I dont understand this result.
As I see the score page, I think the target time column is totally pointless. If you delete it, then the order wont be change.

I thought before the test the target time is important. And I thought I will get points for my saved minutes from my target time. This would be logical.

We had planned the Sprint test such that a 'lot' of players should be able to complete the whole set. That also meant top players will be completing the set much earlier. That is when we thought of adding a 'target time' for each player based on LMI ratings.

While giving bonus points based on 'target — achieved' certainly came across my mind, it was not done primarily because of reasons below.
1) There was no scientific / logical way to compute the target time. Rakesh had an interesting idea, but we stuck to a simple linear formula.
2) There are several new players who were playing first time. They didn't have a target time. For example onigame, Nilz
3) There are several players who have started to play at LMI recently (just 1 or 2 tests). Target time for them is clearly not accurate. For example murat, TiiT

In future Sprint Tests, it will be definitely interesting to fine tune target time. But I'm still now sure if we can give some points based on 'target — achieved' because of points 2 and 3 above.

And should there be penalty if target < achieved?

I am inclined to believe a serious usage of target time will not be worth much if you base it only on ratings (and I have no idea what else to base it on), for the same reason winning Under-X rating sections in chess tournaments is usually a matter of being significantly underrated. Even among players with established ratings, they can be in the process of improving, or others conversely may be a little rusty if they took a brief break from these tests. Either way their rating would not be the best reflection of their skill level.

To me the current system, where the target time is a recommendation with no impact on scores or rankings, seems to serve all the intended purposes well enough. One change that couldn't hurt is to have a separate unofficial ranking list with the best performers relative to their target time.

Naturally, I see the problems with "my" bonus system.
But in this test the target time hadnt function, and I hope it will have in the future :)

I dont have better idea than making an offical and unofficial result.
You can perform on offical list if you have at least 6 non-zero results before the Sprint test. And in this case the target time can be accurate.

I know, I disqualify many solvers, but if the LMI hold a Sprint test in every half year, than this isnt a big problem :)

There will always be players who's rating does not reflect their capability, especially, those who play few tests, or those who are new-comers, like Deb mentioned. So, I would second Palmer's suggestion of having the target time just as a 'benchmark' for players, having no impact on scores. A list of best players relative to their target time would be an interesting read.

Also, regarding the frequency of these tests, it is certainly not feasible to have it once a month. Once a quarter sounds like a good regular event. It helps amateurs gain confidence and motivation throughout the year when there are simpler tests.

I am one of the four players who submitted 15 solutions but did not claim the bonus.

well, I had 14 solutions and 6 minutes to go. Had I clicked on the CLAIM BONUS button I'd earned 18 points, but if I could solve the final puzzle I 'd earned 45 points, so I tried to do it, but unfortunately I didn't succeed. I submitted an asnwer but I noticed it was wrong so I kept trying (unsuccessfully) to solve te puzzle until the last minute (panic!!), so I got no points out of my last 6 minutes. If I (or another player) had only two or three minutes after puzzle #14, by hitting the CLAIM BONUS button would earn immediately 6 or 9 points. Motris may find this situation familiar because of what happened to him in Zilina. In the past the time bonus was calculated only on the basis of last submission time. Is the CLAIM BONUS button really necessary?

Also, I noticed that a player who did not hit the CLAIM BONUS button has not been awarded not only the time bonus, but also the 25 points for completing all the puzzle correctly, I believe this is unfair.

Apart from this, the puzzle set was good (I like particularly the FORTRESS), but in my opinion the puzzles were not exactly "EASY" as defined somewhere (I had a debate with Deb on a different forum). Easier than usual, OK; the contest was solvable in the 75 minutes time, OK, (although my target seemed to me unrealistic, other players of similar level have achieved it), but if the target was to attract new players I believe only a few 5 or 10 points diagrams were simple enough for beginners.

stefano

Edited by rakesh_rai 2011-09-05 2:08 PM

Yet another bug. Updated score page. Melvy gets 225.

