MORE BOLLOCKS! / THE RECKONING!
Count it properly.
Then count the cost.
SNAPSMACK by the numbers: measured source, working scope, security practice, and what a company would likely spend to commission the same thing under conventional pre-AI software economics.
The fair question
You’re interested in running SNAPSMACK. But how do you know it’s serious software—not disposable AI slop wearing a clever website?
How do you know there’s something substantial underneath it? Fair questions. We brought numbers.
The codebase
These figures come from the Git-tracked repository on August 12, 2026. CLOC 2.10 classified source, comments, and blank lines separately. Dependencies, generated files, and untracked work are excluded.
What would a company pay?
Estimated conventional commercial development cost for a company commissioning SNAPSMACK at its current scope.
This is a replacement-cost estimate, not a sale price or company valuation. It covers product planning, core and desktop development, design, federation and infrastructure, security engineering, QA, compatibility testing, documentation, release engineering, project management, contingency, and vendor overhead.
The underlying estimate is approximately 26,500-46,000 hours, or 13-23 person-years. It is based on replacement workstreams and commercial delivery costs, not a dollars-per-line formula.
The FOOD NETWORK problem
FOOD NETWORK once served people who wanted to cook. A larger audience preferred watching food-adjacent entertainment, so instruction gave way to contests and challenges. Photography software followed a similar path: the larger market wants feeds, video, engagement, and performance. The smaller audience still wants to practise photography, publish photographs, and keep an archive under its own control.
Under conventional economics, SNAPSMACK sits in the abandoned middle: unusually broad for a small volunteer FOSS project, too expensive for a company to commission sensibly, and aimed at a market too specialized to promise venture-scale returns. AI-assisted development changed that equation. It made substantial software for a limited community practical without requiring either mass-market revenue or a large unpaid contributor base.
Security work
Security engineering is included in the replacement-cost estimate. SNAPSMACK uses layered controls covering authentication, authorization, abuse prevention, signed distribution, integrity monitoring, breach containment, recovery, and federation.
The published audits document material findings, incomplete fixes, accepted trade-offs, and scheduled work, not merely successful checks. These are internal AI-assisted reviews rather than certification or independent penetration testing. Broader adversarial testing is planned before version 1.0.
Read why SNAPSMACK is built this way → Read the closed security audits →
How it was produced
SNAPSMACK uses a human-directed, AI-assisted production model. Sean McCormick supplies the photography workflow, requirements, priorities, aesthetic judgment, testing, and acceptance decisions. Claude, Gemini, and OpenAI Codex contribute implementation, analysis, review, and iteration.
The useful conclusion is not that one person replaced a team in every sense. AI changed which parts of software production required scarce specialist labour. Human responsibility for goals, judgment, verification, and consequences remained.
Counting note: The headline source figures exclude 46,739 comment lines and 40,516 blank lines. The broader CLOC-classified total, including documentation and configuration, is 262,815 lines. Counts measure repository scale, not quality or effort.