USPC 2009

Update: Well, it turns out I was off by 3 on Window Pain (doh!), but Nick was nice enough not to penalize me for the late submission on Magic Puzzle'Rs - which means I actually ended up with a score of 95 and rank of 105 (about a 50 place drop from last year). Which actually is a bit heartening, because if I didn't make stupid mistakes, a score of 125 would put me at the 44-48 rank and being on top of things really would have netted me a top 20 finish this year. There is hope for 2010!

Well, I knew I probably wouldn't do as well as last year, given the little prep time I had this year. Overall, though, it wasn't bad. I think if I had spent an hour or two reading over the instructions so that I knew what puzzles would be nice to do and strategies to take, (and how to put in the answers) along with being in top shape, I could have scored in the 180 range.

Here's the breakdown:

1. Battleships - I finished this first and very quickly, due to a (correct) early guess on the second cruiser position. I forget my writing scheme, so I had to break out the white-out (doh!).

2. Sudoku - I wasn't really on top of my sudoku game, so I took a crack at this and stalled a few minutes in. I left it alone, came back later, made one guess and extended until I found a contradiction, and then changed it and solved the rest. Took about 20 minutes. (blargh)

3. KenKen - The instructions were somewhat confusing, but I figured out that 2 square cages *could* be multiplication or addition, and 3 square cages *must* be multiplication or addition. It should have been worded differently (ie. 2 square cages can be any operation, but there is a constraint on 3 square cages). Anyway, the singleton 2 helped by forcing the 4+6 on the 10 in the same row. 20 and 40 had to be multiplication, which forced some early numbers, and then there was a bit of casework on the bottom and top rows to get the rest.

4. Sum Thing - I guessed that the center node was the key, since it intersected with 4 pairs of lines - forcing it to be 4. I struggled a bit with placing the rest, until I realized that the 5 had to be in the only unconnected node to the center and it came around nicely.

5. Eminent D'OHmain - In retrospect, rather easy, if a bit time consuming. Probably took me around 15 minutes (after the test) to solve. (which really isn't that much better than my sudoku solve for the same number of points)

6. Switch Cheese - Really easy this time: I got all 10 differences in a few minutes using the good old cut in half, flip over and vdiff.

7. Writer's Block - I initially started looking at 3 down without significant progress and then went to 2 across and filled in everything nicely. Unfortunately I crossed off the MICHENER by accident and messed up the bottom left area. Ended up answering incorrectly. Doh...

8. Corral - I am not a fan of corral puzzles: they take me way too much time - so I left this out, spent about 20 minutes afterwards and still didn't get it. (I think I'm close, though.)

9. Coordinate Pairs - I started by looking at N and its possible pairs. Realized (3,0) was out of the question, as was (2,1) and (4,1), which left (4,2), and resulted in the solution.

10. Triangular Skyscrapers - Looked too complicated - regular skyscrapers is already pushing my point/minute rate - so I left this one alone. It took me about 30 min to get afterwards, once I was able to fix the positions of the 8 and 7's.

11. Window Pain - Halfway through counting, I realized what I was doing and sped things up a bit. I think the strategy here is to look at different sizes of squares and figure out possible positions for the upper left corner. Take the number of valid x-coordinates and multiply by the number of valid y-coordinates to get the solution. Bit annoying to add everything together.

12. Masyu - Spent 1 minute starting, and realized this was a much bigger Masyu than I had done in the past, with no real recent practice, so I left this one alone. I spent about 40 min afterward resolving everything but the middle of the right side, which probably would have been a good place to start. Still unsolved. (but I think I'm close)

13. C Note - I didn't have a good strategy for this one, so I left it alone. Afterwards, I realized you could start with the middle row, since exactly 2 of the 3 givens have to be ones digits. Some casework results in 56 for the middle of the right column, and some guessing led me to the answer. Also about 20 min.

14. Lucky Sevens - No complaints, but seemed real familiar. Actually in the same format as 2003 - 8. Unlucky Sevens - with many of the same words.

