| E.T. Club | Naivoj
Kwon-Tom Addict
Puzzles: 314
Best Total: 33m 50s | Posted - 2008.06.13 08:12:33
Any being solving user beast #73, #88 or #99 (posted in Page 2 of "User Beast Puzzles"), without computer program assistance, is invited to post here to join.
- Beware cheaters, you will be abducted!
Please provide an approximate active solving time (time between session doesn't matter), the number of sessions and the name of the star where you are coming from.
The first two aliens are Pqg and MondSemmel:
- Please post your active time, #sessions and galaxy/star/planet names.
Last edited by Naivoj - 2008.08.04 01:40:13 | MondSemmel
Kwon-Tom Obsessive
Puzzles: 6159
Best Total: 7m 47s | Posted - 2008.06.13 14:29:24
As I said, I haven't solved it yet. I'll post here as soon as I'm done with it. | jasonharper
Kwon-Tom Obsessive
Puzzles: 1349
Best Total: 25m 18s | Posted - 2008.06.14 22:56:19
Solved - in 1 day, 14 hours, 18 minutes, and 37 seconds. I probably slept for a while in there...
I come from a rather unremarkable star at 179°56'39.4" longitude, +0°2'46.2" latitude, 7940 parsecs from the galactic center.
can I haz my life back now, plz? | tobiwan
Kwon-Tom Obsessive
Puzzles: 6228
Best Total: 13m 30s | Posted - 2008.06.15 21:17:15
Oh dear.
LOLcat AND Kwontomloop? I'm not sure you can have it back you know. | MondSemmel
Kwon-Tom Obsessive
Puzzles: 6159
Best Total: 7m 47s | Posted - 2008.06.22 13:39:41
I've finally solved the puzzle, too. When I try to solve puzzles above a certain difficulty level, I become very self-conscious, thinking I might have made a mistake somewhere - and even a single mistake can obviously doom the entire puzzle.
I honestly have no idea how much time I needed to solve this puzzle. I solve puzzles using Loopy, saving before I explore a new part in the puzzle, using saving&loading for trial&error, etc. I used more saves for this puzzle than for any other, but that was probably mainly because I was afraid of making mistakes. That notwithstanding, this puzzle is easily one of the hardest puzzles posted, but its difficulty doesn't extend to the whole puzzle - it's a puzzle of 40x30, but the leftmost 10x30 can be solved trivially.
I still think the puzzle is not as hard as it could be, simply because of that huge part of the puzzle.
I might have needed a lot longer if the highlander pattern for the rightmost part hadn't been mentioned before.
I still have all the saves I used when solving, so I could post them if anybody is interested. If you don't want solutions to the puzzle posted here so far, I'll post them sometime in the future. Still, all of the more advanced deductions I used involved bigger parts of the puzzle and then mostly counting lines (i.e. the basic 'there must be an even number of lines going into every part of the puzzle' - applying this principle makes trial & error a lot more effective) or highlander deductions (e.g., an actual example, there were 6 lines (A, B, C, D, E, F) going into a section on the right border which was otherwise completely closed off - when connecting B and C, there was a highlander pattern involving the rest, giving two solutions for BC, AD, EF and two solutions for BC, AF, DE. Therefore, B cannot be connected to C. There was only one unique solution for this whole part, making this the correct solution according to the highlander argument.
Well, having said all that, I still used lots of trial & error and spent a lot of time on this puzzle. I think from now on I'll try to record my solving times for your beasts, but sadly, I don't have them for this puzzle. If I had to guess (and I'm horrible at guessing), I'd say anywhere between 5 and 10 hours? Well, in any case, I can post what's left of my solving process in the form of lots of saves, so if anyone is interested, just tell me and I'll upload my 25 saves (of which most were necessary and a few only exist because I'm obsessive about saving literally all the time  ) to some file hoster. | astrokath
Kwon-Tom Obsessive
Puzzles: 3258
Best Total: 13m 42s | Posted - 2008.06.22 16:09:21
I've tried and failed this puzzle twice.
