The real proof is provided using sha256 hashes. The site commits to a specific server-side secret before the player does anything by publishing the hash of that secret. Then the player picks their own client-side seed and both seeds are used to generate the rolls. When the player is done playing he is shown the server-side secret. The player can hash that secret and verify that its hash matches the hash he was shown back at the start.


Not sure what you mean by that. I'm a programmer and I know that there's no way the server can cheat the player. I think the problem here is that either you don't understand how it works, or you do understand how it works but want to try to persuade others that it doesn't work for some reason. I've seen a few "traditional" casinos recently trying to tarnish the reputation of provable fairness so that they can justify refusing to adopt it.


The algorithm is public. The code isn't. The outputs of the algorithm are public. I think you're suggesting that there is some way we could be running code which implements a different algorithm but which gets the same results as the published algorithm? If so, would it matter? All we need to prove is that your rolls were predetermined, by you, in a way that we couldn't affect. And that's what we do prove.


I agree. It's an amazing innovation. For the first time it is possible for players to be mathematically guaranteed of a fair game. There's nothing to "fall for".


The site publishes the hash of the server seed before any play happens. ie. before any result. And before the player picks their own seed. That's the proof - because later the site publishes its server seed and the player verifies that the published server seed does in fact hash to the hash that was published in advance.

The only way the site could cheat the player would be if they could find multiple seeds which hash to the same thing, and as far as I am aware that has never been done. Finding hash collisions is meant to be very hard. And when that is no longer the case, new hash functions are developed to restore that property.


The server shows you what you need to see to verify that everything is fair, but doesn't show you enough for you to be able to cheat. How else would you have it?


Thanks for quoting the posts before they were deleted. It's useful to see the various ways the people who are scared of provable fairness use to try to discredit it.

Note that staking is random, much like Bitcoin mining. If your expected time to find a block is 3 days and you go 2 days without finding one, it doesn't mean you will find one on the 3rd day. It just means that you have been unlucky. You don't make "progress" towards staking, and your expected time to find a block doesn't decrease over time.


The chance of each output staking is proportional to its value, yes.

According to my estimates, each 450 CLAM you have will stake about once per day. So if you have 100 CLAMs you can expect to stake every 4 or 5 days. And so on.

Version 1.4.16 added a dynamic peer list. It will automatically connect to the most reliable peers so you shouldn't need to copy/paste lists of IP addresses:

    https://github.com/nochowderforyou/clams/releases/tag/v1.4.16

And my bootstrap file here is usually no more than a few weeks old:

    https://bitcointalk.org/index.php?topic=623147.msg9772191#msg9772191

That post also tells you where to find a series of partial bootstrap files, so you can download just the pieces for the blocks you're missing and stick them together yourself into a custom bootstrap file.

Oh, yeah. Duh.

iven ised? I can't begin to guess what you're trying to say there.

It's possible they marked your withdrawal as 'processed' without actually sending anything. It's cheaper that way, because that way they get to keep your coins. Maybe if you complain loudly enough they will pay you, and maybe they won't.

The fundamental problem here is that you sent your coins to a probable scam operation. You shouldn't be surprised if they refuse to return them to you.

This is honestly the best you can do?


Let me help. The following is closer to English. Is it what you intended?


You will always get 20 rolls below 50 if you play at 70% for long enough. On average it will take about 40 rolls to get 20 under 50, since half of them will be under 50. I can't figure out what you mean by "opposite sequence of the multiplier" or "the method I choose". I expect you could clarify your terms if you wanted to. Am I to assume that your "dice is rigged lol" argument was simply trolling and when challenged you want to run away? It's OK if that's the case.

You're asking why you don't get such long losing streaks when you play with a higher chance of winning? When you have a 70% chance of winning each roll, you only have a 30% chance of losing each roll, and so the chance of getting long losing streaks is lower. I somehow don't think that's really what you are asking about though, because it seems too obvious.

Please try to state your question clearly.

Did you ever play at Just-Dice? If so, tell me your user number and I will take a look at your betting history. We'll see how many times you rolled under 50 and over 50, and see that they're probably about the same, for instance. Maybe with a concrete dataset we can more easily pinpoint the problem you're having.


Forget about doubling. The stake doesn't change the outcome. Any 30 losses at 50% is just as unlikely as any other, no matter whether people are doubling their stakes or not.

That article is very out of date. Take a look at how provable fairness works today and you'll find that the points made in that old reddit post no longer apply. In particular modern sites don't change the server seed every roll, and allow the user to pick their client seed *after* the server picks the server seed, and so no manipulation is possible.

