SatoshiDICE.com - The World's Most Popular Bitcoin Game

[evoorhees]

dude I'm telling you people are cheating, you can raise your edge to 50% and it wouldn't matter, you are only donating your btc to the cheaters.

Site is back in the black. I've had a couple statisticians review this now, and all systems seem normal. There are some interesting phenomena I'll need to address and tweaks I'm going to make, but the site is running as it should. I do not believe anyone is or can cheat, though certainly it's a risk I take.

We're still making some technical improvements to make performance better, including a database upgrade and a new way to handle some of the scripting. Onward!

Also, LOL http://blockchain.info/charts/n-transactions

[1714057761]

1714057761

[ccliu]

dude I'm telling you people are cheating, you can raise your edge to 50% and it wouldn't matter, you are only donating your btc to the cheaters.

Site is back in the black. I've had a couple statisticians review this now, and all systems seem normal. There are some interesting phenomena I'll need to address and tweaks I'm going to make, but the site is running as it should. I do not believe anyone is or can cheat, though certainly it's a risk I take.

We're still making some technical improvements to make performance better, including a database upgrade and a new way to handle some of the scripting. Onward!

Also, LOL http://blockchain.info/charts/n-transactions

money money money 4 u

[Otoh]

Quick update:  Since our house run of luck has been terrible (though we still believe, within the bounds of statistical possibility without cheating), we are increasing the house edge just slightly, up to 1% from .818%. We may bring the edge back down in the near future. We need to do this as the house pool is not large enough to enable people to be placing the size of bets they'd like so this should help with that aspect.

It's probably due to people using a progressive betting strategy. This doesn't reduce your edge, but it does change the flow. What will happen is that instead of having a small trickle of profit, you will have a small trickle of loss with the occasional spike of huge profit. You will need to increase your reserve, but you should be quite happy to do that.

+1

depending on the strategy the punters may enjoy modest & accumulating wins say about 98% of the time, but play long enough &/or enough of them playing that, then martingale will eventually bitch slap their collective wallets in the house's favor

[Electricbees]

http://blockchain.info/charts/n-transactions


I lol'd a lot... Great game, by the way!

[bitlotto]

dude I'm telling you people are cheating, you can raise your edge to 50% and it wouldn't matter, you are only donating your btc to the cheaters.

That seems impossible to me. I don't see how people could possibly cheat.

Whoever knows/steals the secret being used to figure out the winner could cheat.
SatoshiDICE would keep this highly secure and secret, so I doubt anyone else has it.

[mem]

I have been doing some sims myself, perhaps swap the secret less ?

I only have a small amount of recieve txid's (from my own wallet) to sim against, but with that limited data set of 40 (lol) it does seem that some secrets will weight the hash results in favour of the gambler for awhile before swinging back to the house.

Sample output:

Code:

step 37: rand: -2, hash: -4
step 38: rand: -3, hash: -5
step 39: rand: -4, hash: -6
step 40: rand: -5, hash: -5
step 41: rand: -4, hash: -6
step 42: rand: -5, hash: -7
step 43: rand: -4, hash: -6

Its fairly simple, it works through all the txids in your wallet (can be provided in a file), then compares a random results vs a generated result using the same method you do.
It then increments (play wins) or decrements (house wins) the count you see (negative means results in houses favour).

If you want a copy of the code to run against your recorded tx's for a basic sim, send me a pm.

[fireduck]

I'm pleased to announce that the Satoshidice backend processing has been upgraded to a new system that uses a distributed transaction database stored on Amazon DynamoDB.  It is thus far faster and importantly should be much more stable and less prone to inconsistencies taking down the system for hours at a time.

I have some further work to do.  The next step is to have a few peer nodes that do nothing but pick up broadcast transactions and put them in the database.  This will make the system quicker and more reliable in how it picks up pending bets.

[mem]

I'm pleased to announce that the Satoshidice backend processing has been upgraded to a new system that uses a distributed transaction database stored on Amazon DynamoDB.  It is thus far faster and importantly should be much more stable and less prone to inconsistencies taking down the system for hours at a time.

