Razz -- worst hand wins on Absolute - Page 5

silvergoose · #41 09-02-2007, 12:02 AM

Honestly, I think that the fault lies in two places:

First, I think the software that takes two (or more) hands and determines a winner incorrectly thought that aces, being better than 4's, lost. However, the part that determines the best hand did so correctly. That's my guess.

Second, support has no idea how to play Razz and is desperately trying to use the rules on the site to figure out how the software was working.

AggressiveCall · #42 09-02-2007, 01:14 AM

[ QUOTE ]
Honestly, I think that the fault lies in two places:

First, I think the software that takes two (or more) hands and determines a winner incorrectly thought that aces, being better than 4's, lost. However, the part that determines the best hand did so correctly. That's my guess.

Second, support has no idea how to play Razz and is desperately trying to use the rules on the site to figure out how the software was working.

[/ QUOTE ]

Sounds right to me. Which is strange, come to think of it: the code which forms the best 5 card hand must have something internal which allows it to determine what the best hand is, obviously. It seems strange that there are apparently two different blocks of code to do the same thing. Wouldn't they have just made one class and re-used it to help avoid these sorts of problems?

electrical · #43 09-02-2007, 01:20 AM

Looks like they used a lowball hand sorter to choose the five card showdown hand, then used a normal hi-hand sorting algorithm, with an instruction to return the "losing" hand as the "winner."

Ha ha lame.

AggressiveCall · #44 09-02-2007, 02:10 AM

[ QUOTE ]
Looks like they used a lowball hand sorter to choose the five card showdown hand, then used a normal hi-hand sorting algorithm, with an instruction to return the "losing" hand as the "winner."

Ha ha lame.

[/ QUOTE ]

Except, A2345 would lose to 23456 then. And it would have all kinds of issues with straights and flushes. You almost have to *try* to get this so badly screwed up, and yet in such a rare instance.

adanthar · #45 09-02-2007, 12:44 PM

[ QUOTE ]
[ QUOTE ]
Looks like they used a lowball hand sorter to choose the five card showdown hand, then used a normal hi-hand sorting algorithm, with an instruction to return the "losing" hand as the "winner."

Ha ha lame.

[/ QUOTE ]

Except, A2345 would lose to 23456 then. And it would have all kinds of issues with straights and flushes. You almost have to *try* to get this so badly screwed up, and yet in such a rare instance.

[/ QUOTE ]

I think the code just doesn't differentiate between one ace and a pair of aces, with the results that two aces are the "lowest" pair, lol.

electrical · #46 09-02-2007, 01:10 PM

I just realized what is going on. Absolute has invented a new game, 7-card Razz Badugi. Villain had a 4-card Seven, Hero had a 4-card Nine. Villain wins, obviously.

Andy B · #47 09-02-2007, 01:13 PM

This thread makes my head hurt.

SGspecial · #48 09-03-2007, 12:56 AM

[ QUOTE ]
This thread makes my head hurt.

[/ QUOTE ]
Ah but any thread that can bring the silver goose out of mothballs has just bumped to double platinum IMHO

Phat Mack · #49 09-03-2007, 04:01 PM

I've been following this thread with great interest, and I guess I am the only one that's on Absolute's side in this matter. Can't anyone else see that the algorithm is right; it's the game that's wrong.

As a guy who's worked in software development, I can tell you that nothing is more annoying than to produce a fine program, and then have everything screwed up by the end users and their ridiculous "needs assessments."

I'm going to propose a solution. Instead of worrying about pairs and high cards, lets just add up the pips and let the lowest score win--sorta rip off Mike Caro and his Caro Dots game. As Rainbow Warrior (63-6 over No-Co!) predicted, the best hand will be AAA2, the worst, a handful of paint. Easy to program, easy to understand. To make things easier, we'll limit the hands to five cards: if we dealt six cards, we'd have to figure the best of 6 possibilities (a calculating nightmare), and if we dealt seven cards, we'd have to run 21 permutations--God only knows what that would do to their processing time, much less what would happen to the 486 box they use as a server.

No need for thanks, I'm all about you...

Mack

RustyBrooks · #50 09-03-2007, 04:18 PM

I know you're joking but the algorithm for determining the best hand (edit: in your proposed game, of course) doesn't require you to consider all the permutations - just take the 5 lowest cards.

(Proof: start with the 5 lowest cards and sum them up. Which card from the remaining 2 can you swap with one of the lowest 5 to get a better score?)

Also, speaking as someone who has written a razz scorer, it can't be easier... put all the cards in order. Go through this list an add each card to the final hand. If you've already seen "this" card in the final hand put it in an alternate stack. When you're done, if your final hand is not 5 cards, take the lowest card from the alternate stack. Continue until you have a 5 card hand (although you need an alternate stack for this also in case your first alternate stack has pairs in it). Anyway, the algorithm is iterative, short and simple.

Way to go, absolute!