Lisbon Finals is averaging around 330 flights a day now. Over 41,000 logged since May 3. That number used to feel abstract. This week it stopped being abstract, because 330 chances a day to measure something correctly is also 330 chances a day to measure something wrong.
We caught a few of the wrong ones. A sensor reading spikes for a fraction of a second (an electrical hiccup, not a jet), and if nothing catches it, that single bad number can sit in the data forever, or worse, become "the loudest flight ever recorded" on the homepage. It happened. We found it, traced it, and fixed the pages that were showing it.
The fix that matters more than the correction itself: since we started keeping a continuous ambient sound stream a few days ago (originally built for a data-sharing project with an Italian researcher who runs his own sensor network and already publishes on this, more on that when it's real), every new peak now gets checked against that stream before it's trusted. Not "does it look plausible", but "is there actual evidence around it". No evidence, no free pass: flagged, not deleted.
None of this is about doubting what we've measured so far. It's the opposite. The bigger this gets, the more it deserves real scrutiny instead of blind trust, ours included. That's the deal as Lisbon Finals grows: more flights, more data quality checks, more redundancy, and eventually more stations watching from other rooftops. Same goal as day one: give you the most honest numbers we can, and say so clearly when we're not 100% sure yet.