Topic
How engagement-prediction models select and order content, and how interaction signals are weighted.
European Commission findings of July 10, 2026 say Facebook and Instagram may breach the DSA through infinite scroll, autoplay and personalised recommendations. That adds interface features to ranking logic as a legal question for the largest platforms.
Reality
ByteDance's SequenceO1 model, now serving full traffic on Douyin and Douyin Lite, compresses each user's 100,000-interaction history into one cached sketch. Every candidate video is scored against that sketch, so request cost on a cache hit stays flat as history grows.
Reality
A federal jury in Oakland is hearing that Meta built addictive products for children. After a $6m bellwether loss, retention design now sits in the same risk register as privacy.
Reality
A Fast Company column argues that most deployed corporate AI stores what happened without learning what its own decisions caused, and it offers a check on whether your accumulated expertise would outlive the model you swap out tomorrow.
Reality
An InfoQ pattern piece puts consent, fatigue, channel sensitivity and cost inside the ranking path, with five trust actions that adjust a candidate's score and a response payload naming the tier and the rules behind it.
Reality
A dev.to post lays out a four-layer recommender and a 100 millisecond p99 budget with no slot for a generative call in the hot path. The figures are its author's own allocation, not a measurement from a running system.
Reality
The model writing each page's recommendations only ever saw the candidates the code handed it, and when similarity matching found nothing that list was the most-rated games in the catalog. A probability model now scores the candidates first.
Reality
Australia's draft Digital Duty of Care Bill threatens fines up to A$109 million if platforms fail to offer the choice, and it defines the alternative as friends and follows, so whether the opt-out holds depends on how good that feed turns out to be.
Reality
A cosine, a count of resolved threads and a ladder-level delta cannot be averaged. So the fusion reads rank only, and a hard capacity gate decides the rest.
Reality
GenRec, a fine-tuned open-weight model that reads watch history as text, beat Netflix's hand-built pipeline offline and in a live A/B test. The gains are small; the cost story is the interesting one.
Reality
Researchers report that X's For You algorithm promotes value-incongruent posts because replies signal disagreement, and that the mismatch is over four times larger for Democrats.
Reality