Friday 24 July 2020
I'll start by saying this is a subject that I remain a bit unsure about and I am going to be careful not to come to too many conclusions. If there's one thing I'd like to tell my dearest readers before carrying on, it's that if you are going to use internet tools to make versions of my puzzles, please can you do me the courtesy of telling me you have done so, and (as per the licence of puzzles published on this blog) please can you ensure that you (a) credit me and (b) provide a link to the licence each time you share a link.
By Tom Collyer at 16:49
Friday 17 July 2020
The purpose of this post is to see whether I can manage to roughly approximate the halcyon days of the nikoli time trials. The idea is for anyone who is interested to turn up to this blog at a fixed date and time, and then race to see who can get the fastest solve time for a particular puzzle. For the time being, the particular puzzle will very likely be a classic sudoku.
In retrospect this doesn't sound anything like a time trial. Therefore I will call my version "Puzzle Sprints".
The way it will (hopefully!) work, at least initially, is as follows.
- I will schedule a post at a fixed date and time with a link to a Google form, which contains all you need. For the real thing, the fixed date and time will be advertised in advance.
- The form requires you to be signed in to Google (in order for me to impose a limit of one response per participant).
- Therefore, to participate you will need both a Google account, and to be logged into that Google account.
- Once the form loads, you should see a short description, together with:
- A link to the puzzle to be solved in an external online solver.
- Space to enter your email and nickname.
- A static image of the puzzle with the two answer keys marked.
- Space to enter the two answer keys into the form.
- To take part in the Puzzle Sprint, open up the link to the online solver and solve the puzzle.
- Once you have solved the puzzle, come back to the form, and then enter in the corresponding answer keys.
- When you are done, hit the "Submit" button.
- Once you have submitted, you can click the "View Score" button to instantly see if the answer codes you submitted were correct.
- After a certain amount of time has passed, I'll declare a cut-off and publish the results.
For those interested in this pilot version, please note there is absolutely nothing at stake here. You are more than welcome to poke around, but for the next few hours after this initial post, it is highly unlikely you will be getting anything out of this, as I will be making plenty of edits and updates. It is mostly for my own benefit for now.
Here is the link: https://forms.gle/WQYhioNFFfCiUC2TA
Once I am happy with the way things seem to be working (I will edit in a message to this effect within this post when I am happy), please do feel free to have a go. If nothing else, submitting an entry to this pilot version is a good way for me to see who might be interested in this kind of thing.
There are several limitations to the approach as it currently stands.
- I don't like people having to log into Google to be able to do this.
- I don't like having to worry about collecting people's emails.
- Ideally there'd be some kind of self-contained and robust log-in system akin to things like nikoli, fed, croco and so on. This would completely negate the need to use something like Google forms in the first place.
- I am not 100% satisfied with any online solving interface for the purposes of these Puzzle Sprints, where the emphasis is very much on speed solving.
- I don't want an undo function
- I don't want a trial and error function
- I do want much better pencil-marks
- I also want the online solver to directly interact with the form so there's no extra faff switching between solver and form. (This is no longer an issue once you eliminate the need for a form...)
- If this were to ever get very serious, I would also want some way for winning solves to be verified. (Most likely this would involve saving some kind of solve history and having replay functionality...)
- Now the cat is out of the bag, anyone can come along and imitate my idea.
There are the following advantages to the approach.
- Everything is more or less under my control.
- The format gets to the heart of what is really exciting about speed-solving.
- My reputation as a good puzzle author.
- My reputation as a good sudoku competition director.
- I am willing to eventually provide cash prizes (for UK based participants)
That's it for now. I'm intrigued to see whether anyone is going to be interested in this.
By Tom Collyer at 15:00
Friday 10 July 2020
So I put out some puzzles earlier today, and I called them Friday Puzzles because it happens to be a Friday. However I don't think it's going to be a revival of a regular series of posting. I suspect that whenever I do post a puzzle, I'll retain the numbering, but I won't be too bothered about things if the day in the week happens not to be a Friday.
