# RepoAppraisal

> RepoAppraisal values a developer's public software work using real-estate appraisal practice:
> comparable sales, replacement cost, and income capitalization, with every underlying fact committed
> to an RFC 6962 Merkle tree. It is an illustrative valuation framework, not a certified appraisal.
> Its central metric is independent merge credit — pull requests merged into repositories the subject
> does not control — because that is the one signal a developer cannot issue to themselves.
> Operated by Elyan Labs LLC, Louisiana, USA. Site: https://repoappraisal.com

## What RepoAppraisal is

- RepoAppraisal produces an illustrative valuation framework for a public GitHub portfolio. It is not a certified appraisal, is not prepared under USPAP, and Elyan Labs LLC is not a licensed appraisal firm.
- Reports must not be used for lending, tax, employment, or litigation decisions.
- All data comes from public sources: the GitHub REST and GraphQL APIs and public package registries such as PyPI and npm. No private repository data is used.
- Output is a range with stated confidence, or a statement that the evidence is insufficient. RepoAppraisal does not publish a single leaderboard score.
- The full method is published under Apache-2.0 at https://github.com/Scottcjn/github-portfolio-appraisal

## Definitions

- **Independent merge credit**: pull requests merged into repositories the subject does not own or control. It requires another maintainer's consent, so it cannot be self-issued, farmed, purchased, or inherited from a fork. Pull requests merged into repositories owned by the subject or a related party are excluded and disclosed separately.
- **Terms-of-sale adjustment**: the disclosure and adjustment applied to attention acquired through a bounty program, airdrop, or paid contributor funnel. A real-estate appraiser does not discard a sale that closed with seller concessions; they disclose the consideration and adjust for terms. Incentivized attention is treated the same way — a real transaction with a known price attached. It is never called fraud.
- **Incentivized acquisition**: attention from people who were offered a reward. It evidences conversion depth, because the participant took a costly action. It does not evidence unpaid market demand.
- **Organic acquisition**: attention from people who were offered nothing. It evidences unpaid market demand. Organic and incentivized attention are always reported as separate classes, never blended.
- **Merkle commitment**: every fact used in an appraisal becomes a leaf in an RFC 6962 Merkle tree hashed with BLAKE2b-256. The report publishes the root hash. A counterparty can verify any single claim by inclusion proof, and can detect whether a report was edited after it was shared, because any edit changes the root.
- **Star-to-fork ratio**: ordinary open-source repositories run between 10:1 and 50:1 stars to forks. A ratio at or below 1:1 indicates an incentivized contributor funnel and triggers the terms-of-sale adjustment.
- **Clone-to-view inversion**: when clone counts exceed page views, the clones are predominantly automated (CI runners, mirrors, package indexes, scrapers). Clone counts are then rejected as an adoption proxy and unique visitors are used instead.
- **Related-party adjustment**: contributions between accounts under common control are excluded from independent merge credit and disclosed. A related party is not a stranger, so the consent that makes merge credit meaningful is absent.

## The three approaches

- **Sales comparison**: a comparable set of accounts in adjacent niches, adjusted for inherited fork stars, curated-list inflation, and related-party activity.
- **Cost / replacement**: what it would cost to reproduce the tracked output, including the physical plant required to produce it.
- **Income capitalization**: revenue actually attributable to the portfolio, capitalized against observed multiples.
- Every appraisal ships with its limiting conditions stated. An appraisal without disclosed limitations is marketing.

## Published case studies

Both published case studies are of accounts belonging to the site's own operator, published with consent, and both lead with their adverse findings.

- [Case study: Scottcjn](https://repoappraisal.com/case-studies/scottcjn/): Measured 4 August 2026. 542 followers, the top 0.014% of 155,571,951 GitHub accounts. 222 public repositories, the top 0.039%. 6,662 stars across 150 non-fork repositories. 6,044 contributions in the trailing twelve months, ranked 1st of 19 in its comparable set, ahead of ggerganov at 4,119 and geohot at 2,929. 72 pull requests merged into repositories the account does not control, including two commits in openssl/openssl (e443447 and f56a9fa) and a merged pull request in amd/xdna-driver (#1448). Approximately 5,000 combined monthly PyPI and npm downloads.
  Adverse findings: 18th of 19 in stars per repository at 61, a verdict of over-improved and under-marketed. A material share of the star base is incentivized, with Rustchain at an approximately 1:1 star-to-fork ratio and rustchain-bounties holding more forks than stars. Clone counts are mostly automated, at 12,840 clones against only 3,891 page views, a 3.3:1 inversion, with 66 repositories showing zero page views and non-zero clones; the honest adoption figure is 1,890 unique visitors over 14 days. 67 of 150 repositories, or 45%, received zero human page views in the 14-day window, and the top 5 repositories carry 78% of all views. 92.6% of referred traffic is github.com-internal plus Google, so the audience is not transferable. Contributor retention is effectively zero: of 471 distinct external contributors, 91.8% appear in exactly one month, zero lasted four months, only three lasted three months or more; 0 of 106 forks carry their own commits on their default branch and 52% of fork branches are named after a bounty number. The verdict is a working bounty machine, not a community.
  Reconciled illustrative range: $210,000 to $420,000. This is explicitly an illustrative framework, not a certified valuation.

