Articles

  1. The Open Bandit shortfall was not shown on the two confirmatory campaigns

    A pre-registered check of the Open Bandit Dataset benchmark: the published prior is not the policy that was logged, and on the two confirmatory campaigns the off-policy estimate is not shown to fall short of the on-policy click rate.

    Note ope-checks 9 min read

    Technical report, PDF, not peer-reviewed A Pre-Registered Check of the Off-Policy Evaluation Benchmark on the Open Bandit Dataset

  2. A low PBO can mean the search asked an easy question

    A PBO of 0.05 came almost entirely from the long-only versus long/short choice; within the long-only half it is 0.42, in a post hoc, exploratory analysis.

    Note backtest-lab 5 min read

  3. Locking the vault means ending a process

    JavaScript strings cannot be wiped, so Local Ledger ends the processes that held them: the page's renderer on every lock, unlock and committed password change, and on every lock the process that holds the vault. Tests check that they have exited.

    Note local-ledger 7 min read

  4. The cryptography was the part that held

    A self-assessment of an end-to-end encrypted file sharing service: none of its five High findings involved cryptography; four came from the upload lifecycle under concurrency, retries and deletes, and the fifth from cleanup when the database and the storage disagreed.

    Note vault-share 7 min read

  5. The order book was 2% of the CPU

    The matching core applied a command in 144 to 198 ns, but end to end the order book was 1.4% to 2% of the engine's CPU, and the latency gains came from a JSON key check and an fsync moved off the matching goroutine.

    Note matching-engine 6 min read