My pregnant brain clearly isn't up to it! | Zyntax
Kwon-Tom Obsessive
Puzzles: 6534
Best Total: 13m 6s | Posted - 2008.06.22 16:42:33
I can't seem to find this one. Maybe I'm in the wrong area, but I'm looking in Archives/User Puzzles. | astrokath
Kwon-Tom Obsessive
Puzzles: 3258
Best Total: 13m 42s | Posted - 2008.06.22 17:37:03
Quote: Originally Posted by zyntax I can't seem to find this one. Maybe I'm in the wrong area, but I'm looking in Archives/User Puzzles. |
You want to look in this thread:
http://www.kwontomloop.com/forum.php?a=topic&topic_id=292&pg=1 | astrokath
Kwon-Tom Obsessive
Puzzles: 3258
Best Total: 13m 42s | Posted - 2008.06.23 14:57:25
Third time lucky!  | Naivoj
Kwon-Tom Addict
Puzzles: 314
Best Total: 33m 50s | Posted - 2008.06.23 15:24:53
Quote: Originally Posted by mondsemmel I've finally solved the puzzle, too. ... If I had to guess (and I'm horrible at guessing), I'd say anywhere between 5 and 10 hours? ... |
You should probably tell from which star you're from (as a safety), since they only abduct humans.
Btw Astrokath, Nobody is really surprise about your E.T. origin. 
Last edited by Naivoj - 2008.06.23 15:31:12 | pqg
Kwon-Tom Obsessive
Puzzles: 6384
Best Total: 15m 37s | Posted - 2008.07.07 16:31:33
Just finished #88 in 5h. Sorry Naivoj, but I thought it was easier than #73... (mainly because it broke down into smaller sections) | jamin
Kwon-Tom Obsessive
Puzzles: 1233
Best Total: 27m 9s | Posted - 2008.07.08 08:58:32
After spending 10 hours getting a good way through #73 (I've been away for 2 months and am really rusty) using the flash version of kwontom loop, I went to do a weekend puzzle and realised that I needed to turn off flash (I like my right-button crosses). I instinctively went back to the main screen by saving the currrent game and lost #73 as a consequence.
Bleah.
Last edited by jamin - 2008.07.08 08:59:31 | MondSemmel
Kwon-Tom Obsessive
Puzzles: 6159
Best Total: 7m 47s | Posted - 2008.07.08 12:12:09
Quote: Originally Posted by jamin After spending 10 hours getting a good way through #73 (I've been away for 2 months and am really rusty) using the flash version of kwontom loop, I went to do a weekend puzzle and realised that I needed to turn off flash (I like my right-button crosses). I instinctively went back to the main screen by saving the currrent game and lost #73 as a consequence. Bleah. |
If you want to, you could try using one of my earlier savegames in the #73 thread. Download the Loopy application and go through my savegames until you find one that corresponds somewhat well to the situation you had before you lost your progress (if there is one - different people solve the same puzzle in a different order).
EDIT: And that's exactly why I don't use the flash solver for puzzles which need more than e.g. half an hour of my time. The flash solver is great for the small puzzles, but it doesn't fare well for Naivoj's user beasts.
Last edited by MondSemmel - 2008.07.08 12:13:30 | MondSemmel
Kwon-Tom Obsessive
Puzzles: 6159
Best Total: 7m 47s | Posted - 2008.07.08 16:04:13
Solved the puzzle in two sessions (the first session only lasted ~20-30 minutes or something like that, and I restarted the puzzle in the second session because I'm better when beginning with an empty puzzle instead of a slightly filled puzzle). The second session lasted ~100 minutes, with both sessions totalling a bit more than two hours.
I definitely needed longer for #73, but in the case of #88 I had already read the statement that #88 was easier, so I decided not to stop until I'd solved it (well, I would have had to stop if it had taken much longer).
I had some luck with explorations, but I do have to agree that the puzzle seemed easier than #73 - or maybe it felt easier because I had already solved a puzzle of comparable difficulty?
That being said, basically a single session for such a puzzle (as I said, I restarted as stated above) is really exhausting. I solved #73 in many, many steps (which took much more time than #88 ) but at least it wasn't as draining^^. I only need 30 minutes at most for the beasts of the month (although that number is skewed somewhat because I have already solved most of those at least twice, which makes solving them much faster), but I do these in one session.
I usually never do your user beasts in one session but instead spread them in multiple sessions. It's more relaxing that way, although it makes solving times longer.
With #88 solved, I have solved all your posted user beasts so far (again) - I think I'll never learn how to ration them...
I'm not sure I just want to post the star coordinates of the earth, so I'll leave you with the fact that "MondSemmel" basically means "Moon rolls" (e.g. the bread rolls you eat, although there are no "moon" bread rolls), so I guess at least my nickname is literally extraterrestrial.
Last edited by MondSemmel - 2008.07.08 16:04:48 | Naivoj
Kwon-Tom Addict
Puzzles: 314
Best Total: 33m 50s | Posted - 2008.07.12 09:09:34
Quote: Originally Posted by mondsemmel ...
I had some luck with explorations, but I do have to agree that the puzzle seemed easier than #73 - or maybe it felt easier because I had already solved a puzzle of comparable difficulty?
...
