While it's difficult to exactly quantify the number of these, as I mirror the wsprnet database (via wsprdaemon.org), I've been able to dig in and examine them. Here are a few of my observations:

• they're more false decodes than I expected
• I have a script that will access qrz.com to check call signs but with the spot volumes we now have that seems impractical, even if I keep a local database of already validated call signs
• I have been examining the reporters who have the high unique counts and digging into what looks like a suspicious call sign
  
• From those examinations I found that my "balloon filter" for my "8088 box" is already adequate but that there is a sizeable amount of just bad decodes on some stations
  
• I am slowly building a bad call sign filter but that is lengthy and tedious and can disrupt the normal 8088 pages as it has to be tested with live data
• Every bad call sign that is not trapped, results in a bad unique which corrupts the integrity of the ranking system
• Other ranking and analysis sites have similar issues with filtering the bad stuff out
• When we configure our systems, we need to be careful that the "quest for a few more spots" doesn't result in bad decodes and thus bogus uniques

-Jim

WA2ZKD

Attached is a partial list of the potentially bogus uniques I see. Comments welcome 

On 8/7/22 10:35, Jim Lill wrote:

While it's difficult to exactly quantify the number of these, as I mirror the wsprnet database (via wsprdaemon.org), I've been able to dig in and examine them. Here are a few of my observations:

• they're more false decodes than I expected
• I have a script that will access qrz.com to check call signs but with the spot volumes we now have that seems impractical, even if I keep a local database of already validated call signs
• I have been examining the reporters who have the high unique counts and digging into what looks like a suspicious call sign
  
• From those examinations I found that my "balloon filter" for my "8088 box" is already adequate but that there is a sizeable amount of just bad decodes on some stations
  
• I am slowly building a bad call sign filter but that is lengthy and tedious and can disrupt the normal 8088 pages as it has to be tested with live data
• Every bad call sign that is not trapped, results in a bad unique which corrupts the integrity of the ranking system
• Other ranking and analysis sites have similar issues with filtering the bad stuff out
• When we configure our systems, we need to be careful that the "quest for a few more spots" doesn't result in bad decodes and thus bogus uniques

-Jim

WA2ZKD

3D2UR is the University of the South Pacific club call, which is used for its WSPR beacon run by Antoine (3D2AG).  So that one's that isn't a bad decode.

EH

On 8/7/22 11:11, Jim Lill wrote:

Attached is a partial list of the potentially bogus uniques I see. Comments welcome 

On 8/7/22 10:35, Jim Lill wrote:

While it's difficult to exactly quantify the number of these, as I mirror the wsprnet database (via wsprdaemon.org), I've been able to dig in and examine them. Here are a few of my observations:

• they're more false decodes than I expected
• I have a script that will access qrz.com to check call signs but with the spot volumes we now have that seems impractical, even if I keep a local database of already validated call signs
• I have been examining the reporters who have the high unique counts and digging into what looks like a suspicious call sign
  
• From those examinations I found that my "balloon filter" for my "8088 box" is already adequate but that there is a sizeable amount of just bad decodes on some stations
  
• I am slowly building a bad call sign filter but that is lengthy and tedious and can disrupt the normal 8088 pages as it has to be tested with live data
• Every bad call sign that is not trapped, results in a bad unique which corrupts the integrity of the ranking system
• Other ranking and analysis sites have similar issues with filtering the bad stuff out
• When we configure our systems, we need to be careful that the "quest for a few more spots" doesn't result in bad decodes and thus bogus uniques

-Jim

WA2ZKD