Skip to content

Career / Public Proof-of-Work

Public Proof-of-Work — the new résumé.

Build logs, write-ups, and open artifacts as the most credible signal an operator can produce. Why public iteration outperforms private polish — and how to close the gap between genuine capability and visible capability.

Code CRR/03Domain SignalStage OngoingPosture Operator
§ 01The Default

Most capability is invisible to the market by default

The default condition for a technical practitioner is private competence. Work accumulates — systems built, problems solved, decisions made — but the accumulated work is invisible outside the immediate context where it was produced. The client saw the output. The employer saw the output. No one else did. The practitioner becomes progressively more capable without becoming progressively more visible. Capability and market signal decouple.

This is not a failure of the practitioner. It is the natural output of a system optimized for delivery. Every hour spent delivering is an hour not spent producing a public artifact that would make the delivery visible. The incentives of the labor model — discussed in the Labor → Leverage framework — actively suppress artifact production. The client pays for output, not for the documentation of output. The documentation has to happen in addition to the work, with no immediate return.

The market consequence is that technical capability is proxied by signals that are weakly correlated with it: credentials, institutional affiliation, references from known contacts. These signals are legible to buyers but they describe where the practitioner has been, not what they can do. A practitioner ten years into independent work, with no institutional affiliation and no credential post-dating their initial training, is structurally harder to evaluate than a practitioner six months out of a recognizable institution — regardless of relative capability.

Proof of work is the alternative signal architecture. Instead of proxies — where have you been, who do you know — it provides direct evidence: here is the work, here is how I think about it, here is what I built and what it does. The buyer evaluating an artifact is evaluating the practitioner directly, without the information asymmetry that proxies create. Public proof of work is structurally advantageous for any practitioner whose actual capability exceeds what their credentials can represent — which, after the first several years of practice, is most serious practitioners.

§ 02Failure Modes

Three ways proof-of-work strategy fails

The decision to produce public proof of work is the easy part. The execution fails in three structurally distinct ways that are worth naming precisely — because they feel productive while producing nothing of value.

Polish over iteration

The practitioner waits until the work is finished before publishing. The finished state is perpetually deferred — every artifact has one more revision cycle before it is ready to show. The result is a private archive of almost-published work and no public signal. The error is treating publication as a quality gate rather than a practice. The artifacts that produce the most opportunity are frequently not the most polished — they are the most honest. A build log that documents what broke is more credible than a case study that describes only what worked, because the former requires the author to have actually built something.

Credentials over artifacts

The practitioner invests in certifications, courses, and institutional affiliations as the primary signal of capability. These have legitimate value in contexts where buyers rely on credentialed intermediaries to evaluate practitioners. In the markets where independent technical practitioners compete for high-value work, credentials are table stakes — they establish a floor, not a ceiling. A buyer selecting a practitioner to build and operate AI systems infrastructure is not primarily looking for the practitioner with the most certifications. They are looking for the practitioner who has demonstrably built and operated AI systems infrastructure. The artifact is the credential.

Activity over artifact

The practitioner is publicly active — sharing commentary, responding to discussions, participating in communities — but produces no standalone artifacts. Activity creates visibility without creating gravity. A social post disappears from attention within hours. A build log, a framework document, or a system architecture write-up is findable months or years after it was published, by buyers who are actively looking for exactly that content. Activity and artifact production are not the same investment. Only one of them compounds.

§ 03The Progression

Invisible → Emerging → Established

There are three structural stages in the proof-of-work progression. The stages are defined not by time or effort but by market signal: what a buyer can determine about the practitioner without a prior introduction. Most practitioners spend their entire career in the first stage without realizing it. The transition between stages is driven by artifact production — specifically, the type and quality of artifact.

Invisible: No public artifacts exist. Capability is real but unverifiable by any buyer who has not already encountered the practitioner through personal referral. Market access depends entirely on network — which means it is bounded by the size of the network and does not compound. Emerging: Public artifacts exist but are inconsistent in focus or frequency. The market has partial signal — enough to confirm the practitioner exists and has produced something, not enough to characterize what they do or select them on the strength of the work alone. Established: Multiple high-signal artifacts form a coherent pattern. Buyers can evaluate the practitioner from the artifact set without requiring an introduction. Inbound contact is generated by the artifacts, not by active outreach.

The type of artifact determines which stage is reachable. The diagram below maps artifact types by visibility — how findable they are — and buyer signal strength — how much they tell a serious buyer about actual capability. The claim is structural: most practitioners invest in artifacts that are high on one dimension and low on the other. Moving to Established requires artifacts that are high on both.

VISIBILITY →privatepublicBUYER SIGNAL →weakstrongINVISIBLEEMERGINGESTABLISHEDprivate notesinternal docscredentialssocial postscase studiesframeworks+ write-upsbuild logshonest iterationopensystemsthe gap to close
Fig. CRR/03 — Artifact signal spectrum. Visibility on x, buyer signal strength on y. Stage zones overlay the field. Build logs and open systems are the only artifact types that reach the Established zone.

