How about not waiting for confirmations, but using the output of the transaction I paid you with as an input of your payment to me.  Then you can do fast payouts like a little-known dice game does.

I think maybe it just has to not be disabled...  The following remembers my input:


Using a copy of the input element from your form also remembers my input.  Perhaps something in one of the javascript libraries you're using is turning it off.

Note that it is still saying "After you pay, click here to see if you won!" even after it has told me that I won.

Received.  Thanks.  Exactly covers my two losing bets.  


I'm using Chromium 18.0.1025.151 (Developer Build 130497 Linux) Ubuntu 12.04 if it's important.

Usually input fields automatically keep a history of what I've typed into them, but the wallet address one doesn't.  The 'Subject:' box at the top of this page I'm using to compose the message, for instance, if I blank it out and type 'R', I get a big list of previous subject lines I replied to.

I went to bet 40 BTC to win 42 BTC (then changed my mind before submitting).

"You have a 85.87% chance of winning. In order to win, your guess must be within 4294 of our chosen number."

If I guess 4294, then everything from 0000 to 8588 is within 4294 of my guess.  That's 8589 winning guesses out of 10k, or 85.89%.

You report 85.87%, which makes me think that you're not counting 0 and 8588 as being winning numbers.  That's going to piss somebody off one day...

Either that or the displayed percentage is just a little out.

It just told me: "Hash or timestamp has been tampered with.
To keep things fair, games are only valid for 2 minutes, please generate a new hash."

I didn't tamper with anything.

I did hang around longer than 10 minutes though, looking at the odds it offered me.

I'm guessing the 2 minute limit is to prevent me brute-forcing the hash to find out what the number is.  How about making the 'junk' an extra character or two longer, and having the games valid for an hour instead.  2 minutes arguably isn't long enough for a new user to find their wallet address, copy it into the form, get familiar with how the site works, decide what numbers to type into the boxes, etc.

And instead of making me click a link to "generate a new hash", how about just generating a new one right there and then.

Also, I keep having to re-type my wallet address.  The input field isn't remembering the one I used last time, like input fields usually remember things.  I keep having to go back to my wallet and copy/paste it each time.

Oh, and where's that dime you promised me?  

I'm not sure about having the odds get worse the more I bet.

That encourages me to split my bet into 10 smaller bets to get better odds.

The (?) after the hash at the bottom of the page is underlined, making it look like it is a hyperlink.  That caused me to click on it, which in turn prevented the tooltip from popping up.

It appears that I'm meant to just hover over it, not click it.  Removing the underline would help prevent people clicking on it I think.

The string that was hashed for me was:


I would suggest putting spaces in there between the junk, the date, and the guess, just to make it more legible.

Is 9999 'within 10 of' 0001?

ie. do the numbers wrap around?

Edit: I can answer that now.  I lost my bet, with "Guess: 3536, Number: 8930, Distance: 4606, Max Distance: 3536".

8930 - 3536 = 5394
13536 - 8930 = 4606

So they do wrap around - the reported distance is the shorter of the two.  It would be worth mentioning that (more prominently, if I missed it).

I sent 0.05 BTC, and guessed 3536.

I'm clicking "After you pay, click here to see if you won!" but the same page keeps coming up over and over.  Is it waiting for confirmations, or did something go wrong?

Edit: it took over 5 minutes to let me know if I won, but did eventually tell me.  "Guess: 3536, Number: 8930, Distance: 4606, Max Distance: 3536".  I lost my 70.71 shot...

It says the game will be valid until Mon Jun 11 2012 22:56:54 GMT-0700 (PDT) and it's currently only Mon Jun 11 21:03:06 PDT 2012, so it's valid for almost 2 more hours.

Or "the game ends as soon as there are x blocks without a deposit" - then the game can go on for a long time, but stops when people stop playing.


I don't see the trust issue in that case.  You don't know anything that we don't.  Every player can see all the bets, and can decide to bet more or not, just like you can.  We have to trust that you'll pay the winner, but we have to trust satoshidice to pay their winners too.

