Erwin,

Since false decodes can result from various mechanisms, a precise relationship is not possible. However we can come close by looking at a day's false tally for a station.  For 22 Sep, every reporter in the top 200 had at least one false. When I was using values of -C 1000 o-4, I would typically have 1-2 false spots. Yesterday, was a very active day and with my current -C 3000 -o 4, I had 6 false spots. That data may be viewed for a current day at:

http://www.jimlill.com:8088/data/tally.false

and for previous days at

http://www.jimlill.com:8088/previous. 

The impact of false decodes on overall spot count is trivial, < .02% typically.  However, false decodes  are very impactful on accuracy of the unique transmitter count since every false appears as a unique. My data integrity effort on my "8088" page solved that data corruption with its use of an adaptive whitelist scheme and unique errors are essentially zero, perhaps 1-2 overall per day, < .04%

Jim

I have stopped my WSPR decoders as for this time I feel that the competition is to much disturbed by some reporting stations who let their system running outputing way too many false decodes.
Last few weeks we see some stations experimenting with the deep decoding and I do not object to this experimenting. What I do think to be  very disturbing is that these stations should reckognise that they are reporting lot's of false decodes. In this way competition in WSPR decoding is now more becoming an illusion with stations reporting much more than 1 or 2 false uniques. Some have more that 100 false unique decodes over several bands.
 
The WSPR system always had some minor false decodes but now we see several stations with more than 10 false uniques reported. Looking at the map of the reported uniques it's obvious that many reports are false as being middle in the ocean or in a place where no activity exists.

It would be acceptable that during such an experiment a station reports more than 10 false uniques for one or at the most two days. We now see stations reporting much more false decodes for many days after each other. In that way competition becomes an illusion no longer worth to put in some effort.

That is the moment that I do not longer want to participate in this illusion and I deccided to shutdown my WSPR RX.

I have been on listening and reporting for WSPR at least 4 years and had fun in improving my setup, mainly hardware wise, better antenna and better receiver and also better software.
For now it is enough.  Hope for better times and if so perhaps I will be back.

73's   Hans DL/PA0EHG

Please note that both my own http://jimlill.com:8088  and kb9amg  pages excise false decodes thus any ranking is accurate.

-Jim

There are other pages showing the ranking which do not correct for false decodes.
I am using : http://wspr.rocks/topspotters/topb.html

-Hans

Mail to WA2TP

Hello Thomas,

I noticed that you are cheating for more than one month in WSPR receiving and sending very many false spots. You are top score nr 1 for the whole month in the falls spots as reported at: http://jimlill.com:8088/previous/2022-10-08/tally.false In practical you have even more falls spots, not all are found at jimlill.com.

My use of the available -C values built into the WSJTX and WSPRDAEMON software are by no means any form of cheating. In fact, those values help decode poor SNR spots in my noisy environment.

They are available to use by anyone and are currently utilized by many who understand the software.

Since we have never spoken, and you are likely unaware of the construction of my site, here is some background.

At my site with 64RX channels, many RX antennas, and lots of processing power, it leads to roughly a .04% output of false decodes as proven by Jim Lill.

These false decodes are simply a product of the JT9 decoder which identifies it can be improved: this is goal for much of this work being done by others.

While these decodes are not deliberate on my part whatsoever: rather a product of the technology, your actions are not only deliberate: they are in a way direct cyberattack on WSPRNET and the entire WSPRNET community by way of using my callsign without my consent, intentionally uploading erroneous data to WSPRNET, and holding the wsprnet community hostage to this false data. This does constitute fraud.

Sadly, this is far from the comradery that Amateur Radio is known for.

What to do?

Do I contact FCC and CEPT? Or do I let this go provided that you cease and desist from this deliberate activity?

In the spirit of the hobby, I am more than willing to let this dissipate and move on without further action.

I will close by saying that I have invested a significant amount of time and money, (well over 10k US) in just the past 2 years alone, on processing and rx equipment solely for WSPR.

I contribute significantly to the community: I have provided test platforms form WSPRDAEMON development, as well as other sites to use the data from these systems such as Jim Lill and WSPRrocks. 

The most recent contribution was the use of my systems by many others, namely Gwyn Griffiths, for the analysis of spectral width and spreading effects on 4W modes, which helped uncover several bugs within the KIWI GPS integration which resulted in significant frequency drift.

 This bug was affecting ALL KIWI users and has now been fixed thanks to these efforts. This one kiwi bug fix has improved kiwi spot reporting by orders of magnitude.

All of my actions are with the focus of improving receiving capabilities at suburban sites challenged by manmade RFI.

Hello Tom and Rob,

I made a complaint about the false spots at 24 september in a way that I accept experimenting with the software and that it can lead to false spots. I truly think that should be part of our hobby. What I don't think to be acceptable is running systems for many days after each other producing false spots in a setup were we also have some kind of competition to stimulate each other to improve our setups.

After my mail in this group nothing happened, the hams causing lot's of false spots remained in this way of working. In my opinion it's causing just as much corruption in the data base as my action.

Looking at last month I see that Tom has corrupted the database with at least 3000 false unique decoded spots. This is a lot more than I caused by sending spots in 2 days.
So if you say I am corrupting the database than all reporters with more than average false spots are also corrupting the database.

I will stop my action for the reason that my point should be clear to all involved in sending much more false spots then ever meant to be in WSPR in a chase for more decodes. If you look at http://jimlill.com:8088/previous/2022-10-08/tally.false you can see there are much more reporters having way above average false spots during the whole month.
For one or two days as result of an experiment could be acceptable but not for many days or weeks after each other without doing anything about it.

