Tuesday 22 May 2012

Grids? Where We're Going, We Don't Need Grids

I'll keep it brief in the "miscellaneous crossword happenings and musings" section. Started working on my speed-solving a bit for seemingly no reason at all. I've been posting my times at Dan Feyer's blog, but it's embarrassing so you should also post your times if you're not the greatest speed-solver. Really, though, I'm hoping that posting my times and being embarrassed will help me to get better. I'll keep you posted.

Today's puzzle is the first diagramless of the Cross Nerd series. I actually rarely solve diagramless puzzles these days, truth be told, but I grew up solving the ones in the old Dell magazines during the summers at the lake (if you saw my skin, you wouldn't be surprised that I shunned the beach for puzzles as a kid). Come to think of it, I don't really remember solving any of the regular crosswords. Who made those anyway? None of the constructor bio blurbs I've read have mentioned Dell, if memory serves. I'll have to check out some of the bylines in them the next time I'm out at my parents' cabin. Anyway, I hadn't really intended to make one but I had a neat grid design in mind and I thought it might work if carefully clued. The theme is really loose in this one. Some might say inelegant, but my stated goal was to use the entries to sort of paint a picture of the thematic idea (I'll try to keep the background non-specific so as to not give anything away, but skip this paragraph if you're really spoiler-paranoid). I tried to cram a lot into the grid and imposed a number of constraints to solidify the delivery, but I'm afriad I may have shot too high and eschewed the WWPBD (What Would Patrick Berry Do?) approach a few too many times. This was certainly the most difficult grid I've tried to fill (in the end, though, there's nothing in the grid that makes me cringe - so-so stuff en masse, mind you), and I struggled right till the end with tying the thematic stuff together in the clues and making sure I had satisfied all of the constraints. I caught a few biggies that necessitated substantial rewrites, but in the end was still left with one irreparable and grievous transgression. So, I added another equally grievous one to offset it, and highlighted them in the clues, as a another nudge to the solver. Also, I thought of another layer to add near the very end, but realized that comprehensively executing it was infeasible given the time constraints, if not impossible altogether without a complete do-over. So that's kind of left dangling in the puzzle, but I don't think it will be very evident at all to the solver. Anyway, this is kind of a different one. It's hard for me to know, but as a solver I feel that I would find a good satisfying challenge in this puzzle, but I'm sure some of you will hate it. Let me know!

For tyros:

Solving a diagramless crossword is like solving a regular crossword and a jigsaw puzzle at the same time. Maybe a bit like solving one of those murder mystery jigsaws. I think I had a Sue Grafton one at one point; "P is for Poison" or something like that. You know, one of that series that has given us 26 ready-made entries such as JIS, XIS, IIS, etc. Anyway, here are some tips for working a diagramless:
  • Don't be afraid; deep knowledge of construction or common grid layouts is not required. The grids are often unusual to trip you up, anyway. You will need to know 2 things, though:
    1. The puzzle be symmetric in any number of ways, or not at all. Rotational symmetry (flip your paper 180 degrees and it looks the same) is the most common, but symmetry about either of the diagonals, the N-S axis, or the E-W axis are possible as well. Although you can't always be sure (without using the hint, at least, which is not "cheating" so don't feel bad), it's helpful to watch for it.
    2. You'll also need to understand the numbering scheme. It's easy. If a white square is below and/or to the right of either a black square or the border, it gets a number. The numbering begins with 1 in the upper-left, and proceeds consecutively from left-to-right, top-to-bottom (incrementing only on the squares that should be numbered). If you're still confused, take a look at another puzzle (there are several on this site, for instance) and the scheme should become apparent.
  • Speaking of numbering, the big gimme is the length of 1-Across. Every puzzle has a 1-Across, but the number of the second clue varies depending on the length of 1-A. This is true of every entry, in a way, but here we can know what that length is because every letter in 1-A must be numbered (because we know they all abut the border of the grid). For instance, in this puzzle the 2nd clue is 8-Across, so we know that 1-Across is 7 letters long. As a bonus, we know that the Xth letter in 1-A must be the first letter of X-Down, which will help us get a solid block of letters in place, hopefully.
  • If you do manage to get a block of letters, look at the next few across clues and try to find answers that will cross that block. Beyond 1-A you can't always be sure of the length of any entry based on the numbers alone, so you won't know which should fit, but if you do get a couple of acrosses in place and notice that there are extra acrosses that won't fit in the block, you know that there is another open space at another position on the corresponding row.
  • If you get stuck working outward from 1-A, look for pairs of across and down clues that share a number. You can be sure that they also share first letters, and any crossings are helpful at this point. If you get a bunch of these in one area, you can start connecting some of them.
  • At least in this puzzle, every row and column has at least one entry (i.e. no row or column is entirely blacked-out)
  • At least in this puzzle, the clues are fairly easy to make it fair. Don't be afraid to take your best guess at a clue!

You'll probably want to use some or all of the hints. Knowing the symmetry or lack thereof is par for the course, so check that for sure unless you're hardcore. Often, the first square is offered as well. I figured that knowing the location of the last entry might help as well, so I've given you that (note that me saying this reveals nothing of the symmetry. There is no pair of first and last entries such that you can determine the axis of symmetry (or if the grid is asymmetric) just from their first squares. Think about it. While you're at it, can you think of a case where you could determine the symmetry if you also knew the lengths of the entries? There's at least one)

Hints (highlight the white space after the title to view):

Symmetry: The grid is asymmetrical
Location of 1-Across: Top row, 18th square from the left
Location of 80-Across: Bottom row, 6th square from the left
Grid hint (mild): The grid layout is meaningful and thematic
Grid hint (severe): The grid depicts two well-known and related symbols.

More words, crossed and otherwise, next Tuesday.

Puzzle: Diagramless #1
Rating: XW-PG13

Print-out only this week, as Across Lite doesn't support diagramlesses, to my knowledge. I'll look into Crossword Solver for the next one, though, as I think it supports these wacky puzzles.
**UPDATE** as of Tuesday morning: stay tuned for the electronic version! As Dan pointed out in the comments, AL V2.0 does indeed support diagramless. I remember seeing that in their documentation as well, but I must've forgotten. I guess my real point is that Crossword Compiler doesn't export diagramless puzzles to PUZ format, but again I could be wrong. Anyway, I'll whip up the PUZ later today (probably after work) and post it for y'all. Thanks Dan!
**ANOTHER UPDATE** I've added a dl link for the PUZ file. See below.

Download the PDF here and get a blank grid here. Or, you can download the Across Lite file here. Note that you won't be able to solve it in AL without revealing the grid, but it does offer a convenient printable format. V2.0 only.

1 comment:

Dan said...

Across Lite certainly does support diagramlesses, as that's how the NYT publishes theirs. No idea how you make the .puz, but perhaps Dr. Google can help.

I like knowing the location of 1-Across (saves trial and error) but not the symmetry...