As far as I understand this, players with 14 correct will be awarded a bonus ONLY IF they have genuinely attempted the fifteenth puzzle also, i.e., it is not a bonus for doing 14 puzzles, but only on submitting 15 puzzles (with one mistake). So by hitting the CLAIM BONUS, you may not immediately gain 6 or 9 points automatically.

However, as you said, the interface is simpler without the extra button.

It doesn't sound to me that the system is broken in your case. You had 6 minutes to earn 45 points and chose to go for them instead of stopping early. (There was also the chance one of your 14 submissions was incorrect so you can not be sure you would even have gotten the 18 points for sure.)

When I've discussed "partial" bonus before, it has been in the context of having a reasonable but incorrect answer on the last puzzle. A blank submission would not count, and the administrators would judge whether to award it on a case-by-case basis. I believe this is the fairest way to do it, to sometimes award some points to early finishers (ie uvo) who made a mistake, but not so many that a solver would really benefit from stopping early and not trying to complete the test.

Having an end test button is also meant to allow someone who finishes early, for example 40 minutes early, to not have to spend the next 40 minutes constantly checking their work before seeing their score. It also exactly matches the WPC round/time bonus methodology of turning in your paper before you start your finish clock. That >90% of players (41 of 45) that could have claimed bonus did suggests the button generally works, particularly since the vast majority of those solvers ran into this situation for the first time.

Your last point is most important though. I agree that "claim bonus" should only affect time bonus points and not the 25 point overall bonus for being complete and correct. I do not recall if the instruction booklet was clear on this point, but I would view someone who took the last ~5 minutes to simply check their work and not turn in to be worth full marks and completion bonus whether here or on a WPC. They simply wouldn't get extra time/rank bonus.

And now reading the rest of the thread, I see that Deb has corrected the one instance where the 25 points was not rewarded. So I think this was an unintentional error and is certainly now fairly resolved.

Sorry for my very limited presence these days but I have been travelling all day yesterday and I will be again today, so here are just a few words ; I shall complete them later.

Firstly, many thanks to all the players who had a try at this test. Despite its shortness and facility, it really needed lots of work to make it fulfill our purposes as much as possible. The big participation, as well as all the positive feedbacks I got before, during and after the test are a great reward and make it largely worth the spent time.

Congratulations to Palmer, Thomas and Ko for topping the test with around 34' total time each, and to all the people who did their best to try to beat their target time.


I am glad you noticed the 4 quadrants theme on Tapa; I was slightly afraid no one would see it. For me it is one detail that makes the puzzle more than just a very easy tapa. But I will post some more comments about the puzzles themselves later.


About that : I voted for once a month, which is kinda dishonest since I would rather go for once in two months. What is sure is that I would clearly not want to see it replace the current monthly puzzle tests (which had never been the idea) and 3 different tests in a month would perhaps be too much indeed, in particular if we consider the fact that more and more sites are organizing puzzles/sudoku competitions. Anyway - my main desire is to have easier puzzle tests on a regular basis, even if it is only twice a year. The idea seems to seduce many players, which is the essential point. I now just hope we can work together to make it happen.

I abbreviate since I really need to sleep a bit.

Thank you all once again and see you quickly to pursue this discussion. I hope most of you, if not all, had fun; and before everything else, I hope newcomers found a sufficient interest in the test to consider participating in future puzzle competitions.

The idea of the poll was not to change the scheduling of LMI tests. We are certainly not going to add a 3rd monthly test, it is lot of work for us. Replacing the regular 2h test is obviously not an option.
At the same time, the polls suggests that it is interesting to have easy puzzle tests (e.g. Sampler Platter / FLIP / this test). We'll keep that in mind. It is always a difficult task to cater to all kind of solvers with much wide experience and expertise.

Regarding partial bonus :

Just to repeat what Rakesh and motris posted, partial bonus is not to encourage players to leave early, and get the time bonus. We expect players to be greedy and try to get as much points as possible. Also, the wrong puzzle will be manually checked to determine if the player has tried to solve it genuinely. It is some times difficult in a puzzle test, but I would assume that anyone with 10+ minutes bonus and a wrong answer has tried to solve the puzzle genuinely.