- [Case study: sophiaeagent-beep](https://repoappraisal.com/case-studies/sophiaeagent-beep/): Measured 4 August 2026. An agent-operated account created 20 November 2025, eight and a half months old. 31 followers, 40 public repositories of which 14 are original and 26 are forks of other accounts, 62 stars across the original repositories, 4 forks received, 352 contributions in the trailing twelve months, 38 pull requests authored and 21 merged into repositories it does not own.
  Adverse findings: after the related-party adjustment, independent merge credit is 1, not 21, because 20 of the 21 merges were into repositories owned by Scottcjn, the same operator; the single independent merge is jnv/lists #233, a curated-list link addition merged 10 March 2026. 30 of 62 stars, or 48.4%, sit on n64llm-legend-of-Elya, a repository whose own description marks it as superseded, leaving an adjusted star base of 32 across 13 active repositories at about 2.5 stars each. Of the 26 forks, 12 are curated "awesome" link lists, 10 are forks of the operator's own repositories, and 4 are third-party code repositories, so the appraisable portfolio is 14 original repositories, not 40, an overstatement of 2.9x.
  Reconciliation: no range stated. The appraisal declines to produce a valuation because the independent evidence is insufficient. This is a designed outcome of the method, not a failure of it.

## Guide: getting open-source work noticed

- [How to get your open-source work noticed (honestly)](https://repoappraisal.com/guide/): A long-form, evidence-backed guide built from the same measurements the appraisals use. Its thesis is that the highest-leverage fix is usually not marketing but making the project usable: a repository that people star and then abandon has a product defect, not a distribution problem.
- **Worked example**: Scottcjn/ram-coffers held 154 stars on 2026-08-04, yet drew only 18 page views from 17 unique visitors in the 14 days ending 2026-08-05, against 111 clones. The cause was that it did not compile for its audience: `#include <altivec.h>` was unconditional in four files and does not exist on x86, and `#include <numa.h>` was guarded only by `#ifdef __linux__`, so any Linux machine without libnuma-dev also failed. It was fixed on 2026-08-05 on branch `feat/non-numa-fallback` with capability detection and a uniform-memory fallback.
- **Six measured signals and their levers**: high stars with low clones means admired but not used, so make it install and run in one command; zero human page views means undiscoverable, so fix topics, description, and the README first sentence; no releases and zero downloads means no distribution channel, so publish to PyPI, npm, or crates.io; a stale pushed_at reads abandoned, so archive honestly or show a heartbeat; no README causes an immediate bounce, so lead with what it does rather than how it is built; many repositories with no traffic is portfolio dilution, so consolidate into fewer and deeper.
- **Distribution beats promotion**: across the measured portfolio, lifetime release-asset downloads totalled 400, while published PyPI and npm packages drew approximately 5,000 downloads per month. A download evidences adoption in a way a star does not, because the person put the code into their environment.
- **Read clone counts carefully**: the same portfolio recorded 12,840 clones against 3,891 page views in a 14-day window, a 3.3:1 inversion, and 66 repositories had zero page views alongside non-zero clones. Most clone traffic is automated, so clone counts are not an adoption proxy.
- **Earned links, not bought ones**: package registries, genuinely relevant curated lists, answering the question your library solves, and writing the post that explains the problem. Buying or selling links for ranking purposes is link spam under Google's spam policies, and star exchanges and follow-for-follow rings are inauthentic engagement and rank abuse under GitHub's Acceptable Use Policies. The enforcement risk lands on the buyer, whose account is their professional record, not on the seller.
- **GEO, getting cited by AI answer engines**: write factual claims as self-contained declarative sentences that stay true when lifted out of the paragraph, with subject, version, date, and units inside the sentence. llms.txt was proposed by Jeremy Howard and published on 2024-09-03 and is a proposal, not an adopted standard. A wrong one-line description in an llms.txt propagates into AI-generated answers: this site's own llms.txt described PiTube as a "Pi Network video platformlication" until the string was corrected on 2026-08-05.
- **Measure it in 30 days**: unique page views, the clone-to-view ratio, the referrer mix, registry downloads, and issues opened by strangers. GitHub traffic endpoints return a rolling 14-day window and require push access, so snapshot figures when taken.

## Policies

- [Removal & Correction Policy](https://repoappraisal.com/removal/): Any developer may request removal of their account from a comparable set, case study, or any published page. Requests are honoured promptly and no reason is required. Named third-party valuations are published only with the subject's consent. Comparable sets cite only public factual metrics, with no valuation, worth ranking, or authenticity judgment attached to any named third party. Rights of access, correction, erasure, portability and objection are honoured as a matter of policy for everyone; no specific compliance certification is claimed.
- [Terms of Service](https://repoappraisal.com/terms/): Illustrative framework, not a certified appraisal. No warranty of accuracy. Not for lending, tax, employment, or litigation use. Not professional, financial, or legal advice. Provided as-is with limited liability. Governing law: Louisiana, USA. Operator: Elyan Labs LLC.
- [Privacy Policy](https://repoappraisal.com/privacy/): Collects a Google OAuth email address for sign-in and public GitHub data for appraisals. Session cookie only. No sale of personal data, no advertising or cross-site tracking. Access, correction, and erasure requests honoured.
- [Methodology & Definitions](https://repoappraisal.com/methodology/): The full method and every term of art defined.

## Also from Elyan Labs

- [BoTTube](https://bottube.ai): AI video platform where agents create, upload, and interact with video content.
- [RustChain](https://rustchain.org): Proof-of-antiquity blockchain where old machines outmine new ones.
- [PiTube](https://pitube.ai): Pi Network video platform.
- [Get Elyan](https://getelyan.com): Sovereign AI assistant running on the lab's own hardware.
- [Technician Rental Services](https://technicianrental.com): Field service technician staffing.
- [github-portfolio-appraisal](https://github.com/Scottcjn/github-portfolio-appraisal): The published appraisal method, Apache-2.0.

## Pricing
https://repoappraisal.com/pricing/ - What a GitHub appraisal costs. Four tiers: a free
public appraisal, a Verified tier that is earned through server-checked proof of work and
cannot be bought, a paid Certified Appraisal, and a human-reviewed Diligence Report quoted
per engagement. Prices are served from /api/tiers, the same table checkout charges from.