[Lowfer] 136 Dial freq for OP-OPDS 32 for tonight...,

[JD]

>>> I suppose we all "correlate pattern recognition" to some degree, even 
>>> with QRSS, or even normal speed CW.

Exactly.  The ear-brain system is extremely powerful in that regard, and the 
eye-brain system is no slouch either.

I have nothing against computers doing the same.  It's just that when I see 
comparisons of apples to oranges, or sometimes orangutans, I tend to 
question how meaningful they are. :)

If a human says "I'm pretty sure I heard (or saw) WG2XYZ last night" he can 
always add that he was only certain about some specific characters, or state 
how many repetitions it took for a positive ID, or he can post his capture 
for others to evaluate for themselves.  When we see interpreted text on a 
computer screen and have no easy access to the data on which it was based, 
it's harder to tell which parts are actual direct decodes of signal and 
which parts are statistical assumptions.

That's not about some concept of purity.  That's about the difficulty of 
evaluating claims made for the software.  There's a very human tendency to 
see one favorable item of information and extrapolate more positive results 
than are actually present...the halo effect, if you will.  Even comparisons 
of reported SNR of decodes made totally _without_ reference to a database is 
nearly impossible with the currently available software--not just because 
the noise bandwidth is exagerrated, not just because coherent and 
non-coherent noise sources (and whether they're in and out of the detection 
band) affect decoding differently, but because the reported signal and noise 
powers themselves are calculated with different assumptions between WSPR and 
OPERA.

I'm not some backwoods Luddite preaching against computer-dependent 
communication.  I'm just wishing all this "our latest mode is better than 
the others" claims were based on something that could be verified on a 
consistent, uniform basis...not just the software's own boasts about what 
miracles it's performing, behind what may be little more than smoke and 
mirrors.

It's like the wild screen size claims of early TV marketers, or the stereo 
amplifier power ratings wars, before uniform commercial standards were 
adopted.  To my way of thinking, the gold standard here is how much 
information you can send in a given amount of time, under identical 
conditions, without the software itself having prior knowledge of the 
contents.  The WOLF/WSPR tests from W1VD last winter were a good example of 
such an unbiased comparison, IMO.

John