With #88 solved, I have solved all your posted user beasts so far (again) - I think I'll never learn how to ration them
... |
These techniques and you experience with #73 made #88 easier, I think.
Pqg and Petit-Pain-de-Lune, tell me which one is the hardest beast?
I would not be surprised that if Pqg redo his choice of hardest beast now, he would resolve it a lot faster because of the experience and technique acquired solving #73 and #88.
- Maybe not for Petit-Pain-de-Lune, because he's redoing them all the time.
Thanks for the E.T. anecdote!
I should be posting a user beasts batch in the next few days. | MondSemmel
Kwon-Tom Obsessive
Puzzles: 6159
Best Total: 7m 47s | Posted - 2008.07.12 12:47:34
Quote: Originally Posted by naivoj #88 is by far the hardest beasts for our solver, it's the only one that requires 2 levels of recursion. Our solver don't make highlander deductions (even simple ones) it can only apply highlander patterns, but you beings have developed out of this world highlander techniques.
These techniques and you experience with #73 made #88 easier, I think. |
The highlander problem is quite difficult - essentially, highlander deductions are probably the easiest deductions for a human in huge puzzles in terms of finding them versus how much they help - applying these usually helps solve substantial amounts of the puzzles.
Quote: Pqg and Petit-Pain-de-Lune, tell me which one is the hardest beast?
I would not be surprised that if Pqg redo his choice of hardest beast now, he would resolve it a lot faster because of the experience and technique acquired solving #73 and #88.
- Maybe not for Petit-Pain-de-Lune, because he's redoing them all the time. |
I had to blink when I saw your translation^^. It's fitting, though. I learned French in school for several years, but sadly I've forgotten most of it. In comparison, I probably read more English than German (my mother tongue) on the Internet.
At the moment, I think #73 is the hardest. But commenting on the earlier puzzles is really difficult due to accumulated experience/techniques and the fact that solving puzzles again is much easier than solving them the first time.
I probably phrased this poorly:
Quote: Originally Posted by mondsemmel With #88 solved, I have solved all your posted user beasts so far (again) - I think I'll never learn how to ration them
... |
I have solved all of your beasts once and about fourty (and none of the last ~25 puzzles) of them twice.
I'm looking forward to the next batch! | Nis
Kwon-Tom Obsessive
Puzzles: 2208
Best Total: 22m 1s | Posted - 2008.07.15 20:24:40
Just finished #73 in ... effective time no less than 30 hours, probably a bit more. | Naivoj
Kwon-Tom Addict
Puzzles: 314
Best Total: 33m 50s | Posted - 2008.08.28 01:52:04
Quote: Originally Posted by pqg OK, all done! I finished all the archives, user puzzles and BotMs some time ago, leaving just the user beasts, which I started working through about a month ago. I had done about a dozen before that, but since I couldn't fully remember which ones, I did them all in order this time to be sure I didn't miss any, culminating in #99 which I've just finished. |
| pqg
Kwon-Tom Obsessive
Puzzles: 6384
Best Total: 15m 37s | Posted - 2008.08.28 12:11:59
Quote: Originally Posted by naivoj Pqg, could you comment on how difficult was #99 for you compare to #73, #88 and any other beasts you consider hard. |
Warning: long post coming, including possible spoiler for general approach to #99!
Well, as I went through all your user beasts, I came to the conclusion that your difficulty ratings (which I presume are based on solver stats) were, in general, not to be relied on for human solvers. While the 'ET' ones are certainly among the most difficult, they are definitely not the orders of magnitude tougher that you imply. Conversely, many of the ones you rate as easy/medium took me longer than some of your hard/harder ones. Some ideas for reasons why this might be:
Things that are tough for human solvers but not so difficult for programs:
(or perhaps more accurately, things I find tough)
Finding a pattern-based deduction when there is only one (or two) available at a time in a very large puzzle. Sometimes, you can waste so much time searching for these that's it's quicker to resort to T&E. You have rated any puzzle that can (theoretically) be done without T&E as Easy, but if it throws up repeated points like this it can be quite tricky. I think this is the main reason why I found even your Easy ones more difficult than most of the BotMs. (Bizarrely, this situation is more likely to occur in easier puzzles - when large sections are completed at once, it's more likely that you'll miss a deduction and have to waste time looking for it later. This is because the deduction tree has so many branches it's not so easy to keep track of them. eg. You use a pattern to give you a line (or cross) A, which implies B, C and D. each of them implies a few more things, and so on...
In the more difficult ones you only get to deduce small sections at a time, A -> B -> C... so you're less likely to miss bits.)
