WSJT v9 possibly available soon

[G0DJA]

The following is part of a message sent to the WSJT development reflector by Joe (K1JT)

"As I mentioned in a message dated June 29, I've been working recently on putting some of the lessons learned from WSJT8 into further development of WSJT7. I have now started to call the result WSJT 9.0. A number of users have been quietly testing this version in recent weeks -- in North America, Europe, and Australia -- and these tests will continue. You can get a general idea of what's in the new version by reading the "WSJT 9.0 Supplement to User's Guide" at
http://physics.princeton.edu/pulsar/K1J ... lement.pdf

My present guess is that a public release of the new version might be
made in September."

[G0DJA]

Given that the new test release has not been made available on a wider basis than a few well placed and relatively high powered stations, and also given the criticisms of people who mistrust the 'deep search' system for stations with lower power stations, I've decided that WSJT is no longer a mode that I want to support.

I've used the WSJT modes from 2M down to 4M (plus the JT65-HF program on various HF bands) and deep search has always worried me, so I have always run with that option switched off.

However, given that WSJT9 is now only being tested with stations running high power and who are well sited and well set up, I have to ask if these data modes are compatible with low power MS and ES operating?

If they are, then why is all the testing being done with high power stations? Is this a case of correlating the results with the tests that you want to carry out and then feed that into the deep search results that you want to happen in the future?

If it is not, as I wanted to believe and used to support up until now, then why are only high power and well set up stations the only ones able to take part in testing WSJT9?

[G4KLX]

Hi Dave

You are talking total nonsense.

The newer versions of WSJT were available to those with technical ability to build the software, and are involved with the development. This is not an unreasonable thing to do. The fact that they weren't widely distributed was so that new modes could be tested and feedback given without (a) the new modes becoming standards prematurely, and (b) having to support the technically challenged with the new modes.

There is absolutely nothing wrong with the deep search algorithm. It uses the well known and mathematically proven technique of correlation (AKA matched filter). The people who complain about it are those not technically capable of understanding it, and a number of very vocal but ill informed people have given it a bad name, undeservedly.

For a supposedly technical hobby, we can be very very backwards at times.

Jonathan G4KLX

[G6NJR]

Linux version or source code to move to Linux no M$ Corp rubbish here and no Jobs junk ..

[M0NKC]

There is absolutely nothing wrong with the deep search algorithm. It uses the well known and mathematically proven technique of correlation (AKA matched filter). The people who complain about it are those not technically capable of understanding it, and a number of very vocal but ill informed people have given it a bad name, undeservedly.

So the new deep search algorithm has nothing to do with what is described for example in http://www.sm2cew.com/Digital%20communications%20using%20minimal%20transfer.pdf? What a relief!

BTW the above link was one of the first few results Google produced for "wsjt "deep search"", the link to it is published (together with some other interesting reading) on http://www.sm2cew.com/jt65.html. Having read these few articles I haven't quite got the impression the JT65-non-believers are so lacking the technical capability to understand it...

On another note, are you deliberately muddying the proverbial waters in your reply or am I just not reading your article right? I can't recall the issue of maths utilised being referred to anywhere in this thread, what the posters are complaining is that the software is trying to "guess" the callsigns in a QSO (or in white noise) by using a predefined list of valid callsigns. Is your point that the latest & greatest WSJT no longer does (or should I rather say "depends on") this?

73
Kristijan
M0NKC

[M0BOP]

...and I found it a powerful and convincing argument. Moreover, I understood it quite easily.

So I expect to have a better than 50% chance of understanding a full and fair explanation about why deep search leads to valid contacts, as I understand them, i.e. that both callsigns, a report and a roger are transmitted and received in full over the air, and not by using shorthand messages.

Can anyone provide such an explanation?