15. 2-3 Maze - I wasted about 20 min darkening unusable squares, screwed up a couple times, but eventually got the answer. Probably could have saved a bit of time, but oh well.

16. Sweet Sixteen - I have some intuition about trying to fit the middle 6 numbers, but never found a workable solution after some 20 minutes post-test.

17. Yajilin - Did a little bit, messed up - figured it was the same situation as Masyu (larger than expected and little practice), so I left it alone.

18. Wolves and Sheep in Fences - Also not a fan of fences.

19. Magic Puzzle'Rs - ARGH: spent over an hour trying to fit things in. I realized early that Merrell was key - since it had to be placed diagonally. I wasted about 20 minutes of trying to fit it in until I realized it would fit in the other diagonal location. Then wasted another 30 minutes trying to fit in the 7-letter names to no avail. Came back with about 25 minutes left and solved with 6 seconds to spare, but then spent 17 seconds to put in the answer. (Hey Nick, can you number the columns for me next time?) Overtime, but worth the penalty for the 10 extra points (20 for solving, -10 for overtime)

20. SuDOUku - Worked on it for a few minutes - screwed up - then solved afterwards in about 25 minutes, which is pretty good for the points award. DOH. The key was realizing which empty cells forced 2 of one number in a column or row, which then required no collisions in all other columns/rows. For instance, there is a forced 4 collision in column 7 or 8, which means all other columns contain exactly one 4, which can be placed in row 1 for column 6, which conveniently also means all the collisions for 4 are known (row, column, cell). The tricky part is realizing that there are 2 collisions for 7 along rows, columns, and cells.

21. Di-Agony - Haven't tried it and no real intention to, either...

22. Four Square - Spent about 10 minutes post test looking at some clues, but I've never really seen Snake before, so I left this alone.

23. Inside/Outside Corral - Remember what I said about Corral?

Results: I think 100 points. (Hopefully I counted correctly for 11.) I probably should've gotten 125 points (-15 for 7. and -10 for 19.) Again, if I had been able to prep and were in practice, probably could have also squeezed in KenKen, C Note, and SuDOUku for another 50 points. Ideally, maybe also Masyu or Yajilin, but I'm ok with trying to the potential 175 mark I think is doable for next year. (with appropriate point scaling)

Why I hate programming

There are generally two approaches to programming: The first approach (which is highly recommended) involves having a mental model of what the code will do, for which the only bugs that appear will be mistakes in typing, or algorithmic errors. These two types of errors are fairly easily distinguished, especially since typing mistakes usually pop up as errors during compiling.

The second approach involves writing code in very short sections at a time, and then fixing any bugs that may appear during compiling or running. One of the main problems is simply that bugs may be due either because a bit of code didn't do what you wanted it to do OR your algorithm was incorrect, in which case major pieces of code need to be rewritten. Another problem is simply that this method is extremely tedious. It is analogous to solving algebraic equations through trial-and-error. Sure it works, even if you don't know algebra, but ultimately we end up teaching everyone algebra.

This analogy also gets at the heart of the difficulty with the first approach: in addition to requiring would-be programmers to learn a language's syntax, they must also have a good (i.e. accurate!) mental model of how the language works. Once this model has been internalized, developing code becomes simpler and straightforward.

The problem I have with certain programming languages is that sometimes they do some things that are non-intuitive to me. Consider the following bit of code in Matlab:

A = ones(3, 5);

which creates a matrix of 1's with 3 rows and 5 columns (yes, it's silly, but bear with me.) Now, consider a slight variation on the code:

B = ones(3);

My initial mental conversation for what this code does might go something like this: "well, in the first example, we gave two arguments and got a two-dimensional matrix, so it makes sense that in this example, with only one argument, we should probably get a one-dimensional matrix. Since the first argument in the first example dictated the number of rows, B should have 3 rows, and be a column vector."

