Spider solitaire

[Quartz]

I'm pretty good at the 4-suit game, winning nearly 60% of the games - and I'd win more if I could save the game under Vista without exitting - but I'm wondering: are all the games presented winnable?

[Sam Stone]

How the heck do you win 60% of the games with 4 suits? I think my win average is about 1/10 of that.

I must be doing something wrong.

[Snarky_Kong]

Originally Posted by Sam Stone

How the heck do you win 60% of the games with 4 suits? I think my win average is about 1/10 of that.

I must be doing something wrong.

Yeah, even with taking back all the turns to find the optimal cards to overturn I'd be nowhere near that.

[Quartz]

It can take over 1500 moves - I don't think I've cracked 2000 yet. One trick is to move only part of a set and not the whole set. Another, more critical, objective is two blank columns.

[Snarky_Kong]

That's way way more time than I'd care to spend.

[jacquilynne]

It's kind of hard to prove that every possible deal of a computer game is winnable. But proving that they're not is usually much easier -- just come up with a deal that's not winnable.

In this case, imagine one where the last row of cards to be dealt up is all, say, Kings and Jacks. The kings can't go anywhere because there are no empty spaces. The jacks can't go anywhere because they don't play on Kings. Hence, you can do nothing with that deal and you lose.

Whether all the games that your version of the software presents you have been pre-determined to be winnable can vary -- there are solitaire programs out there that let you set 'winnable deals only' as an option. That's fairly non-standard, though and it would most likely be a menu option in the software, not a default.

[divemaster]

Originally Posted by Sam Stone

How the heck do you win 60% of the games with 4 suits? I think my win average is about 1/10 of that.

I must be doing something wrong.

I love Spider, but this is more in line with my experience. 60%? WTF? There are games I've gotten pretty far on and then get hung up--those I recognize are likely winnable if I had made different moves. But there are plenty of times I can't for the life of me see any way to win certain set-ups. At least not given the parameters I play under. Are there variations to allow an easier game?

[jacquilynne]

I find that number fairly surprising, as well. SolSuite lists the chance of winning a straight game of Spider at about 5%.

[ZipperJJ]

Hot damn! I stay solidly around 25% for two suits and I play a LOT.

You are a Spider Solitare genius

[Brown Eyed Girl]

Originally Posted by ZipperJJ

Hot damn! I stay solidly around 25% for two suits and I play a LOT.

You are a Spider Solitare genius

Word. I have never won a four-suit game (admittedly, I get so frustrated, I don't bother trying much) and my highest score in a two suit game is 1194. I can't even break 1200 and my win rate hovers around 20%.

I suck.

Help?

[NoCoolUserName]

I only play 4-suit and I probably win 25% of my games. I don't restart games if I lose and might win more if I did. I just play to kill time while letting some long-running junk run in the background. I find 60% unlikely but not impossible. I generally play about 500 moves--never look at the points, and don't really count how many I win, so 25% is a WAG at best.

[Quartz]

Originally Posted by jacquilynne

It's kind of hard to prove that every possible deal of a computer game is winnable. But proving that they're not is usually much easier -- just come up with a deal that's not winnable.

In this case, imagine one where the last row of cards to be dealt up is all, say, Kings and Jacks. The kings can't go anywhere because there are no empty spaces. The jacks can't go anywhere because they don't play on Kings. Hence, you can do nothing with that deal and you lose.

The theory is true but your example is not: you just need to make sure that you end up with Queens where the Jacks will lie, so you move the stack of Q+J onto a King.

I'm well aware that you can create an unwinnable game - you just repeat your example for each deal. But I'm asking if the Spider Solitaire can do this: will it present an unwinnable game?

[Quartz]

It's actually 58% (pic). Note that this is a recent reinstallation of Vista so I haven't played many games.

[Blue Mood]

Originally Posted by NoCoolUserName

I find 60% unlikely but not impossible.

I win more than 80% but it's because I cheat. I undo to my heart's content and save before dealing in case I make a mistake - then I can revert to the saved game.

[Hockey Monkey]

Originally Posted by Quartz

It's actually 58% (pic). Note that this is a recent reinstallation of Vista so I haven't played many games.

Was anyone else tempted to hit the reset button in the screenshot?

[jacquilynne]

Originally Posted by Quartz

I'm well aware that you can create an unwinnable game - you just repeat your example for each deal. But I'm asking if the Spider Solitaire can do this: will it present an unwinnable game?

As I said, generally solitaire games that have 'present only winnable deals' as an option if they offer it at all. If you haven't set 'present only winnable deals' in your software, there's a very low chance that it's doing that.

[Quartz]

Originally Posted by Blue Mood

I win more than 80% but it's because I cheat. I undo to my heart's content and save before dealing in case I make a mistake - then I can revert to the saved game.

In the Vista version, you can undo a deal but not save.

[Quartz]

Originally Posted by Hockey Monkey

Was anyone else tempted to hit the reset button in the screenshot?

Rotter!

[Don't Call Me Shirley]

OK, I am astounded by the claims of win percentages here. I win maybe 25% of my two-suited games. I tried a 4 suited game once. Once.