There are a couple of things to look out for in the near future of this blog.
The first thing will be an exploration of my interest in ratings systems. Generally the ratings systems used in the likes of Football, Tennis, Chess, various computer games all leave at least something to be desired, and I have some ideas which might go some way to addressing these short-comings. There is a lot of good, freely available, data in the results of various puzzle and sudoku competitions covering a good period of time, and so this is probably where I'll make a first go of things. Needless to say this will be strictly unofficial, in association with no-one other than myself, and on a just-for-fun basis.
My exact mechanism might need a few goes to get the details exactly right, and will probably reveal its own set of shortcomings, but I am excited to be giving that a go in the coming weeks. I'll probably put out a blog post outlining some of the theory behind my ideas as well.
My second idea is to revive something like the old Nikoli.com time trials, whereby I'll put out a puzzle at a given time, and then anyone who happens to be around races to register the first finishing time, which i affectionately remember being referred to as a "1get". This won't be a 100% polished experience without any further expertise and collaboration, but on the other hand I think plenty is achievable with the likes to google forms and pzv.jp. I'd probably look to be handing out some kind of prizes to UK participants at least. Note that this would be purely on my own behalf, and not as part of any organisation.
My third idea might look to start some kind of a puzzle club that eventually ends up replacing regular puzzles being posted here. This may well feature sudoku variations that are on the harder end of the difficulty scale, and that may cross over the "solves like" criteria of the "What is a Sudoku?" view of the world. Maybe this gets splits up into seasons, or even just over the course of a year, and again with some kind of self-funded prizes being up for grabs. This is probably the least developed idea I have for now, and I can certainly see it ending up down one of many possible initial choice of routes.
Time will tell as to how far are get with any of those three ideas, but I am quite excited by the idea of doing things on my own terms and seeing what happens. It feels like reviving the spirit from the good old days a bit, and i think will help me feel a bit better about the puzzling world. Hopefully that will end up being the case!
By Tom Collyer at 20:22
So by now more or less everyone has heard of Cracking The Cryptic. Lots of the puzzles featured these days are what I'd describe as curiosities, sometimes with strange mish-mashes of constraints paired together to produce interesting solving gimmicks, sometimes with constraints and solving mechanics that go way beyond anything that resembles classic sudoku. In any case they represent sudoku variations probably wouldn't retain much in the way of interest once the initial novelty had worn off. Not that it isn't great to see so many different people I'd never heard of before producing these interesting novelties of course!
One idea that does seem to have a few more legs appeared in a more fiendish form that presented below. Ahaupt's idea is for you to construct an irregular grid from a regionless start to the puzzle, using clues which give the sums of numbers to be placed in each of the cells the clue sees in both the horizontal and vertical directions up to the border of a marked region, including the cell the clue is contained in.
I think the idea is perfectly nice if you are given the courtesy of the irregular grid to start off with. You end up with something that is related to Jigsaw Killer, where you combine Irregular Sudoku with Killer Sudoku, but something that I think is a little bit nicer, and with enough differences to warrant calling it a separate Sudoku variation. I'll call it Signpost Sums Sudoku.
The trade-off with Signpost Sums is that you no longer have the cages represented visually, which is a bit of an inconvenience, but you do have the comfort that you don't have to scan across different regions when looking at the sums, and you get the possibility of having some of the now implied cages to be overlapping. As each of the sums are entirely contained within one region, it means that you still get to keep the no-repeats condition with the sums.
I've tried a puzzle using a normal 3x3 box layout, and whilst possible I don't think it was all that enjoyable. Instead I've gone with a couple of irregular grids, one a warm-up, and one towards the harder end of difficulty you might see at a well-balance competition. I'm kind of hoping I've taken enough care with the clues and testing this myself for these to have come out uniquely, but if not I'm sure one of my dearest readers will let me know soon enough.