HF Propagation predictions - is there any point?

Most of us have too little time to waste.  As a result, fine people, many with impeccable qualifications, attempt to make the most of our precious time by issuing propagation predictions.

But are these predictions relevant to a world where digimodes provide effective operation way beyond the noise level?

On the ITU-based prediction sites out there on the internet, you can choosea 'SNR' (signal-to-noise ratio) output.  But this effectively has a cut-off of 0dB SNR.  I regularly operate JT65A down into the lower minus 20s dB, WSPR down to around -30dB, sometimes deeper, and FT8 down to around -25dB.  Even CW 'by ear' gets us down to around -18dB.

To underline the argument, -30dB SNR is a signal one thousand times weaker than the noise level (as specified for the rather odd, ham radio way of doing things, a 2500Hz bandwidth - see here for a useful explanation).

So what is the point of continuing to use 0dB as a lower marker, with its implication that there is no point in operating when the prediction yields that result?  I have started to think, perhaps wrongly, that predictors aren't keeping up with digimode developments, and are continuing to issue predictions for the SSB operator. 

Steve Nichols' blogsite currently has a similar sort of prediction.  It, like Gwyn Williams' predictions for May 2018 (see RadCom), strongly suggests that there is nothing much to be found at 12m, even at very moderate distances.

This isn't the reality.  My logbook has perhaps a couple of hundred 12m FT8 QSOs so far for May, with especially good days on the weekend of 12-13th May out to Bolivia (at +01 dB) and ZP9 (+9dB).  OK, I'm not using a dipole with a (unrealistic) 0dBi gain, but perhaps that itself is too restrictive a model.  You can alter antenna and noise settings with online models.  But printed monthly summary forecasts are pretty much fixed.

Even if dipoles for 24MHz had been in operation at both ends, and we take the ZP9 signal as an example, there was still around 30dB 'spare' for a potential QSO, at around the limiting SNR of -25dB, with FT8.  JT65 or JT9 could have offered even deeper plunges into the noise.

Now, to be fair to both men, their predictions make specific assumptions on mode of propagation and gain of antennas at both ends.  So it would be unjust to say that their predictions are wrong simply because a good day of multi-Es hops in fact allowed working distant regions for hours on end using a 3-element Yagi.

And there's the rub.  Is there any point in presenting predictions based solely on F-layer propagation when the summer months are well known to bring Es conditions, for example? 

There is also the fact that ham operators are quite sheep-like in their conduct.  The majority follow the DX cluster or predictions like those under discussion.  Even if the intention is not to say 'don't bother operating when the prediction is simply a blank line', the effect will tend to be that operators will receive and act on the prediction in precisely that way.  You only have to see what happens on 12m to prove that.  Take an empty band.  Call CQ and get a random response.  Announce it on the cluster.  Within seconds, the band will be full of signals.  Hams are simply waiting for others to do the work for them, when it comes to probing empty bands.

Another criticism I have is that propagation can, and often is, very different from various regions of the UK, further limiting the usefulness of predictions.  A station in southern England will often have very different conditions to those in the north of Scotland, or, indeed, in the north of Wales.

Overall, even where the predictions might be generally correct for their given set of assumptions, they are very rarely, in my experience, particularly reliable indicators of which regions of the planet can and will be working each other in practice.

That is why I never even bother looking at predictions.  If I did, I would miss out on a lot of QSOs for thinking it would be a waste of time in trying.