papers/zoo-experience-ledger/sections/08-security-analysis.tex

\section{Security Analysis}


\subsection{Threat Model}

We consider adversaries who may:
\begin{enumerate}
    \item Tamper with stored experiences
    \item Return incorrect retrieval results
    \item Attempt Sybil attacks on the network
    \item Perform denial-of-service attacks
\end{enumerate}

\subsection{Security Properties}

\begin{theorem}[Tamper Evidence]
Any modification to an experience $e$ produces a different CID with probability $1 - 2^{-256}$.
\end{theorem}

\begin{theorem}[Retrieval Integrity]
With Merkle proofs, a client can detect server-returned experiences not in the committed library with probability $1 - 2^{-256}$.
\end{theorem}

\begin{theorem}[Availability]
With $n$ nodes pinning the library and at most $f < n/2$ Byzantine failures, the library remains available.
\end{theorem}