#50 · Why YouTube views got stuck at 301
Every view is an event, and bots can fake them. YouTube froze public counts at 301 while it verified views, until 2015. Eventual consistency and sharded counters.
Ek line yaad rakho: counter thoda peeche, par aakhir mein sahi.
The 301 freeze
For years, a fast-growing video's public view count would stick at 301 while YouTube verified that the views came from real people, then jump. On 5 Aug 2015 YouTube announced it was "saying goodbye to 301+": views it is confident are real now count as they happen, and the rest keep being reviewed (Fortune, 6 Aug 2015).
Why counting views is hard
- Every view is an event in a huge stream; some are bots, refresh loops or fraud.
- Validating takes time, so the displayed count can lag the true count.
- A single counter row would be a write hotspot for a viral video.
The interview answer
- Eventual consistency: accept that the shown count lags; it converges.
- Sharded counters: split one counter into N shards, increment a random shard, sum on read (or cache the sum).
- Pipeline: events → stream (Kafka) → dedup/fraud filter → windowed aggregation → counter store → cache.
- Batch re-counts later can correct numbers (that's why counts sometimes drop).
Words to say: event stream, idempotent/dedup, fraud filtering, eventual consistency, sharded counters.
Numbers after the jump and shard values in the reel are illustrations.
Go deeper
Finished reading?
Mark it done to track your progress in Reels.
#38 · CrowdStrike outage: one file, 8.5M PCs
Case study of the 19 July 2024 CrowdStrike outage — a 21-vs-20 field mismatch, an out-of-bounds read, a boot loop, and why canary rollouts exist.
#49 · How Find My Device finds an offline phone
Your offline phone whispers a Bluetooth ID; any nearby Android phone relays its own location, end-to-end encrypted. A crowdsourced network with privacy built in.