False Decodes Tally

Jim Lill

The False Decode trap/filter seems to be running well on my 8088 page. Its adaptive whitelist scheme is effective and my daily audit has noticed only bout 1 escape a day, literally.  It is applied to all incoming spots during the scrape process, before any ranking or counting processes occur.

With it now seemingly stable, I pulled together a few metrics for 2 Sept 2022:

Out of the 3.8+ million spots, there were only 825 false WSPR decodes, an insignificant 0.02%

I then looked at the rejected false decodes and determined that all those decodes would have been new uniques.   With about 5000 uniques, that would have created about a 20% error.  Many of them are 1-2 on a single reporter but some reporters has in excess of 20 and that would have boosted their unique count by 1-2%, an impactful amount.