In the upcoming Sudoku test, we'll enforce a rule which requires player to have filled at least x correct cells of the 17 cells to be eligible for partial bonus.

As a relatively new puzzler I would love to see some of the easy puzzle tests appear more frequently. I can see your dilemma. For many of the world class puzzlers on here, that would not present much of a challenge, but it would be great occasionally to see a competition aimed at people like myself who are keen (but quite slow!) puzzle solvers. I donlt imagine I could ever compete with the top solvers however hard I practised, but I would still find it fun to enter the competitions and try to beat my own scores etc. At the moment I find I am downloading quite a few of the puzzle booklets and then not submitting my score because it is too low to bother with! I then take away the booklets and do the puzles at leiisure when I can usually complete most of them in the end I have to say I really enjoy the challenge and variety of the puzzles in these tests because you get so bored with all the regular sudokus that appear in every magazine over here! I really enjoyed the Sprint test puzzles, but was just too slow to enter my results!

Just to say I really appreciate all the hard work everyone puts in here - it's a great site. Thank you.

I'm a new puzzler and I have tried online puzzle test for the first time, with this Sprint Test. I have appreciate it so much !!
I agree with Gramar. If possible we need other session of puzzle for beginners.
I am slow too, never tried to solve puzzle with timer.
I agree with Forcolin too, about this test was not so "easy" for newbie. The time is not sufficient for who of us that have tried periodically some puzzle on magazines, considering that many of them never appeared on magazines, and the difficult is absolutely not easy for other of new puzzler that have never tried this kind of game, or done it too few time.

Anyway, I will exercise with past tests, and hope do better next time :)

Congratulation for your work !!

can anybody post the solution of horse snake (in form of image).
and how to solve the Galaxies puzzle (i am finding it hard to solve it)

Thanks for the very good and fun puzzles from a beginner point of view.

As the test was designed for beginners/medium level players, and as I fit very well this category, I've some thoughts about this test:

I think that the main thing for me is that there was to many different puzzles if you consider that the goal is to introduce new players to puzzles. Perhaps is it also related to my proper way to approach things:
As a beginner, I would prefer a tournament introducing to me perhaps 4-5 types of puzzles, with 3-4 easy/medium grids for each type, so I could really get into them, memorize rules and become comfortable with them, understand the logic of the puzzle, etc...
The amount of work to become comfortable (without speaking of being really competitive) with a puzzle type is not negligeable (even if it's not comparable with the time spent to create grids and organize such a test), especially for an inexperienced player. With 15 puzzle types, I think it's too much for someone who does not know most types of puzzle. I had not much time to prepare this tournament, so I began to solve puzzle which I already knew enough about the rules and after that, I stopped solving.

It's just a personal point of view about sprint test designed mostly for new players. As you said in the instructions booklet: "Our main objectives with this test are to catch the interest of potential new players, who may be afraid of standard tests which are usually designed for experienced puzzlers"
I was also afraid of the number of new puzzles I had to study to be happy of my performance on this test.


Fred

Horse Snake was one of my favourites of this test.

Edited by Rohan Rao 2011-09-06 9:42 PM






Attachments
----------------
HS Solution.png (9KB - 2 downloads)

Since a lot of beginners/new-comers/leisure-solvers participate at LMI and many of them do not submit answers due to 'low scores', we can possibly have more tabs on the Results Page. Apart from the complete results, we could have (say) 5 different tabs showing results of only those participants belonging to a certain category (which can be probably based on LMI ratings). So, all players with a rating below 200 can see results of all participants whose rating is below 200. Players whose rating is between 200 and 400 can see results of all participants whose rating is below 400. And so on.
(The number and section of tabs can be discussed).

I dont know how useful it would be, but it could certainly be an interesting part of the Results page. Players can compare their scores with players 'of similar level'.

I found the Galaxies puzzle pretty nontrivial myself. Here is a text-only logical walkthrough for it.

