Today’s blog is more of an incentive commentary—two intriguing thoughts about Didi’s (the China ride-hailing app) matching mechanism.
fake impatience
First, it appears that the matching algorithm has a peculiar bias against patient customers. If a someone is willing to wait—say, for a car that’s more than 15 minutes away—the system seems to interpret this as a sign of infinite patience and begins assigning worse matches over time. In other words, the algorithm rewards impatience. So, as an individual, the optimal strategy is to cancel immediately if assigned a far-off car, signaling your “impatience.” Over the long run, the system does “learn” you’re high-maintenance and start offering better matches.
Who knew feigned impatience could be such an asset…
’newline'
Second, let’s talk about peak hours—like when a concert ends and hordes of people flood the app, creating a virtual stampede for rides. In such situations, you might find yourself stuck behind 50+ requests, facing a wait time of over half an hour. But here’s the hack: simply walk one block away from the concert hall and call your ride from, say, a quiet café or an unassuming crossroads. Voila, you’re suddenly at the front of the line! It’s almost like magic—though it’s really just exploiting how the system prioritizes locations.
What’s particularly amusing about these “nontruthful” strategies is how contagious they are. Once one person figures them out, the entire crowd catches on faster than a viral TikTok trend. Soon, everyone’s gaming the system, and chaos ensues. It’s reminiscent of the famous case of British birds learning to peck open milk bottle caps—a behavior that spread like wildfire.
I did quick search but none of the mainstream media seem to mention these little algorithmic loopholes yet. But if Didi’s algorithm designers don’t fix them soon, I have no doubt they’ll surface in no time. After all, when a system incentivizes strategic misbehavior, it’s only a matter of time before everyone becomes a master tactician.
Cheers!