"Hah," Matlab says, "you expected a column vector, but instead, I'm going to give you a square matrix, with 3 rows and 3 columns."

To which I respond, WTF. If I wanted a 3x3 matrix, I could have just as easily called ones(3, 3); BUT, for some reason, probably historical, "people" expect ones(3) to return a square matrix rather than a vector, so that's what Matlab is going to give you. If you actually DID want a vector, you would use ones(3, 1) or ones(1, 3) depending upon your preference for rows vs. columns.

I should note that Matlab does indeed support 3-dimensional matrices:

C = ones(3, 5, 4)

which is equivalent to 4 3x5 matrices stacked on top of each other. When you need two dimensions, you don't need to specify 1 as a third argument, but when you only need one dimension, you DO need to specify 1 as a second argument. Not only is nonintuitive (to me), but it now becomes inconsistent as well.

Now let's take a look at a beautiful R example I just saw today:

x <- 1:10
length(x) # returns 10
length(x) <- 20 # now extends the length of vector x to be 20 elements

First of all, I should mention that R traditionally uses <- as an assignment operator. The introduction mentions that "In most contexts the '=' operator can be used as [an] alternative." It does NOT say that they are equivalent (or why would it say "most contexts"), but it also fails to mention cases where <- and = might work differently. To this, I am left puzzled.

The problem I have with this example is the third line. In the second line, length(x) returned 10, which we can guess intuitively returns the length of the vector x. However, the notation length(x) indicates that length() is a function, NOT a parameter for an object. The usage of a function is fundamentally different from simple value assignment, such as:

y <- 20

In the latter example, y is a variable, and thus is the target of assignment rather than something to be evaluated. A function, on the other hand, is different, because it is, well, a function. Users can write their own functions, with appropriately specified return values. Note that the user-defined functions operate in ONE direction only: arguments are specified, some stuff is done, and sometimes, a value is returned. To have that value then be the target of an assignment completely boggles the mind. In addition, it cannot be done with all functions. For instance, the following code gives an error:

x <- 1:10
sum(x) # returns 55
sum(x) <- 20 # error:
Error in sum(x) <- 20 : could not find function "sum<-"

In fact, the error does tell us something about the internal model for R: the message about "sum<-" not existing suggests that what is actually going on in the first example is that there are two DIFFERENT functions called length, and that "length(x) =" is actually syntactic sugar for some other function called "length<-".

The very nature of syntactic sugar should be to make a language easier to learn/type. However, in this case, it has only made me more confused...

USPC 2008 wrapup

Edit: score update

I was a little worried going into this year's competition, given the quantity of grid-based puzzles (black pearl, arrow ring, masyu, kuromasu, fences, corral, etc.). I had tried practicing using last year's Masyu (which I almost completed in time last year), but ended up wasting a lot of time trying to work it out.

Anyway, this year I felt a lot better on my time management. Initially, I thought maybe I shouldn't have spent so much time on KakurOh or Ampers& Crisscross without getting any solutions and instead focused on fences, but I ended up taking way too much time to complete fences after the contest was over, so I think some of the harder difficulty puzzles of that time end up eating a lot of my time.

Battleships was straightforward, as was Sudoku. I probably lost some time on Sudoku with a guess in the middle that didn't pan out, after which I had to start over, since I didn't distinguish my guesses from my solves like I normally do using pen vs. pencil.

Sum figure was pretty quick, given the quick constraints on 10 and 30. That leaves 10 as either 0 1 3 6 or 0 1 2 7. Heuristically, these should probably be placed according to the size of the sum they need to be in, which worked out well.

Arrow ring was also reasonably quick, since all my guesses turned out to be correct. I think this is one of those puzzles where my heuristic approach yields some really good results. I practiced on last year's arrow ring to great success, so I think I can try a looser strategy next time to gain back some time.

Double feature was rather nice this time, as I figured out the differences much faster.

