SudokuSolver Forum

Hi all

I have come to think that the boundary between HS puzzles and very hard assassins has become very thin.

To me the idea was the difference between human intuition and programmed number crunching.

I would welcome all your views on this.

Cheers

Maurice

Well, what I would consider "human solvable" is probably different from what most of the rest of you consider "human solvable". Haha! Especially when it comes to killers. Regarding gattai and extra-constraint variants, I have concluded that X-wings and quads provide enough challenge for me that I haven't made much attempt to learn more difficult techniques.

The JSudoku column includes the solver order that I use and the number is the "Complex intersection" that JS uses, plus any chains. Not allowed to attached the solver order to this post but email me if you want it. Download it to your JSudoku software folder to get JS to use them. Uses far fewer chains. Can then compare it to a table I made for 10 assassins.

Observations.
1. HS starts with a technique and designs a puzzle around it. My As, puzzle first, then try and solve it.
2. HS wants the two machines to fail (but not always, see HS 16). My As, at least one of machines succeeds on initial puzzle, and gives more options for variety of WTs.
3. HS is okay with one-trick pony. My As, try to also make hard middle and ideally, a hard end. So, big difference in Andrew's step average, between HS and A. This may be why Andrew's ratings for Assassins are slightly higher overall.
4. HS must be interesting. As must be interesting. Neither is hard just for the sake of it.
5. HS and As on a spectrum. So HATMANS thought about a thin line is probably correct. But keep the HS coming if you can HATMAN! But surely, we have to run out of finding new techniques!!

+-----------------------------------+----------+---------------+-------+--------------+
| Puzzle              WT sub-steps #|  Andrew  | Andrew rating | Score |   JSudoku    | (Complex Intersections)
+-----------------------------------+----------+---------------+-------+--------------+
| HS 15                             |    25    |     E1.5      |  1.95 |     DNF      |
| HS 16                             |    49    |      1.5      |  2.35 |      0       |
| HS 17     variant                 |          |               |       |              |
| HS 18                             |    95    |     E1.5      |  DNF  |     DNF      |
| HS 19                             |    94    |      1.5      |       |     ??       |
| HS 20 Hard                        |    64    |     H1.25     |  DNF  |     DNF      |
| HS 21      variant                |          |               |       |              |
| HS 22                             |    73    |    no rate    |1.70(R)| 7 + 1 chain  |
| HS 23                             |    52    |    no rate    |  DNF  |     DNF      |
| HS 24                             |    48    |      1.5(?)   |2.55(R)|      9       |
| HS 25                             |    81    |      1.5      |  2.75 | 7 + 1 chain  |
| HS 26                             |    75    |      1.25     |  DNF  |     DNF      |
+-----------------------------------+----------+---------------+-------+--------------+

+-----------------------------------+----------+------------+-------+----------+---------------+-------+--------------+
| Puzzle              WT sub-steps #|  Andrew  | wellbeback |  Ed   | Average  | Andrew rating | Score |   JSudoku    |
+-----------------------------------+----------+------------+-------+----------+---------------+-------+--------------+
| Assassin 380                      |   104    |   85   (!!)|   76  |    88    |     H1.50     |  1.90 |      3       |
| Assassin 381                      |    76    |   63   (!!)|   89  |    76    |      1.50     |  1.60 |      0       |
| Assassin 382                      |   106    |  105 (!!!!)|  101  |   104    |      1.50     |  1.75 |      3       |
| Assassin 383                      |    97    |   86 (!!!!)|  107  |    97    |     H1.50     |  2.00 |   4 + 1 chain|
| Assassin 384                      |    88    |   46  (!!!)|   79  |    71    |     H1.50     |  1.95 |      4       |
| Assassin 385                      |    91    |   78    (!)|   58  |    76    |      1.50     |  2.10 |      4       |
| Assassin 386                      |   105    |   67  (!!!)|   72  |    81    |      1.50     |  1.45 |      6       |
| Assassin 387                      |    77    |   75     ()|   81  |    78    |     H1.25     |  1.55 |      5       |
| Assassin 388                      |    99    |    no WT   |   64  |    82    |      1.50     |  1.55 |      5       |
| Assassin 389                      |   110    |  104   (!!)|   89  |   101    |     E1.50     |  1.60 |      2       |
+-----------------------------------+----------+------------+-------+----------+---------------+-------+--------------+

I don't know if I agree with HATMAN's original thought. To me, the difference between a good HS and a good Assassin is that an HS is relatively simple to solve if and only if you spot the step(s) that the solvers don't.

I don't know if this is a good example or not since I don't know what the solvers are capable of, but in the latest HS 26 I used the fact that (56) in n8 are either both in the 14(4) or both in the 28(4). Is that something solvers can stumble on?