What is the secret? When does it make sense to play a red card onto a black card? I'm sure that is where I am going wrong. Obviously it is always good to uncover a new card, but is it worth screwing up a long string of same-suited cards to do so?

[NoCoolUserName]

Originally Posted by Blue Mood

I win more than 80% but it's because I cheat. I undo to my heart's content and save before dealing in case I make a mistake - then I can revert to the saved game.

OK, I stand* corrected. Clearly %60 is not that unlikely.

* Actually I sit corrected.

[NoCoolUserName]

Originally Posted by Don't Call Me Shirley

...What is the secret?...

The next time I'm waiting on a long-running process I'll screen-shot some examples and post a lesson. Although clearly there are others who might have better techniques.

[Mizz.Lucifer]

That game is soooo boring

[Quartz]

Originally Posted by Don't Call Me Shirley

Obviously it is always good to uncover a new card,

No it isn't!

is it worth screwing up a long string of same-suited cards to do so?

Sometimes.

Some further tips: don't automatically move Kings to blank columns, remember that you can't put anything onto an Ace, recognise when you've lost control, undo is your friend, and try to get two blank columns. Having two blank columns makes the job of reorganising the other 6 much easier. Don't be afraid to move columns around either.

[Shalmanese]

My strategy: Your first goal is to uncover a blank column. Get one of these at almost any cost, then liberally use undo to see if you can uncover it in a cleaner way. Try as hard as possible to keep this column clear, put only a single stack on there and work as hard as you can to clear it after every deal.

After that, there are 2 phases I alternate between, tidying and uncovering. Using the blank columns, it's possible to move around the cards so that same suits end up together and piles get uncovered. Once it's no longer possible to tidy, sacrifice the blank columns to try and uncover other cards but make sure you end up with only a single pile in each blank column before you deal again.

[Blue Mood]

Originally Posted by Shalmanese

Your first goal is to uncover a blank column.

Exactly. Kings don't go into an empty slot unless doing so vacates a different column. Suit doesn't matter much in the beginning - it's much more important to clear those columns. Also I usually have a "junk" pile - starting with a king, I place cards there that I don't want anywhere else.

[Voyager]

Originally Posted by Quartz

No it isn't!

Sometimes.

Some further tips: don't automatically move Kings to blank columns, remember that you can't put anything onto an Ace, recognise when you've lost control, undo is your friend, and try to get two blank columns. Having two blank columns makes the job of reorganising the other 6 much easier. Don't be afraid to move columns around either.

The only real backtracking I do is to find which card is best, if any, to uncover when there are choices. But I'm a person and not a backtracking algorithm. Writing code to play Spider wouldn't be that hard with this kind of brute force approach, so that kills my interest.

[Manduck]

Okay, what is considered a good percentage if you don't do backtracking? I was feeling pretty pleased with my 11% until I opened this thread.

[Quartz]

If you're not doing rampant backtracking (including save / restore) then I think 11% is pretty good.

[NoCoolUserName]

Let's summarize:

* Open spots are gold. Do anything to get them. One is OK, 2 is good and 3 is super.
* Once open spots are available, clean up the stacks--get suits together as much as possible, then continue.
* Backtrack to find the best choice. If you have 3 aces and only 2 deuce, try all three (and all cards uncovered thereby) to find the best choice before continuing.
* Put all the junk in one pile. I used to distribute odds-and-ends across all piles to keep things even--that's less effective than piling everything on one place.
* My rule of thumb regarding matching suits is that (given a choice) I'll take the non-matching one if it uncovers more than 1 extra card.

A hint for making play easier: look for consecutive numbers that are not together. If you have 4-3-2-A on top of a pile, look for an open 5 (or one that you can make open).

[Quartz]

Originally Posted by NoCoolUserName

* Put all the junk in one pile.

I disagree with this: you mustn't let piles get too big otherwise it's too difficult to get at the card you require unless the big pile is well ordered. Think about the number of sequences (not cards) you'd have to move to get that card.

[Charlie Tan]

I just checked and I have 34 % playing with four suits. I also think a great game should have no more than 300 moves. A good game should have under 400 and if I'm over 500, I consider it a failure.

[Don't Call Me Shirley]

A hint for making play easier: look for consecutive numbers that are not together. If you have 4-3-2-A on top of a pile, look for an open 5 (or one that you can make open).

This seems less like a helpful hint, and more like the whole point of the damn game to me.

But maybe that's why I suck at it.

[NoCoolUserName]

Forgot one:

* Don't put a king in an open spot until the end of that round. Any other card you might be able to move, but a king kills that puppy as being open forever.

I'll stand by the "junk pile" strategy. Yeah, it's tough to un-stack, but it makes the likelihood of the all-valuable open spot greater if you don't have junk all over the place.

The "look for consecutive cards" hint--yeah, it seems obvious. However, when you're consolidating suits it's not as easy to see as one might think. If you've got a 6 on the top of one stack, look for a 5 in the middle of some other stack, then work to bring them together.

[Guinastasia]

Not hijack this thread-but how the hell does this game work? Even reading the directions, I've never been able to figure it out.