Black Pearl was a quick solve, but I made two major screwups. First, I misinterpreted the directions for answer submission, and counted the lengths of the paths between all 1's in the sequence. The diagram doesn't differentiate between this interpretation and the actual answer format. Then I miscounted, which is unfortunate, since otherwise I might have been able to request some points back. My solution (in diagram form) is correct, though.

Masyu was much quicker this time, because of the increased number of constraints.

Nebijok was a standard puzzle type that went pretty well.

I didn't try Number Blocks until late in the contest, since usually I'm bad at the math ones, but this one was very open to a successful attack. The first puzzle, heuristically should have the 3 24's as the bottom sum and the two large products. This quickly yields most of the digits. The second puzzle also worked nicely, since the product has to be 1 x 2 x 8, and 8 is the only number that will work for the top row. The rest fell into place quickly after that.

I didn't try Kuromasu, it seemed very corral-esque to me, and I suck at corral.

Murder No. 6 was fun. It was slow going at first, but picked up speed quickly. As mentioned by others, the 15's and 14's and even the occasional 13 helped a lot.

Obviously I skipped Corral as well.

Tilted Weights I didn't bother trying as I usually suck at these, although a glance after the contest suggests that it should be quickly solved. The bottom two balances are pretty well constrained, and to balance the overall puzzle, the leftmost balance needs to be two relatively large consecutive numbers.

Distances also seemed too close to Corral for comfort.

Dot Triangles looked pretty impossible, and given Tom Snyder's reaction to it, I think skipping it was an excellent choice.

Crisscross Pairs worked out nicely. I think there are probably some faster methods (such as focusing on the rings rather than the 4th vertical word which has more constraints.)

Fences I almost got near the end. All but the bottom-right corner worked out quickly, and it was just a matter of trying to get the middle worked out, while keeping all the sheep on the same side of the fence. I ended up solving this after the contest after maybe 30 minutes due to a number of mistakes and lack of time pressure at that point.

SuDON'Tku is one I probably should have worked harder on. I only spent a few minutes, made a mistake, and decided it could be very risky to continue to try this puzzle given the number of mistakes I might make. I tried it a little bit after the contest and made more mistakes, but eventually figured out the strategy, so I think I should be pretty quick now.

I wasted a lot of time on KakurOh, only getting about 1/6 of the puzzle. There are generally fewer quick logic-solves in the beginning for these harder KakurOh's, which is something I'm not used to, I think. Either that, or I missed some vital clues.

Ampers& Crisscross is something I also should've gotten. I had an initial guess that didn't pan out, but that was probably because I was taking a more local strategy for this (which works well for me in other crisscrosses). I think the approach of working out the rings is probably the quickest, since that gives many more constraints on the connecting words and ring location.

Total Solved Points: 150 145 (Um, it's a good thing I can add correctly during the competition?)
Likely Resulting Points: 135 (-10 incorrect Black Pearl answer, -5 incorrect answer penalty) 145 (SOMEHOW, they gave me credit for Black Pearl, EVEN THOUGH I made the mistakes in submission above (and also dropped a comma apparently).

Improvement over last year: LOTS.

Mwa ha ha

I was catching up on my RSS feeds, when I found this post on Boing Boing on IQ differences between groups using different browsers on different OS'es.

So of course, I took it, knowing full well the stupidity of an online IQ test as well all its inherent biases.

Well, the result was certainly unexpected.