I have some further work to do.  The next step is to have a few peer nodes that do nothing but pick up broadcast transactions and put them in the database.  This will make the system quicker and more reliable in how it picks up pending bets.

nice work there

[Kitemike]

 I really enjoy satoshidice, only wish it wouldn't hang for an hour occasionally. When it’s clicking along with almost instant replies it’s really great.  I even wrote a little satoshidice simulator to try out some betting strategies.  You can follow my signature link if you’d like a copy.  I’m (not) sorry to say I’m ahead so far 

[Stephen Gornick]

I even wrote a little satoshidice simulator to try out some betting strategies.  You can follow my signature link if you’d like a copy.

I just want to add, as a public service announcement, to anyone who might not be aware of the risks.

When installing software you are trusting that the software doesn't include a keylogger, a wallet.dat stealer or any other malware.

That's one thing that is so awesome about SatoshiDICE -- no software necessary.

[ccliu]

Props to satoshi dice for making the site much faster!

However I miss the days when I would get the tx back near instantly, hopefully that will be soon.

[piotr_n]

However I miss the days when I would get the tx back near instantly, hopefully that will be soon.

Yeah. Right after the house fee reaches 10%...

[Kitemike]

When installing software you are trusting that the software doesn't include a keylogger, a wallet.dat stealer or any other malware.

I agree, don't trust any software on the web.  Always download and install into a test machine or virtual machine.  Once you have the proper sandbox, you can install even known virus ridden crapware without any worries. 

[evoorhees]

Hello SatoshiDicers!

I'd like to make you aware of some changes to the site today.

Due to the very low house edge, and resulting high variance in our balance, the site is a fairly risky liability thus far. To date, it is in the red slightly, though it is fully within the expected variance. This has been a bit nerve-wracking for me as the owner with large swings up and down, and I need to run this as a reliable and long-term business, not as a gamble.

So, after discussions with a few people, the house edge has been changed from the current 1% to 1.5%. This means players now have an expected return of 98.5% instead of 99% (payout to winners of 97.5% plus the payout to losers of 1%)

Also, the fee for return payments will now be subtracted from these payments (all payments will basically be .0005 btc less). However, our fee is occasionally higher than .0005, based on our inputs. In these cases, we will pay that additional amount (the player will not bear any cost more than .0005 for the return fee).

With these changes, the variance to us should be calmed, our ROI marginally higher, and I can be comfortable about keeping capital in the system indefinitely. If the wallet attains a healthier balance in the future, I will consider bringing the house edge back to 1%. For now, I hope the new rules will give players further confidence that SatoshiDice will be around a long time (I have a few exciting development plans for it as well...). I've been extremely pleased with the response and interest thus far, and I really appreciate all the feedback and active participation.

Also, the Ghost of Satoshi has blessed all players with good luck for the remainder of the day.

[mem]

Roulette is only a 47.37% probability of winning (red/black) with a house edge of 5.26% I dont think people will mind if yours is now 1.5% house edge

[payb.tc]

Roulette is only a 47.37% probability of winning (red/black) with a house edge of 5.26% I dont think people will mind if yours is now 1.5% house edge

yeah and on that note, when is satoshiroulette.com launching?

[SkRRJyTC]

I'm pleased to announce that the Satoshidice backend processing has been upgraded to a new system that uses a distributed transaction database stored on Amazon DynamoDB.  It is thus far faster and importantly should be much more stable and less prone to inconsistencies taking down the system for hours at a time.

I have some further work to do.  The next step is to have a few peer nodes that do nothing but pick up broadcast transactions and put them in the database.  This will make the system quicker and more reliable in how it picks up pending bets.

It does occasionaly hold back a bet for ~15 minutes but most of the bets I have placed today have come back in ~ 20 seconds

Great job, keep up the good work!

[kano]

Seems to have died about an hour ago ...
(Web page isn't updating)

Edit: back to life now
(10 minutes later)

[fireduck]

A few new things:

A list of secrets used for previous days:
http://satoshidice.com/secretlist.php

A way to lookup your transactions:
http://satoshidice.com/lookup.php
(TODO: add other selection criteria, add lookup by address)

[Stephen Gornick]

A list of secrets used for previous days:
http://satoshidice.com/secretlist.php

Kind of jumped the gun there if you were following your specs:
"After a day has been over for at least 24 hours the system will release the secret used."
 - http://satoshidice.com/secrets.php

Though I can't think of any use case where it matters as long as SatoshiDICE itself is the authority as far as when a wager is recognized, and thus there is a zero percent chance of getting a wager that used a revealed secret to cause a transaction to be recognized as if it occurred on the prior day (even if it ended just a second earlier).