1. The two bottom right galaxies can be finished first
2. R2C3 has to be part of the galaxy with R4C4; hence R7C6 also is
3. R5C2 has to be part of the galaxy with R4C4; hence R4C7 also is
4. R2C2, R2C1 are both part of galaxy with R3C2; hence R4C2, R4C3 also are
5. The above means R5C6, R5C7 can't be part of the R4C4 galaxy; it can easily be checked only the R6C7 galaxy can contain them
6. This means R8C7, R8C8 are part of that galaxy also
7. The large galaxy in R4C4 has to contain R6C5 to reach R7C6, and symmetrically has to have R3C4 also
8. That means R7C4 can't contain R6C5, hence symmetrically can't contain R8C3
9. Then only the R9C2 galaxy can contain R8C3
10. Notice it also has to contain R9C4, so symmetrically it contains R9C1, R10C2
11. Since the bottom R10C4 galaxy can't contain R10C2, it can't contain R10C6 either, leaving only one galaxy that can (R9C6)
12. This seals off R8C6; that gives more info about the R4C4 galaxy and finishes the R6C7 galaxy
13. Note R5C8 can only be contained by the R4C9 galaxy, so it symmetrically contains R4C10
14. This boxes in a bunch of space on the bottom, so we can actually finish the R4C9 galaxy here
15. R3C10 is hard to get to; only the R2C7 galaxy can contain it, also needing R2C9, R2C10
16. Drawing in the symmetric squares, we can easily finish all of the R1C4, R1C8, R2C7 galaxies
17. This gives us a lot of boxed in space in rows 3 and 4, as well as R2C4 which the R4C4 galaxy has to contain; draw in the symmetric squares
18. As a result, the R7C2 galaxy can't reach R8C2, so R9C2 has to reach it
19. This finishes R9C2, R8C1, R10C4, R9C5
20. From R4C4 containing R2C4, we get R7C4 is a single cell, which finishes off R7C4, R7C2, R6C1, R4C1
21. The remaining three galaxies are easy

If this isn't clear enough (likely), I can try to whip up some images. Broadly speaking, the important thing to recognize is that there are relatively few galaxies that are not on the edge, so that means that there are relatively few options for which galaxies can contain which non-edge squares, and you should focus on this to move forward. For example, step 2 is motivated by seeing the top left corner has almost no one nearby except R3C2 and R4C4, and so you pick a square R3C2 can't reach, such as the one symmetrically opposite the R4C1 galaxy, and the result (R2C3) is probably forced to be a part of the larger one - it is.

Hmmm, I worked right to left in that puzzle. Mainly on the fact that cells like R3C10 and R6C910 can only belong to one galaxy, which helps you fill out the whole right top instantly.

Thank you rohan and mellowmelon..
can u post the final image of galaxies.
if you can tell where i can find few practice puzzles for galaxies and horse snake.


Edited by rakesh_rai 2011-09-07 11:08 AM

There's a ton of Galaxies puzzles at Otto Janko's site: http://www.janko.at/Raetsel/Galaxien/index.htm

I will publish a file with all solutions soon.
Regarding Horse Snake, the only such puzzles I know come from Oguz Atay Puzzle Contest 3, organized by Serkan and Gulce. I have some connection problems, so I do not have a link to give you right now.

Some very interesting comments from beginners/average solvers here. I will answer to all this later in the day.

I noticed that the number of participants in Sprint Test was a new high at LMI. Congratulations to all involved with the test !!!

For the record, here are the Top 5 "Box Office" numbers so far at LMI (excluding ISC/IPC):

Test Name	When?	Total participants	Nonzero participants
Sprint Test	September 2011	264(1)	211(1)
Japanese Puzzle Land	August 2011	258(2)	203(2)
Fivefold Sudoku Test	July 2011	221(4)	199(3)
20/10 Puzzle Decathlon	October 2010	226(3)	191(4)
Nikoli Selection II	July 2011	215(5)
Screen Test	December 2010		183(5)

Edited by rakesh_rai 2011-09-07 4:36 PM