USPC 2007

ARGGG. I should've done better this year for the puzzle championship, but instead I did worse. I spent a little too long on 1. Battleships. It should've been a quick 5 minute solve, but ended up taking more like 10, because I missed the third 0 column. *smacks forehead* I missed 2. Diagonal Sudoku, because I did a really stupid end-game fill-in where I put 9 instead of 5. It should've been obvious that it was a 5, because everything else was solved! That makes a 20 point differential right there (15 for solve, -5 for penalty). 3. Sum Figure: Guess and check did the trick. I'm not sure of a methodical algorithm that is fast enough, but I didn't see the 0-9 until a few minutes in. Remind me to read the directions very carefully next time. 4. Count Me In: I think I had a nice method for this, which involves counting hypotenuses by enumerating along possible directions. So I counted by vertical, horizontal, positive and negative slope for the triangles, which worked. The check revealed one I missed, but I managed to get the right answer pretty quickly and am happy about the method. 5. Comic Strips should've been easy. I spent too much time worrying about reflections and rotations, when I should have looked for the key, which was the H in the last strip. It forces only three possibilities for Cho (without reflection), which then makes the solve really easy. That's another 15 points. On 6, I missed J4 for some reason, and put both cells for one of the two-cell changes instead. Not sure how they are going to grade it, could be a 9 (ignore the extra square, no penalty), 15 (include extra square, plus 5 point bonus), or 4 (extra square is wrong, 5 point penalty). 7. Circuit Maze also was quick, even if it looks daunting. From a couple minutes worth of playing around, it is clear that one can work backwards to eliminate dead ends. Meanwhile, using different color pens for the different layers, one can draw some of the necessary segments where there are only two elevators per path segment in one layer. That's another 15 points. 8. All About Rio was pretty straightforward. I accidentally had Maia placed incorrectly near the end, but realized it was probably a mistake and corrected. However, I spent time looking for the rest of the phrases to check, which may have been unnecessary. 9. Round Trip, I finished 5 minutes over, so I didn't submit an answer, but I stopped solving in the middle and returned to it later to finish. I think some more experience with this format and 10. Arrow Ring is going to improve my times on those puzzle types dramatically. So that's 10 points, and possibly 10-15 minutes shaved off if I were better at those types. 11. Tiling, is a piece of cake, and doesn't require cutting out the pieces either. That should have been another easy 15 points. 12. Masyu, I didn't finish, though I spent about 5-10 minutes on it. Wikipedia's tips are rather basic, and I self-discovered some more advanced constraints that would've been helpful if I had known them earlier. Still haven't solved this yet, though. 13. Fence Posts, and 14. Fridge Magnets are both puzzle types I'm bad at, so no solve for those two. 15. Gull from Ipanema: Perhaps I would be faster if I were more methodical. I'm thinking perhaps that a fast way to do these is to use a giant paper cutter to make individual squares for each and then group by some criterion (number and distribution of objects) to find pairs fast, but the labels wouldn't be there, so it might take longer to backtrack to find the coordinates to submit. I am not a big fan of the puzzle type for 16. Corral either, but this may just take some work. 17. Multiplicity looked daunting, so I didn't do it, but it works out rather easily if you enumerate the possibilities for each digit, realize that the 0 forces 5 or 6, and then the rest follows pretty naturally. I glanced briefly at 18. Double Murder for a bit, and realized it would've been hard to get started, much less finish in a short amount of time, so I skipped it. I worked on 19. No Parking twice, and got it the second time with 2 minutes before the end. I think the key here was the limited placement possibilities for "on one's honor". Also, I had the beginnings mostly correct the first time I approached the problem, so I probably would've gotten it marginally quicker if I hadn't switched to other puzzles. Other key phrases were "unobnoxious", "mano a mano", and "economic" for me. Once I filled those in a few of the cross phrases, I knew I was on the right track. 20. ORu Kakuro looked daunting, but I managed to get the bottom-right corner. I then tried the top-right corner to no luck, so I gave up, thinking this puzzle would take too long.

Score:
94, 99, or 105 depending on grading for 6.

Lost points that I might have gotten if I hadn't messed up or wasted time:
105

going to work on 18 and 20 now.

EDIT: It turns out that I forgot to submit when I had answers for 3. and 19., resulting in a score 40 points lower than I had thought. *ouch*. The grading for 6 was middle of the road, as I expected, giving me a total of 59 points, a significant decrease from 2006. However, overall, I felt like I performed better, with a number of puzzles very close to completion this year, as opposed to last year, where I was hoping to get lucky with guess and check.