The offer of the 1 BTC or 80 CLAMs still stands if you can show me any single 30 reds in a row at 50% on Just-Dice.com.

It's true that once we saw a losing streak of length 32 at 49.5% on Just-Dice. People play at 49.5% quite a lot. People almost never play at 50%, and that's why we didn't see a losing streak of length 30 at 50.0% yet.

I'm not stating my opinion here. I am stating facts. I am also offering you free money if you can back up your statement that you always get 30 losses in a row at 50%.

What does "there is 30% chance of winning at the same 2x multiplier for 100000 rolls" even mean? Every 50% roll has a 50% chance of winning. All your rolls are predetermined. You get to pick 'hi' or 'lo' after the roll has already been decided, and so like calling heads or tails on a coin toss you have a 50% chance of calling correctly. I don't really understand how you can think it works any differently. If you can explain in more detail how you think the house can change the chance of you guessing a 50/50 result correctly, then maybe I can figure out where you're going wrong in your thought process.


I try to only quote the parts of messages that I am replying to. I dislike the practice of quoting a big long message and only addressing part of it. But as you wish, here's your whole sentence:


The 2nd part of the sentence claims that you never lost more than 10 bets in a row when playing any chance other than 50%? If that's true, I guess you must be either not playing other chances very often, or when you do, you play at a relatively high chance. If you play at 1% you will almost always get losing streaks longer than 10 in a row for instance. I'm not sure what you want me to say about this 2nd part of your sentence. My main criticism was that you claim to "always get 30 reds at 50%", and that isn't true. To show that it isn't true I offered to pay you 1 BTC if you can show it happening even once on Just-Dice. Then I extended the offer to everyone else as well.


We've had about 3/4 of a million bets in about 3/4 of a year. So we should expect to see 30 losses in a row at 50% about once per year, assuming everyone bets at 50%. But they don't. Nobody bets at 50%...

There is a theory that demand driven by Just-Dice.com has caused the recent spike in the price of CLAM. Most of the CLAMs are in the Just-Dice bankroll, earning around 0.2% per day rather than sitting in sell orders on an exchange earning nothing. That means there's not much supply for people wanting to buy, and so the price tends to increase.

That's right!

If you want to be clicking the 'hi' button for the next 10 years to win a free 1 BTC, be my guest!

That's how often losing streaks of length 30 or more come up if you play at 50.5%.

ashfaq is talking about playing at 50%, and so that's the chance you have to play at to qualify for the prize I'm offering.

It's 1 in 1,073,741,824, since losing at 50% is a bit easier than losing at 50.5%.

Showing my working:

>>> 1 / (1 - 50.5/100) ** 30
1451590214.9428499

>>> 1 / (1 - 50.0/100) ** 30
1073741824.0

Note: this is why I wrote this a while back:

I was specifically making the offer to ashfaq, but sure, why not. I'll open it to anyone.

The first person to roll 30 or more losing bets in a row at 50.0000% on Just-Dice gets a 1 BTC or 80 CLAM prize, no strings attached.

PM me on this forum to claim your prize, telling me the betid of the first loss in the sequence, set your emergency withdrawal address on your Just-Dice account to a BTC or CLAM address depending on how you want the prize, and I'll send it there.

I don't care what stakes you use. You can flat bet zero amounts if you like. Then you'll see just how rare it is to see 30 or more losses in a row at 50%, and be less likely to believe ashfaq's story.

http://worldcoinindex.com/coin/Clam is nice too:

You don't. Nobody ever rolled 30 losses in a row at 50% at Just-Dice. If you can provide a legit screenshot of it happening even once (edit: at Just-Dice) I will send you 1 BTC or 80 CLAMs, your choice.

I won't reply to the rest of your post, because it has already all been expertly refuted by others.

If you are regularly seeing losing streaks of length 30 at 50% on any site, let me know. That would seriously need looking into since it should only happen about once per BILLION rolls.

I just brought it back for you. Reload the page and you'll be able to see it.

Don't believe him. This withdrawal likely never happened.

Not long ago he said he would prove that he had withdrawn over 1000 BTC from scrypt.cc, but the proof never materialised:

Check this bet out. "dental" was aiming for the 10,000 to 1 shot, and hit it on his 331st attempt:



He won ~1000 CLAMs for a 0.1 CLAM stake.

Yes, I changed a bunch of stuff. Most functions and variables aren't exposed in the window object any more.

What is the particular error you're seeing? Hit F12, go to the 'console' tab. Anything there? Or - which bot are you using?