The New Sudoku Players' Forum

[StrmCkr]

Well technically the originally forums had the hardest know puzzle predated his by weeks, within the context of "S.E" cant solve as bases for current known hardest

He left this forum and published his own work self declared hardest puzzle ignoring a lot of work found here and emails from us regarding his lack acceptance to our work in 06.

3rd nov 06
JPF found an earlier sept 06 that could not be solved by 1.1 S.E beating him by weeks

Since then this community has published multiple puzzles that dwarf his in complexity.
Yet none of us here ever tried to seek media attention to rectify his claim.
Do we need to, probably not.

Ie terrik, Metcalf, myself, eleven, plus several others all have 11.4 plus sudoku explained rated puzzles from that era.
Compared To his ~10ish (these even still quash his updated hardest)

When we contacted Nicolas to update and implement more ram to verify our puzzles rating all back in 06 from version 1.1 -> version 1.2
As before that update it failed to complete on ours but could solve his.

three puzzles specifically: Easter Monster, Metcalfs, Mine: all required that update to complete

[SpAce]

StrmCkr, it seems that there's conflicting information about the sequence of events even on this site. On the thread below some people (RW, tarek) do opine that Arto had a claim to fame for a short time. Even you didn't sound quite so certain about the dates back then. I'm just saying that for an outsider it's kind of hard to reach a clear conclusion about what really happened back then, if even the insiders are second-guessing themselves.

http://forum.enjoysudoku.com/has-anyone-ever-tried-solved-the-ai-escargot-t30203.html

[eleven]

If someone wants to recall exact dates and counts, it can be read in the forums "hardest sudoku" threads. These were THE place, where
- a open collection of potentially hardest puzzles was hold
- it was seriously tried to find "the best" rating systems, and lists were hold in parallel for different open rating programs
- it was open for everyone to contribute, discuss, claim and whatever

Arto never published his own rating system (a shame for one having a math degree). So the forum had no possibility for a sincere comparison. He just claimed - and the dummies believed ...

[StrmCkr]

2nd guessing my self not really; I've still stuck to my gut instincts on this very topic: we have the "hardest" before him and still do to this day exact dates I leave that for pat to post {which she usually will} but in this case i updated my post and put links in with dates showing when S.E borked and failed to rate a puzzle confirming what i said.

back on topic:

you asked if there is many people that Manually still solve: yes there is specially mark ups ? not really, i don't use any for any printed/published puzzle very few actually break into the 9+ rating where i might need to draw/map out a move set via links. In which case i drop it into a program that auto generates Pencil marks and highlight the digits/cells i'm focusing in on to connect logic.

asides from the normal RC,[ RN,CN,BN]* space i also have mapped some special mark ups in my coded solver are unique to me:
Mini row/Cols and ERI's {which is the combination of mini Row + Mini col {useful for strong/weak links in chains and empty rectangles}
R Digit C, C Digit R, B digit square{by row}, B digit square{by Col} {which allows for faster finding on Hidden sets}
note* - naked set

advice, besides from learning how something works and putting it in practice.
{ how to find said techniques usually comes down to the following real question is: how do you view a puzzle and can you see both points of view.
spaces left {naked} vs spaces used.{hidden} }

ps. that's also the key for speed solving

that's pretty much all you need to draw links between cells & digits in your mind.

the fun part is this:
what order do you apply it in, as a techniques elimination can make a puzzle harder by removing the application of an easier technique.
and how vast is the tool box you currently have and what are its limits.

[eleven]

Oh please StrmCkr,

it is a fact, that Arto's puzzle was the first one published, which was too hard for SE, in Oct. 2006. All other came later. E.g. read this post, which also shows to me (again), that Arto has been welcome at the forum at that time.
Ocean's puzzle came 2 days later, see here, and the puzzles in JPF's list were not so hard, see here

[m_b_metcalf]

JPF found an earlier sept 06 that could not be solved by 1.1 S.E beating him by weeks

Without wishing to muddy the waters, 'could not be solved by SE' at that time had, IIRC, two meanings:

1) JPF's failed because of a bug in the BUG code, and not because it was 'hard' (se re'born's post in the middle of this page);

2) really hard puzzles 'failed' simply because the amount of space supplied by default was insufficient. That was easily fixed (see near bottom of this page).

HTH

Mike Metcalf

[StrmCkr]

i am aware there was a "bug" with bugs, merely pointing out the tribulations within "se failed" to solve as a bases for his claim,
having the first 10~ rated puzzle once the bugs where fixed would definitely belong to him by 2 days

then its splitting hairs over who's was actually harder out of those 10~'s

that's as much as ill concede since he still hasn't or wont recognize the harder ones found in the threads linked as being harder then his own.

both instances paraphrased above fixed in the last version of s.e released

[eleven]

So hopefully this is clarified now.

ps. that's also the key for speed solving

The key for speed solving hard puzzles is doubtlessly GUESSING.

