How Chinese Delivery Giants Are Making Their Algorithms Transparent

Making algorithms more transparent is becoming a new business ethic for Chinese internet companies.

On November 5, Meituan disclosed its food‑delivery order‑allocation algorithm, detailing how it assigns over 40 million daily orders to millions of riders based on location, time availability, current workload, added distance, and rider experience.

The platform claims this logic selects riders with more spare time and efficient routes, allowing them to earn higher income under reasonable workload.

Riders can refuse or reassign orders when they deem the system’s dispatch unreasonable, enabling decisions that may be more reasonable than the algorithm.

Earlier this year Meituan also revealed its delivery‑time estimation algorithm, which chooses the longest of four models rather than simply dividing distance by speed.

Other major internet firms are following suit: Weibo published its hot‑search ranking rules, and Didi disclosed its dispatch algorithm and commission rules, aiming to break the black‑box perception.

A well‑designed algorithm can respond in milliseconds to massive concurrent demand, underpinning the scalability of internet services.

After decades of evolution, algorithms now permeate e‑commerce, ride‑hailing, news feeds, and gaming, but concerns grow about secrecy, public interest, and the alignment of commercial and social value.

China’s Personal Information Protection Law, effective this month, mandates transparency and fairness in automated decision‑making.

Companies therefore aim to disclose algorithmic principles—not source code—to balance commercial secrets with the public’s right to know.

Meituan’s latest “post‑preparation dispatch” solution uses smart hardware to report when food is ready, allowing the system to schedule riders more accurately.

Riders report better planning, fewer conflicts with merchants, and reduced delivery pressure after this feature’s rollout.

Recognizing algorithmic imperfections, firms increase human intervention: Didi’s “rush‑order reward,” Weibo’s manual top placements, and similar measures aim to mitigate perceived biases.

Technical challenges remain, such as inaccurate rider locations in dense or signal‑poor areas, which can lead to unreasonable dispatches.

Prioritizing human choices over algorithmic decisions may sacrifice some efficiency but enhances fairness and multi‑stakeholder benefit.

Algorithm transparency is a prerequisite for “algorithm centering,” yet it requires broader societal participation and humanistic considerations to truly balance commercial and social values.