Learned within-list scoring and WAND pruning

Characterize whether learned or otherwise non-uniform within-list scoring can enable WAND-style pruning in misi posting-list merges without incurring an unacceptable recall loss.

Background

Misi currently attaches uniform count or inverse-document-frequency weights to postings within each list. The authors indicate that learned within-list impacts could make WAND-style pruning possible, but the resulting trade-off between reduced merge work and retrieval recall has not been mapped.

References

What remains open, in order of theoretical leverage: skipping on id-sorted layouts (measured skip rates pending), learned within-list impacts (would re-enable WAND at a recall price not yet mapped), calibrated sequential stopping, tiled batch execution (bytes amortize; operations do not), and bandwidth-rich substrates --- the GPU measurement above is a first point on that curve, and bank-level processing-in-memory DRAM is a speculative later one: under the same id-block layout the accumulate is a per-bank local histogram and a verification is an in-bank dot product, though no such device has run this workload.

— misi: a Metric Inverted Sample Index  (2608.27422 - Chavez, 27 Aug 2026) in Section 5, “Position”