The stage zones map directly to the artifact clusters. The Invisible stage contains only privately held or low-visibility work — internal documents, personal notes, credentials that require a buyer to already know the practitioner before they become legible. The Emerging stage contains artifacts that are either visible but low-signal (social posts) or moderate on both dimensions (case studies). The Established stage is occupied exclusively by artifact types that make the practitioner's thinking directly evaluable: framework documents, build logs, open systems.

The highest-signal artifacts share a common property: they are evidence of process, not just outcome. A credential says the practitioner passed a test. A case study says a project was completed. A build log says — here is how I think, here is what I tried, here is what failed, here is what I understand now that I did not understand before. The buyer evaluating a build log is evaluating the practitioner's thinking in real time. That is the actual thing being purchased.

§ 04Logs as Signal

Why build logs specifically outperform every other artifact type

For an AI systems practitioner, the build log is the highest-leverage artifact type available. The claim is not aesthetic — it is structural. Build logs produce stronger buyer signal per unit of effort than any alternative because they are the only artifact type that simultaneously demonstrates depth, honesty, and operational reality.

Depth: a build log describing the retrieval architecture of a production RAG system — the chunking decisions, the embedding model tradeoffs, the reranker choice and why — tells a buyer more about the practitioner's technical depth than any credential could. The credential says the practitioner knows RAG systems in the abstract. The build log shows they know this specific system in the specific operational conditions where it was deployed.

Honesty: a build log that documents failures is more credible than one that does not, because every serious buyer knows that real systems produce failures. A practitioner whose public record contains only successes either has not built anything substantial or is not being honest about what they have built. The buyer who has built systems knows this. The practitioner who documents failure signals that they are serious enough to acknowledge it — which is the same quality that makes them trustworthy to work with on the buyer's own systems.

Operational reality: the build log demonstrates that the practitioner does not just understand systems conceptually — they operate them. The Researcher's Path framework describes the practitioner who has moved from knowing to doing to compounding — where depth is demonstrated not through explanation but through the artifacts of practice. A build log is one of the clearest artifacts of practice available.

The publication you are currently reading is itself a proof-of-work artifact. The build logs section documents the systems built in this lab — not as polished case studies, but as operational records. The frameworks describe how the thinking works. The systems describe what was built. The research documents what is still being investigated. A buyer reading any page is evaluating the practitioner directly.

§ 05Public Iteration

Visible iteration changes who reaches out — not just how many

The first-order effect of public iteration is visibility. The second-order effect — the one that matters more economically — is filtering. Public iteration that is specific about domain, method, and the difficulty of the problems being solved attracts buyers who understand that domain, those methods, and those problems. It repels buyers who do not.

This filtering is valuable in a way that is not immediately obvious. A practitioner who produces vague, broadly appealing content attracts a broad set of inquiries, most of which are misaligned — buyers who want something different from what the practitioner does well. Managing misaligned inquiries is expensive. Declining them leaves an impression. A practitioner who produces technically specific, operationally honest artifacts attracts fewer inquiries that are far more likely to be aligned. The conversion rate on aligned inquiries is dramatically higher, and the engagements that result are more likely to produce further artifacts worth adding to the public record.

The composition of inbound shifts in a direction that is legible only in retrospect. Early in the artifact production process, the practitioner cannot distinguish between a buyer who found them via a specific artifact and a buyer who found them via network. Over time, as artifact-driven inbound increases relative to network-driven inbound, the difference becomes measurable: the artifact-driven buyer has already done their evaluation before the first contact. They know what the practitioner does. They have a specific problem. The conversation starts further along.

The filtering also operates in reverse. Artifacts that are technically specific and operationally honest make certain buyers self-select out — buyers who want surface-level engagement, who are not prepared to work at the level the artifacts represent, or who are looking for a vendor relationship rather than a technical partnership. This is not a loss. The practitioner who is explicit about their level of work in their public artifacts has effectively pre-qualified every buyer who reaches out.

§ 06Artifact Gravity

Persistence is the property that generates inbound without outreach

Gravity is the property that makes buyers come to the practitioner rather than the practitioner going to buyers. It is not a function of volume — a practitioner with three high-specificity, deeply substantive artifacts can have more gravity than a practitioner with fifty surface-level ones. It is a function of findability, specificity, and the match between what the artifact addresses and what buyers are actively searching for.

Findability requires that the artifact be indexed, linkable, and searchable — properties that long-form published work has and social activity does not. A framework document published today is findable in three years by a buyer researching the specific problem it addresses. A social post published today is not findable in three days by anyone who was not already watching. The persistence of long-form artifacts is the mechanism behind passive inbound — not engagement, not reach, not network. The buyer finds the artifact while researching the problem. The artifact establishes credibility before any contact occurs.

