A Patch Deployed at 2:00 AM Randomized Every User's Friend List — You Kept the Same Number of Friends, Just Not the Same People — 400 Million Friendship Assignments Were Shuffled Overnight

Patch 3.6.0, deployed at 2:00 AM EST, contained a social graph initialization bug that randomized friend-list assignments across all active accounts. Each account retained its previous friend count but was assigned a completely different set of friends drawn from the global user pool. MetaCity confirmed the bug at 8:45 AM and announced a rollback — which requires users to manually re-confirm every friendship rather than an automatic restoration. Some users logged in this morning to find they were 'close friends' with their real-life former partners. Others discovered they had been removed from longstanding friend groups. A non-trivial number of users have stated they prefer their new list.

Patch 3.6.0 was a maintenance release — the kind of patch that exists to fix accumulated minor issues, improve performance on specific system calls, and clear technical debt that doesn't warrant a major version increment. The patch notes list eleven items. Item seven is 'social graph initialization improvements — stability optimization for large-cohort session starts.' The bug was in item seven. When the patch applied its social graph initialization optimization, it treated the friend-list assignment tables as a fresh initialization state and wrote new values into them — drawing from the global user pool, preserving the count for each account but randomizing the identities. The process completed in approximately 4 minutes at 2:04 AM. 400 million friendships were reassigned before the platform's monitoring systems flagged the anomaly at 2:11 AM. By 2:11 AM, the write operation was complete.

The experience of logging in this morning and finding a completely different set of friends is, by community description, deeply disorienting in a way that's hard to fully convey. MetaCity friend lists are not just contact books. They are social architecture — the people whose content appears in your feed, who see your Stories, who are notified when you come online, who can send you tokens directly, who appear in your 'people who care about you' summary. Waking up to find that architecture completely replaced with strangers is, for many users, the first time they've understood how much of the platform's meaning comes from those specific connections rather than from the platform itself. One user, @Galen_Arc, posted at 7:30 AM: 'I have 200 friends on this platform. Today I have 200 different friends. The number feels the same. Nothing else does.'

400 Million Wrong Friends

The relationship cases are the ones generating the most coverage. MetaCity's friend-list system includes a 'Close Friends' tier — a curated subset of an account's friend list that receives additional visibility into that user's content and activity. The random reassignment populated close-friends tiers from the global pool alongside the main friend list. A non-trivial number of users logged in to find their close-friends tier populated with real-life former partners — people they had removed, blocked, or simply drifted from, returned to a position of elevated access to their platform activity by an initialization bug. The cases that have been publicly shared range from mildly awkward to genuinely distressing. Two users have posted that they discovered a real-life ex had been assigned to their close-friends tier at the same moment they discovered a longstanding friend had been removed from it.

The manual re-confirmation rollback — MetaCity's chosen solution — has generated its own set of complications. The process works like this: the platform shows each user their original friend list and asks them to individually confirm each friendship before it's restored. This means that for every friendship MetaCity intends to restore, both parties must actively re-confirm it. For people in longstanding friend groups, this is tedious but manageable. For people who had accumulated large friend lists organically over years, it is a significant time investment. And for the friendships that one person wants to restore and the other doesn't — the cases where someone discovers their longstanding friend hasn't re-confirmed them, or where someone realizes they've been waiting for a re-confirmation that isn't coming — the rollback process is surfacing a quiet category of social information that many people had never wanted to have.

The 12% opt-out figure — users who have been offered their original friend list and declined it — is the data point that has generated the most philosophical discussion today. These are people who looked at their randomly-assigned friends, compared them to their original ones, and decided they preferred the accident. Some have posted explanations: accounts that had been gradually drifting from their actual interests, restored to a vitality they'd forgotten. Some have posted nothing and simply declined. @Orin_Fade, who has 3,400 followers and has been on MetaCity for six years, posted a thread this morning that has 2.1 million views. The opening line: 'The bug gave me better friends than I chose for myself over six years. I don't know what to do with that.' The thread does not reach a conclusion. It ends with: 'I'm keeping them.' MetaCity has confirmed that opting out of the rollback is permanent.