[SpAce]

asides from the normal RC,[ RN,CN,BN]* space

What's that?

i also have mapped some special mark ups in my coded solver are unique to me:
Mini row/Cols and ERI's {which is the combination of mini Row + Mini col {useful for strong/weak links in chains and empty rectangles}

I have those, if I understand correctly what you mean. I think I have a distinct mark-up for every kind of conjugate linking within boxes, lines, box-lines, and ERIs. If I look at a candidate I can immediately tell what kind of immediate relationships it has with its environment. I have different markings for 1-to-1, 1-to-many, and many-to-many types of strong links within lines and boxes. The ERIs are the most complex because it's functionally different whether there's a candidate in the intersection or not, so the groups need different markings for both cases (and the different intersection shapes, too).

I just counted that I currently have 21 distinct markup symbols for a single candidate, most of which have four or two possible orientations, and up to three can co-exist. So, the number of possible combinations is quite huge. Nevertheless, I find them easy to apply and read because they're designed to be intuitive (for me, at least).

The hardest part was to design them so that they have an additive progression from one type to another because I don't allow erasing. For example, a many-to-many link turns first into a one-to-many (and many-to-one) link, and later into a one-to-one link, and the difference should be noted on both sides without erasing anything. On a computer program you don't have such restrictions and can erase and update any symbols at will.

Another challenge in a manual system is to keep the mark-up fresh and valid whenever something changes. For that I have an in-place book-keeping system and a multi-stage elimination protocol which ensure that whenever something changes, all of its effects are reflected on the grid. In practice the process works almost as reliably as a computer program, except it's much slower, of course.

[David P Bird]

SpAce, I gave you you sufficient time (~36 hours) to retract your previous insulting reply to me that was obviously written in anger, but when you didn't, I reported it to the moderator (who is presumably away at the moment).

To pick on the point that triggered your venomous tirade:

I also find it interesting that you've now made ad hominem attacks on three different people in one thread (first John Welch, then Arto Inkala, and now me).

I challenge you to justify that. If any-one is leaping to conclusions it is you.

When I suggested you conducted a forum searched for Arto I, it was for you to understand why he was not well received, nothing more.
I then summarised the reason

The people that are not well received here are those that make exagerated claims and stubbonly stick to them when they are proved to be so, sometimes because there are commercial aspects.

Where did I say that that was what you were guilty of ? I was just warning you to avoid doing the same.

How you continued was contemptible, which is why I won't engage with you in future.

[SpAce]

SpAce, I gave you you sufficient time (~36 hours) to retract your previous insulting reply to me that was obviously written in anger, but when you didn't, I reported it to the moderator (who is presumably away at the moment).

Threats. Cool. I hope the moderator double-checks what you wrote as well to get a balanced view.

If any-one is leaping to conclusions it is you.

Maybe, maybe not. I acknowledge that there's a possibility that I completely misinterpreted you, and if that's the case I sincerely apologize. It's not yet the only candidate left, however.

When I suggested you conducted a forum searched for Arto I, it was for you to understand why he was not well received, nothing more.

So you were already then implying something similar about my style? Or why else did you think it was necessary for me to search for Arto? How stupid of me that I didn't get your point the first time. I thought you just wanted to let me know that Arto was not held in high regard here because I happened to mention his name. That would have been fine, although I already had a pretty good idea of his standing here. I had no idea you had ulterior motives for your suggestion.

I did, however, conduct a new search as you suggested. In the first thing I found Arto was called a fraud. Maybe he is, maybe he isn't, I don't much care. I agree that his conduct doesn't seem to qualify for my ethical standards. What I do care about, however, is that you imply similarities between me and someone who's considered a fraud by possibly a lot of people here. I happen to take pride in my honesty and you have to have a frigging good case before you make any implications otherwise. Can you understand that? If I get kicked out of here for getting pissed for someone questioning my honesty for no reason, then so be it.

I then summarised the reason

I wrote:
The people that are not well received here are those that make exagerated claims and stubbonly stick to them when they are proved to be so, sometimes because there are commercial aspects.

Where did I say that that was what you were guilty of ? I was just warning you to avoid doing the same.

That's a convenient narrative, but it's not very obvious from what you wrote. I'd be happier to concede a little if you also took a bit of responsibility of what you wrote before putting all the blame on me. Your summary was in a different post and in a context that made it much easier to interpret that you were talking about me instead of Arto -- or us both. You didn't mention his name at all at that time. Instead you made implicating assumptions about my claims:

In that light, I find your choice of words worrying. Surely to say your system is "extreme-capable" is premature, and calling it a "propriety system" has commercial overtones.

So you see why I made the link to myself? You did imply -- or more accurately you stated it as a fact -- that I was making exaggerated claims about my system. So yeah, I think you quite clearly said I was guilty of at least part of what you said before. How can you deny it now? Can you please explain how I should have made a different conclusion?

And, as I asked before, why is it premature to call my system extreme-capable? Can you understand at all why a belittling statement like that, for which you gave no supporting reasons, pissed me off? Especially in a context where you've implied comparisons to someone who's considered a fraud or at least unethical by several people here. I'm pretty good at taking valid criticism, and I'm actually grateful for it, but yours wasn't it. It was insulting, and I responded in kind. I know it wasn't very high-class behavior, but you did push certain red buttons. I hope you understand that.

