2017 Sudoku GP: Round 2 France

Ok, so this was the Bastien Vial-Jaime show.  Having known this I was very confident this would end up have the highest quality puzzles of the entire series - and having now solved them I think this is very like to end up beingtrue.  If there's a better author around than Bastien then I certainly don't know them - his variants are always fresh and innovative, and the classics he writes always solve beautifully.

Nevertheless this round wasn't as fun as round 1 for me, for reasons that have more to do with myself rather than the puzzles themselves.  Very often the clever tricks to the puzzles were lost on me and given the time pressure I had to do a bit of bifurcation (to resurrect my favourite euphemism) to get the puzzles out.  Also there was a Fortress Sudoku which inevitably I cocked up.  It seems to be the rarest of Fortress Sudoku in competitions that I do not break.  For what it's all worth, here are the vital stats for the round:

Enjoyment: 6/10

Classics: 9/10
Favourite Puzzle: Killer Sudoku

So onto the review.  First thing I'll note is something that I didn't notice last time, but which I did this time.  Round 1 the shade of grey was definitely too dark.  For this round we are back on track with an acceptably light grey.  The difference this makes to solving should not be underestimated!

1-5 Classic Sudoku (21, 17, 28, 20, 41 Points)

This being the first page out of the printer, I went straight to the classics.  All in all, they were very high standard, with strong distinctive visual patterns of givens that weren't constrained to the traditional 180 degree clue symmetry and in each case guided by clear and smooth solving paths that weren't trivial even for the easiest puzzles.  I only knocked off a mark here because I'd have liked to have seen another ~40 point puzzle included -- ideally in place of the Almost Palindrome?

6 Diagonal Sudoku (72 points)

A beautiful puzzle that would have been my favourite puzzle on any other round.  Nearly minimal in terms of givens, with a mere 17, they were all populated in the top half of the grid.  This meant that the solve elegantly cascaded down through the grid.

From a presentational point of view, I'm pretty much converted to having solid diagonal lines rather than dashes.

7 Figures Sudoku (66 points)

I had this and an attempt to retry the Fortress with 7 minutes to go, and failed miserably. I tweaked that the figures might actually be extra regions, and used this as a half hearted attempt to get it out in time, but I wasn't successful.  It'd be interesting to actually deduce the logic for this some other time as I can't see how you make any progress without realising this. The other note here is that the cages for this puzzle were formatted differently from the cages for Killer Sudoku, which irritated me a little.

8 Fortress Sudoku (58 points)

As with many other competition Fortress Sudoku puzzles, this was going very well until suddenly I hit a big contradiction.  My attempts to fix this in the last 7 minutes were sadly fruitless.  What was doubly annoying was that I caught myself making a false assumption early on in the solve.  What would be triply annoying would be when I do go back to solve this, it may well turn out this falsely made assumption may have nevertheless not contradicted the final solution.  Ugh.  For what it's worth this felt like it was another excellent puzzle.

9 Renban Groups Sudoku (64 points)

Perhaps I'm being a bit precious on this one, but I don't really like to see this abbreviated simply as "Renban Sudoku".  I corrected this without comment for Round 1, but there, I've made this very minor point here!

Semantics aside, this was a lovely puzzle and probably a candidate for favourite puzzle on another round.  I did have to accelerate progress with this one with a well chosen bifurcation, which admittedly does take something away from the competition experience (for me anyway).

10 Thermo Sudoku (36 points)

A striking design that solved cleanly and elegantly, another excellent puzzle.

11 Killer Sudoku (84 points)

Every now again I come across a Killer that I start raving about, and this is one of them.  I think I just really like it when the interplay between no repeats in cages and the addition really comes off - you can get a similar thing with extra regions/consecutive with Renban Groups but in my opinion there's an extra layer of solving richness with addition.  

This was perhaps slightly tarnished in competition as I had to bifurcate to get myself started which meant that again I missed something, but the bits I did get were enough to confirm to me this was a thing of real beauty.  

I think this is probably one of those puzzles that I might have umm-ed and ahh-ed about including before ultimately liking it too much to throw away, nevertheless I would have to accept this isn't necessarily the same kind of Killer that many less experienced solvers are used to.

12 Odd-Even Pairs Sudoku (56 points)

This variant has been around with various names and various presentation, and I think in general whilst I like the name here, I'm not so keen on the use of "capsules" as a presentation.  I think I'd prefer to see the shaded Renban Groups style here instead, assuming you really wanted to make the cells prominent - or else to stick with the (more) standard "Pairs" style marking of a circle between the two adjacent cells.

That all said, the capsules style presentation for this puzzle did make the puzzle's defining gimmick much easier to spot (i.e. that you had to have odd numbers in R3C3 R3C4 R4C3 and R4C4) and the puzzle again solved cleanly and elegantly after that.  Another excellent puzzle!

13 Almost Palindrome Sudoku (37 points)

I'm not hugely keen on these "liar" or "almost" constraints in general, and although this one solved nicely enough it's definitely one where the constraint was in parallel with the sudoku solving rather than then augmenting it.  A good example of where it does work is with Surplus or Deficit Sudoku.  That's just my personal preference, but I don't think this one quite reached the exceptional heights of the rest of the round.

So, 11/13 solved (still no mistakes!), but I didn't feel like I was close to finishing this one, which I think also takes away a little of the enjoyment from solving.  I suppose the proof will be in the results, but I suspect this will be perhaps quite a bruising experience for less experienced solvers, and one where experienced solvers really need to be on top form to come away happy.  

It feels a bit strange to say that perhaps there were too many "clever" puzzles for a competition, but with the clock ticking away I did have to rely on bifurcation more than I'd have hoped and I'd be interested to hear whether this was the same for other solvers.

Having initially written most of this post on Sunday, and having now seen the results, it looks like my gut feeling was pretty close to the mark.

Top 3 (preliminary results)

1. Shiyu Chen (795.7 points)
2. Tiit Vunk (761 points)
3. Jakub Ondrousek (692.2)

A new name for me is Nicholas Ling of Australia, finishing in 16th place, which would be pretty impressive going for a newcomer.

And for the Brits:

27. Tom Collyer (476)
53. Heather Golding (372)
71. Mark Goodliffe (325)

Well done Neil for having a go!