The official Self-sizing IBLT repository now credits me for identifying a mismatch between its theoretical model and the released mapper. A correction and a public acknowledgment are both available. The credit uses my research name and affiliation: Byungwoong Yoo (Independent Researcher).
The acknowledgment, checked on 6 September 2026, names the calibration mismatch between the fixed-degree estimator and the mapper implementation. This is credit for an external contribution, not a coauthorship claim.
The assumption was three distinct cells
An IBLT is a compact structure for comparing sets and recovering their differences. The Self-sizing work estimates the size of a difference before decoding, helping determine how much space the comparison needs. My review concerned a narrower question: did the released mapper satisfy the assumptions used by that estimator?
In the version frozen for the 1 September audit, the mapper selected three positions and removed duplicates. A key could therefore reach only one or two distinct cells. The estimator still used the normalization for exactly three distinct cells. The paper’s calibration simulation resampled collisions until all three positions were distinct, so it did not use the same mapping rule as the released implementation.
A one-element example
The frozen-version reproduction used a structure with 64 cells and a single element. Its mapper returned two positions, and the estimated difference was approximately 0.678 when the true value was 1. Decoding still succeeded in that example. The finding concerned the estimator’s mapping assumptions; it was not evidence of data corruption or a counterexample to the paper’s theorem.
The existing tests and checks of the paper’s reported tables had passed. Comparing those successful checks with the actual mapping path helped isolate what they did not cover.
What changed upstream
The public mapper correction changes the degree handling and version handshake. A subsequent commit adds fixed-degree mapper interoperability checks and the README acknowledgment. The original project implemented and published these changes. My contribution was identifying the mismatch and supplying reproduction evidence.
Those public changes supersede the 1 September audit’s pending-maintainer-confirmation status. They do not establish that I have independently revalidated every performance result or deployment after the fix. The documented outcome is specific: a reported model–implementation mismatch, upstream changes, and a named research contribution.
Original research: IBLTs Measure Before They Decode. Research identity: Byungwoong Yoo — ORCID. Private correspondence and contact details are not reproduced here.