Specificity amplifies findability. A general overview of AI systems reaches many people and generates weak signal. A precise treatment of hybrid retrieval architecture for production RAG — covering BM25 tradeoffs, reranker calibration, and the failure modes of pure vector search — reaches fewer people and generates strong signal from exactly the buyers who are working on that specific problem. The specificity that limits reach is the same specificity that makes the artifact the authoritative result for the buyer who needs it most.

A publication with multiple high-gravity artifacts develops compounding gravity: each artifact reinforces the others, and the practitioner's positioning becomes legible not just from any single artifact but from the pattern across them. The buyer who reads a build log, finds a corresponding framework document, and finds a corresponding research investigation has developed a comprehensive picture of the practitioner's thinking — at their own pace, without any action from the practitioner. That is what the Consulting as Leverage framework describes when it refers to inbound that arrives pre-qualified: the buyer has already done the evaluation. The conversation starts at a different level.

§ 07Proof-of-Work Audit

Where you currently sit — three-stage diagnostic

Three diagnostic stages. Each stage has a description of its structural characteristics, diagnostic questions, and a concrete next move. The audit does not produce a score. It surfaces an honest description of current market signal and the move that follows from that description.

Stage 01Invisible
Stage 01

No public artifacts. The practitioner's work exists — it is real, often substantial — but it is not visible to any buyer who has not already encountered them through personal referral.

  • 01Has any work you have done in the last 12 months produced a public artifact — a document, a system, a published analysis — that a buyer could find without being introduced to you?
  • 02If someone searched for your name or your area of work, would they find evidence of what you can do?
  • 03Do you have a single piece of written work that represents your current capability at its best?
  • 04Could a buyer who encountered your work for the first time assess your level without speaking to you?

No to any of these: the practitioner is invisible to the market regardless of how deep their capability runs. The next move is singular — produce one artifact, in public, that represents current best work. Not a thread. Not a summary. A complete, standalone piece.

Stage 02Emerging
Stage 02

Public artifacts exist but are inconsistent in quality, infrequent in production, or misaligned with the buyers the practitioner wants to attract. The market has partial signal.

  • 01Do you have more than three public artifacts produced in the last 18 months?
  • 02Are your public artifacts findable by the buyers you want to attract — do they surface in the searches, communities, or channels those buyers use?
  • 03Has any public artifact produced inbound contact from a buyer you did not already know?
  • 04Is there a coherent thread across your public artifacts — a consistent focus, perspective, or domain — that a buyer could use to characterize what you do?

Yes to one or two: emerging. The artifacts exist but are not yet compounding. The next move is consistency and coherence — producing artifacts on a defined focus often enough that a pattern becomes visible to buyers over repeated exposure.

Stage 03Established
Stage 03

Multiple high-signal public artifacts are generating inbound from buyers the practitioner did not actively pursue. The artifact set has gravity — it surfaces opportunities without active outreach.

  • 01Do you receive inbound contact from buyers who cite specific artifacts you produced — not your credentials, not your network, but the work itself?
  • 02Do your artifacts surface when buyers research the domain you work in, independent of your direct promotion?
  • 03Is your inbound volume sufficient that you can be selective about which opportunities to pursue?
  • 04Do the opportunities you attract align with the work you want to do — not just the work you have done?

Yes to three or more: established. The risk at this stage is complacency — the artifact set that built the position represents past capability, not current capability. The compounding continues only if the artifact production continues. Stopping produces a position that decays faster than the artifacts age.

The Argument

Why artifact production compounds and activity does not

The compounding function of proof-of-work is the same function that makes the labor-to-leverage transition worth making: it is back-loaded. The first artifact produces almost nothing. The tenth artifact produced by a practitioner with a consistent focus is ten times more findable than the first, because the pattern of artifacts creates a signal that no single artifact can create. The twentieth is not twenty times more valuable than the first — it is substantially more, because each artifact increases the credibility of every other artifact in the set.

The practitioner who understands this makes different decisions. They do not wait until the artifact is perfect before publishing. They publish at the quality threshold where the artifact is honest and complete, because the cost of delay is not just the lost opportunity of that artifact — it is the delayed start of the compounding. Each month of polishing in private is a month of compounding that does not happen.

They also resist the pressure to produce activity over artifacts. Social activity is easier to produce and yields faster feedback. It feels productive. The feedback loop is tight: post, receive response, feel visible. The artifact feedback loop is slow: write, publish, wait months for the artifact to surface in the right search, wait more months for the pattern of artifacts to develop gravity. The practitioner who can tolerate the slow feedback loop long enough to see compounding begin is the practitioner who eventually does not need to pursue opportunities — because the artifact set does it for them.

That is the full argument for public proof of work: not that it is better than other signals in the short term, but that it compounds in a way that other signals do not. The practitioner who starts now, at whatever quality level their current capability supports, is in a different position in three years than the practitioner who waits until the work is finished.

END · CRR/03

Continue Reading