I suppose that's what I'm exactly what I'm implying re the ratings. The WPF GP has much larger participation and comes with prizes at the end of the road. It was a fair old while since I've done the test, but I can almost certainly guarantee that I wasn't paying much attention to the LMI puzzle ratings when I put together that test.

Looking back at the NS results, what seemed to affect things more than anything else was submitting a wrong answer and losing the time bonus. Ignoring those results with a mistake, those who finished early and didn't get a marathon out were very effectively separated by the time bonus, and I don't think there can be any complaints with the qualitative results (i.e. the rankings).

But I get the feeling we are looking at this from very different perspectives. I have to say I don't fully understand the nature of your complaint. I don't know whether you are saying 1) there should have been no significantly longer and harder puzzles at all or 2) significantly longer and harder puzzles should not have been kept separate from the main section. In case 1) - fair enough. This is pretty much my point of view regarding significantly longer and harder puzzles that aren't separated from the main section of a test. In case 1) my only defence is that of experimentation.

If it's case 2) you are arguing, then I'd appreciate some clarification.

To say a solver who finished the main section wasn't rewarded when placed in the situation of having 15 minutes left in which to solve a 20 minute puzzle isn't true because of the time bonus awarded for finishing the main section. Now, the test valued a minute at ~3 points and the time bonus valued a minute at 1 point, and so there is a quantitative issue if you didn't happen to get a marathon puzzle out. This wouldn't change the rankings unless a solver behind you in the main section got more marathon puzzles out than you did. In which case, if there was freedom to choose to do the marathon puzzles then this other solver would still be finishing ahead of you in the rankings, although perhaps not by as many points.

For better or for worse I was not, and am not, so worried by this quantitative issue. My main aim with the marathon system was to give top solvers something to do after finishing the test, whilst keeping the original rankings in place and possibly allowing for small perturbations in those rankings if something exceptional happened. No-one can (or should) legislate for mistakes wiping out bonuses, so I don't think you can say that the scoring system didn't achieve what it set out to do.

To bring things back to this test then the difference here is that the issue of significantly longer and harder puzzles only affected the top 17, and did so in way so as to minimise the effect on the rankings. By not separating the significantly longer and harder puzzles, the variance issue was exposed to everyone, and had much greater effect on the rankings.

I'd also like to repeat to the Serbian authors that I very much enjoyed the puzzles on this test, and much like Fred the frustrations I had with my own performance have faded. With good reason, the issue of how to balance the relative difficulties and timings (timings are a significant factor I haven't mentioned in my discussion with motris) have been weighing on my mind greatly in the last week or so!