I dont think that adding a remark how much good Tom did for the development of WSPRdaemon is helpfull, nor the remark that it concerns only a very few false decodes. If you look at a map of his reports you can instantly see the amounth of false spots. It's not only corrupting the database but it's also corrupting a competition.

I expect from all involved to serious look into their amount of false spots and analyze their settings if the false spots are too obvious and influencing the results of the competition for many days or weeks.
I also invested in my WSPR setup and enjoyed the work with it, that makes is also unacceptable if other try to win a competition be cheating with large numbers of false spots. We have a competition without rules. Perhaps these rules should be result of this discussion.

As said I will stop sending reports for now but I expect this discussion will lead to some changes. I will not agree in saying that 100 or more false spots per day are only a very few false spots. A normal operating wspr decoder will produce perhaps maximum 2 false spots per band. Looking at the max from Tom at 149 last month is way out of the acceptable value.

Enjoy your hobby but pse don't spoil mine also.

73's  Hans PA0EHG 

Tom,

You gave me some information on your setup, so perhaps it's nice if I do the same.

Up to the moment that I decided to switch off at 29 september I was using one Kiwi receiver  and one Perseus receiver, this last one dedicated to 80 mtr with WSJT-x for decoding.
The Kiwi was used for 7 RX channels on different bands.
Also I had a Red Pitaya receiver using a different callsign, wenn I strated using this it was performing well but since the WSPRdaemon it was giving less perfromance than the Kiwi with WSPRdaemon.

As antenna I am using a single antenna a magnetic loop home build based on a Alford Loop.

I have been working to build a second antenna for the WSPR but if things stay as they are I will not even bother to get things going. I have 2 more Kiwi's waiting to get used but also these have not been used up to now.

My WSPR RX performance on 80 mtr is good to very good with lot of spots decoded and quite often within the top 3 of unique decodes.

Hi Hans,
 
Thank you for that.

I am and always have been willing to share what I have implemented will all: it should not be a secret as that is how we can all learn and grow this hobby.

 
I have taken what I have learned from others and applied it in my own practice. 
I only started WSPR monitoring 2 years ago. Ironically what got me intrigued in enhancing my performance was out of error on how I was trying to monitor a balloon spot, which was brought to my attention by a much more experienced HAM.

From there I got involved with the group and learned quickly I had much work that I could do to enhance my RX capabilities which are strongly impeded by local noise.

Currently I am running 6, 14 Chanel KIWI SDR (All modified with BBAI to achieve 14 RX channels). 5 of them are running in 14 RX mode, while 1 is in 8 ch mode.
Out of the available kiwi RX channels which total 84, I am currently using 72 active rx channels.  I have two spare kiwi, which i utilize on a test system which can upload using WA2TP-1.


Antennas consists of:
ZS6BKW@ 20m height broadside EU
Inverted L's for 80/160 with 64 short radials (nulls due south /north respectively
30m Delta loop vertically polarized apex at 20m
Wellbrook Active loop
Reversable coaxial beverage on ground (short at only 50m in length)
20-10 Hexbeam  @ 1 meter off the ground (yes 1 meter)

all very much compromise antennas, most of which I constructed.

I had a Steppir 2e which was on a home brew telescopic mast @ 10m height, which pulled the chimney off the side of my house in a storm early this year. It is sitting on the ground waiting for the arrival of my tower this fall.


All of my antennas are all quite close together on my very small 30m x 30m lot.  There was quite a bit of interaction which required custom built cores for isolation and matching.

In addition, I have several 25KW broadcast transmitters line of sight from my QTH just  5km away, and a few ~15km away the opposite direction.  I had to custom build some filters once I found commercial ones didn't quite cut it.

In addition to all of this, my area is quite prone to lighting strikes. My detection equipment to date has recorded 24,918 lighting strikes this year alone with a 30km radius.
I had to pay special attention to my lighting protection system which you can see the details of at my QRZ page.

Prior to two years ago, I had zero experience with any operating system other than windows. 
I had to learn UBUNTU, Linux Raspbian etc. all in two short years.

I then built a robust 24 Core I9-1200k Ubuntu server running 20.04LTS to merge all RX channels.

Currently I do have one radio: a Newley acquired 7610 running wsjtx on 10,20, 40 meters being fed by an active splitter using the ground mounted hex beam. 
I am using the preamplification and notch filters available in the 7610, to this to see if the addition of preamplification may offer increased rx performance from the kiwi's  which are notoriously deaf from 20m up.
Currently the only kiwi with pre-emphasis is the well brook loop. all others are run flat.

I would like to add that as a show of good faith, and in the spirit of our HAM community, that as of a few hours ago I have reduced the -C decode value by 30% of its maximum value. We won’t know what that impact is for several days.  

The effects of reducing the -C value to 3000 will align you with many others who run that or less. I would anticipate that the false count will drop although you may still be high as you run so many receivers.

The hard thing to quantify is how many valid "deep" spots you'll miss, if any, running 3000.  I ran a lot of tests here, as did KD2OM with different C values and I was never able to correlate that value with any increase in valid spots. There  are simply too many variables at play, band condx, TX hopping variance etc..  The only method that would have any certainty would be to have multiple identical RX systems hooked to the sample antenna and take a large sample size of spots. Even that has accuracy limits as the one time I tried that, with equal C values, the A/B results were dicey on very low SNR signals.

-Jim

WA2ZKD

Tom, your place is "spacious" compared to mine. My testing was done both here, at one extreme, and at Steve's,  closer to the other. 

Wow! 
First, Chess Grandmaster Hans Niemann's cheating scandal makes international news.  
(He was suspected of being signaled moves with an undetectable wireless vibrator).
Now a cheating scandal rocks the wsprdaemon world! 
What is our world coming to?