’re good at translating technical complexity because it observed you doing this across dozens of conversations, not because your résumé claims “technical program manager.” This twin needs properties current systems lack.
Continuous learning from behavioral streams — not periodic LinkedIn updates, but ongoing ingestion of how you actually work, think, and communicate. The twin gets smarter as you do, capturing capability development in real time rather than waiting for you to remember to update a profile.
Cross-platform data synthesis — professional identity doesn’t live in one place. The twin must combine signals from Slack, email, GitHub, documents, meetings. This requires either platform cooperation (unlikely) or user-controlled data aggregation (complex).
Portability across institutional boundaries — your twin should follow you between companies and communities. When you leave Company A for Company B, your accumulated intelligence transfers. When you join a professional community, your twin enables discovery without rebuilding your profile from scratch.
Privacy controls with selective disclosure — full behavioral access creates surveillance concerns. The twin shouldn’t store everything you’ve done but extract patterns: “asks clarifying questions before proposing solutions,” “strong at cross-domain pattern recognition.” These insights enable matching without exposing raw behavioral data.
Building cognitive twins requires infrastructure that doesn’t exist. Companies owning your work communication have no incentive to enable portability — their moat depends on keeping data trapped. Building from scratch means competing with Slack while trying to build intelligence on top of it. Who owns your cognitive twin? What happens if the platform shuts down? Can you export your twin’s intelligence? Users want portability but companies want lock-in. The structural tension remains unresolved.
Professional matching defaults to credentials because credentials are accessible. Creating comprehensive cognitive profiles requires ongoing engagement with systems that don’t exist yet.
The technology for cognitive twins exists. What’s missing is architecture that makes them practical: platforms willing to share behavioral data, governance frameworks giving users control over their intelligence.
We’re stuck with credentials not because no one’s thought of better approaches, but because building them requires solving technical, economic, and competitive problems simultaneously. That’s why professional matching stays broken while everyone agrees it should be better.