[eleven]

Nice to read an emotional debate from native speakers - it improves my English.
SpAce, please forgive David, his skin is getting thinner and thinner (for the moderator: i say this with sympathy for David).

[eleven]

The hardest part was to design them so that they have an additive progression from one type to another because I don't allow erasing. For example, a many-to-many link turns first into a one-to-many (and many-to-one) link, and later into a one-to-one link, and the difference should be noted on both sides without erasing anything.

Very interesting. I only know one manual solver without a rubber. He never makes a mistake (with the easy puzzles, he is solving).
So if someone makes a mistake with your system, he is fired (has to start from the beginning) ?
(When i make a mistake, it is 50:50, that i can find it some steps back)

But i like the idea. When will it be published ?

[StrmCkr]

StrmCkr wrote:
asides from the normal RC,[ RN,CN,BN]* space
What's that?

i originally had a massive excel spreed sheet with a massive list of techniques where i mapped out each technique individually case by case and was able to solve ~60% of the top1465 list before i ran out of memory in the program and locked my self out of it via saving it. {found out i hit the 2.4 ~ Gig built in memory limit when i tried to reopen it}

{in essence i could drag and drop a grid into it and it would instantly produce the solution in zero time {based on if my solver had the key technique programed into it}

which was pretty epic,

however since then i re-taught my self some Free pascal and moved to coding never been able to replicate the above as all coded works has to cycle each technique one by one instead of all at the same time which is an advantage of excel.


when you are looking at filled in pencil marks you are looking at the summation of Rn,Cn,BN space so its not really something knew just a diffrent vantage point of the solving process by breaking it apart into respective space.

Grid: represents the puzzle as imported storing the Digit placed at each of the [1..81] cells

RC - represents every cell that is not solved [1..81]

RN - represents each row per digit that is active
CN - represents each col per digit that is active
BN - represents each box per digit that is active

to figure out if Cell 1 Pencils marks its sum [ R1N * B1N *C1N] {where N = 1..9) {ps i'm using set-wise operations stormdoku - my program}
this generates an easy view area that shows "naked" singles/sets {ie all other candidates are empty}

I use a variation for naked sets:

RC- storing number set {row, col }
RBS - storing number set { box, square by row ->> left to right} }
CBS - storing number set { box, square by col ->> left to right} }

Hidden sets is a bit more difficult for computers but easier for humans as its the only cells left for digits to be in for a sector. {exactly the inverse of Rn,Cn,Bn: }

RD - Row ,represents each by Digit storing Col
CD - Row, represents each by Digit storing Row

BD - box, represent each Digit storing Square by row->> left to right}
BD - box , represent each Digit storing Square by col->> left to right}

For box/line reduction i use: {mini row/col}
BSR - storing number sert {box, square @ Row }
BsC- Storing number-set {box, Square @ Col}

Empty Rectangle intersection { ERI }
ERI: Cell storing number-set

defined as:
( RD * BD) + (CD * BD) = BD
where (RD or CD) <> box and box <> [empty]

[SpAce]

Very interesting. I only know one manual solver without a rubber. He never makes a mistake (with the easy puzzles, he is solving). So if someone makes a mistake with your system, he is fired (has to start from the beginning) ? (When i make a mistake, it is 50:50, that i can find it some steps back)

No, he's not fired The player can still use an eraser, of course. I only meant that I don't allow the mark-up progression of my system to depend on erasing. That's why all marks only grow, never shrink, when they progress. So theoretically all you need is a pencil, or even a pen if you're really confident. Most of us aren't totally error-free, however, so it's perfectly fine to use an eraser to correct mistakes, and nowadays I do too. I used to use a pen when I had a simpler version of the system and solved only basic puzzles, as mistakes were so rare that I didn't care if one happened and I had to discard the puzzle. Now I mostly use a pencil because my new mark-up details need more sharpness and it's easier to make minor mistakes due to the increased complexity.

The system is designed to minimize the probability of bigger errors and, in case they still happen, to catch them early. Early detection makes it often (but not always) possible to backtrack and correct the mistake and its implications before it's too late. Careless errors mostly happen to me in easier puzzles where I aim for speed and only use basic parts of the system or none at all, but they're still rare. I don't have actual statistics but I don't think I'm lying if I estimate that my catastrophic error rate is less than 1%, and of those my recovery rate is probably about the same as yours (50%). If I catch an error in a difficult puzzle and can't be absolutely sure I've corrected all of its implications, I may rather restart than take the risk of frustration at the end.

But i like the idea. When will it be published ?

Thanks for asking. I'm a bit of a perfectionist so publishing anything is hard, as I always see targets for improvement. For example, right now I'm thinking of ways to add weak-link markers back to the system, but I think it's a dead-end. I used to have them but then ran out of room and kicked them out as the least necessary component. The core has been pretty stable for a while now, though, and it would be nice to hear comments and suggestions if someone's actually interested. Too bad things got a bit sour with David, so perhaps it's better to let the dust settle and see what happens before I publish anything here.