Things that are tough for programs but not so difficult for humans:
(or perhaps more accurately, things that appear to be tough for your program)
Trial and error in general. We can make more intelligent choices of which explorations to try, and when to attempt and/or persist with 2+ply T&E. (usually when there's a small area isolated, so you know conclusions can be reached) We can also use knowledge gained from inconclusive explorations to make better choices of where to investigate next. Remembering inconclusive explorations also comes in useful when we return to that area having deduced something else elsewhere - we know which explorations may be affected by the new information, so can try them again and not waste time with others. Some puzzles are more susceptible to these techniques than others (in particular, it's partly due to things like this that I found #88 easier than #73)
Another thing human solvers use a lot but I gather your program doesn't is highlander deductions (although I believe they could be automated much easier than the other things above)
Now, #99: OK, I'll agree it was the toughest yet, but not by far. I'd say it is no more than 1.5 times as tough as anything else, but probably less than that. (as opposed to your stats which suggest it is many times as difficult) Compared with #73, there was much less of a 'start' given, but it did solve bit-by-bit in quite an orderly manner from top-left to bottom-right (while some explorations got pretty large, they never covered 1/3 of the puzzle, as they did in #73). Thematically, then, it was more like #88. What made it more difficult was that almost all of these parts needed some quite complex reasoning. I used many (non-pattern) highlander deductions, and extensive 'incomplete' counting-based reasoning, combining the 2 at some of the most difficult points. In fairness, I was using all the techniques honed by the first 98 puzzles to the full, so I guess that made it seem easier than if I had been developing the techniques at the same time. (which may have been partly responsible for long solving times on some previous beasts)
All in all, it was a very satisfying puzzle to solve - pushing the boundaries of what's been learnt from the earlier ones, but by no means out of reach for anyone who's done the others. I also liked #98 btw - an easier progression through it, but towards the end it left me with several isolated sub-areas, all of which were interdependent in quite a complex way. I don't think I've had to consider as much at one time in any other puzzle.
As for solving time, well, I have been much slower than Mondsemmel has for these beasts. My best weekly time is over double his, so I guess that's to be expected. I also tend to be deliberately slow and thorough when doing beasts, because making an error and having to backtrack or start over is much worse than in a small puzzle. This means that I mark every X, and when I get to the FP stage of things I double check many deductions based on long T&E explorations - in particular all those involving complex counting/highlander dedcutions. He reports average times of 30 mins. Although I may have got close to that for some of the easiest BotM, they usually took me 50-70m, while few if any of yours were less than 90m and I'd estimate the average at over 2h (bumped up by a fortnight of going slower while I watched the Olympics at the same time as solving!) The longest time for me was a bit over 9h for #99, followed by ~7h for #73 the first time round. There were a handful of others over 6h (one or two of which you had rated as 'Medium'!) and #88 took about 5h. Bear in mind that human solver times in general and mine in particular are not very reliable as statistics compared to program ones - due to our highly varying levels of concentration, distractions, sobriety, etc....
Incidentally, did you add some more patterns to your program at some stage during production of these beasts? I think I noticed a point (somewhere around #55-#60?) after which some more complex patterns needed to be used much more routinely.
Finally, let me make the point that all of these difficult puzzles were made easier by the knowledge that they were difficult! Having been told how tough they are (in fact, IMHO, having had their difficulty over-exaggerated), you know to get the heavier weapons out of your solving armoury from the beginning. Without this knowledge, I think much more time would be wasted searching for easier deductions.
Last edited by pqg - 2008.08.28 12:55:03 | MondSemmel
Kwon-Tom Obsessive
Puzzles: 6159
Best Total: 7m 47s | Posted - 2008.08.28 13:44:26
Wow, someone managed to solve #99? I somehow don't enjoy puzzles that require a solving time of several hours or several sessions as much as the easier ones, so I haven't even solved #98 yet (as half the puzzle is interdependent, as pqg said, it's really hard to solve) - and neither #99, although I already did the top part of the puzzle. Seeing how someone has managed to solve them, I guess I'll have to pull myself together and try them again^^.
I think my record for any normal BotM (the official ones) is below 20 minutes, but this was never in a first attempt, so the numbers are naturally much lower. In any case, I need much, much more time for Naivoj's puzzles^^.
Quote: Originally Posted by pqg I used many (non-pattern) highlander deductions, and extensive 'incomplete' counting-based reasoning, combining the 2 at some of the most difficult points. |
Naivoj, could you post some more beasts soon? I'll solve #98 and #99 soon, but until then, I'd really prefer some "lighter" puzzles (in comparison to those I just mentioned) - this would probably be your difficulty rating of "hard" or "harder" (although that sounds grammatically wrong)? |
|