This article is related, has lots of maths, but may give you some ideas:

  http://www.chapman.edu/ESI/wp/GeneralizedTullockContest-Sheremeta.pdf

See also: http://en.wikipedia.org/wiki/All-pay_auction

I'm not trying to haggle.  I don't even play much roulette.  I've been trying to reach an understanding of the mathematics involved.  You were under the impression that by betting big a player would have an advantage playing 30x deposit+bonus at 5% house edge roulette, but I didn't think that was the case.  Whenever I see something that seems wrong to me, I try to understand it, and point it out.  Sometimes you find that helpful, other times you get mad, and I have a hard time knowing which of the two will happen until it does.

I'm sorry if by arguing the point I upset you.

I don't know if my argument is right, and I don't know what you should do.  I just don't see any reason to not allow 100% of roulette bets to count towards WR.  It doesn't affect me, since I play very little roulette, but I like things to be right, and since roulette has a high house edge it doesn't seem right to make roulette bets count less than other games.

If there's a good reason why they need to count less, I'd like to learn of it, but asking for 150 times playthough with 5% edge (ie. a 750% expected loss) seems to me like the player has about a 0% chance of clearing their bonus before losing their deposit.

It's 14%, not 27%.


Even if we assume the player wins 27%, I think the house still wins:

The player either loses his deposit (-20, 73% of the time) or keeps the bonus (+40, 27% of the time).  You can't count the $20 deposit as "winnings".

Expected profit for player = 40*.27 - 20*.73 = -$3.80.  That is, the player can expect to lose 19% of his deposit.

But in the scenario you proposed, where the player is betting $25*2 each spin to get through the WR without losing, when the player loses they lose $50, not $20.  They're betting $25 on both red and black.  You win $50 73 times and the player wins $40 27 times.

Then the expected profit for the player is 40*.27 - 50*.73 = -$25.70, ie. he can expect to lose 128.5% of his deposit.

Now if we use the real value, 14.28%, the expected profit is 40*.1428 - 50*.8572 = -$37.18, or 185.7% of the deposit.  If the player played all 36 spins, you would expect him to lose more, but he's going to stop playing and forget about the bonus as soon as a zero comes up.

Note that the player could bet $50 on red and black and get through the WR in just 18 spins, with 37.79% chance of success, but a 62.21% chance of losing $100.  The expected profit then is 40*.3779 - 100*.6221 = -$47.094 or 235.5% of initial deposit.

In no situation is allowing 100% of roulette bets to count towards WR going to be a net loser for the house (unless you're comparing it to the current situation, of course).  There's no getting away from the >5% house edge, which when multiplied by 30 gives >150%.


So 0.6% would work too?  Anything over 0.51% means you end up more than tripling your deposit.  I was confused about the 0.71% - it seems like of arbitrary, so why write it to 2 decimal places...

No, you're right.  He foresaw it, but discouraged it:

I've never understood the problem with hedging.  If I bet red and black, I lose 2 units every 38 spins (when the two green zeroes come up).  That gives a house edge of 5.26%, the same as if I was only playing red (where I win 18 units and lose 20 units every 38 spins).  I can expect to lose the exact same amount whether I play red every spin, or red and black every spin.  Is it the reduced variance that you don't like?  I'd have thought that the reduced variance works in your favour.  Losing over 5% on each of 30 playthroughs means the player loses 150% with high probability - ie. is very likely to lose the whole bonus and more before clearing it.  Only counting 20% of bets towards wagering requirements means the player will likely lose over 750% of their bonus before clearing it.


Doesn't that depend on which bonus they are trying to clear?  And I'm not sure what "beating the house edge by 0.71%" means.

Do you mean "reducing the house edge by 0.71% of 0.51 to give 0.506379%?  Or "subtracting 0.71% from 0.51% to give -0.2%"?  I don't see where 0.71 comes from in either case; I guess it depends upon an arbitrary value for "significant".

You used "Martingale" and "safety" in the same sentence. 

I signed up as user 'dooglus'.

There was a thread on these forums not long ago about integrating bitcoin with tahoe lafs.