I tried signing up but didn't get very far.  Observations so far:

* "Send Money To Another Country ?" has a space before the question mark which bugs me.

* https://www.bitcoinary.com/en/users/sign_up selects the 2nd input field by default, "email".  Why not start at the top?

* After filling in the signup form and submitting it, all I got was "We're sorry, but something went wrong."  Would be nice if you told me what kind of a something.  Was my 32 character password too long for you?  I tried again with 'fred' and it gave me a properly formatted error page, with "is too short (minimum is 6 characters)".  I tried using shorter and shorter passwords, but even a 6 character password still gives me the same "something went wrong" screen.

Expected value is the sum of each possible return to you, multiplied by the probability of it happening.

There are 2 possible outcomes:
  * you win 1.343 times your money, with probability 0.732422
  * you win 0.005 times your money, with probability 1 - 0.732422 = 0.267578
(because even when you lose you get a little something back - just to prove they got and processed your bet).

So the expected return is (0.732422 * 1.343) + (0.267578 * 0.005) = 0.984980636 times your money.

That means you lose (1 - 0.984980636) * 100 = 1.5019364% on average per bet.

I wonder if they're waiting to have enough confirmed funds to pay winners now to avoid the other problem that's been going on of paying out with funds that never confirm.

Well, no.  I read that at one point it was resolved.  I guess it's an ongoing issue but it does seem that everyone gets paid out eventually.

I think you're missing something in the maths.

The house edge is the same whatever the bet size, so the bet size doesn't hurt the expected house win rate, it just affects the risk of ruin.  And the risk of ruin is managed by dynamically adjusting the maximum bet size to only accept bets they can afford to lose.

If they only accept bets up to 100 when they can afford to (which makes sense) they still make 1.5% (or whatever the edge is these days - it's advertised differently in different places) on average each time they take a bet for 100 BTC.

By the way, here's where I saw it say the house edge is less than 1%:

To be fair, blockexplorer has only ever - uh - explored blocks.  And so unconfirmed transactions, not being in a block, aren't shown.

It's also sometimes quite a few blocks behind.  But is there other stuff is doesn't show that it should?

This is discussed at length in the main satoshidice thread, but the summary seems to be:

Someone bet 100 BTC and won 1000 BTC, taking a large percentage of the site's coins.  This meant that winners were being paid out with unconfirmed coins from losers, and some of those coins never confirmed, meaning the payouts couldn't confirm either.  This wouldn't usually happen since there are usually enough confirmed coins in the wallet to not have to use unconfirmed coins from other players.

Everything was resolved in the end.

The input looks like it was a donation payment from the P2Pool software.  For some reason P2Pool blocks always use "pay to public key" outputs for the donation address (see https://en.bitcoin.it/wiki/Transactions#Transfer_to_IP_address) rather than the more common "pay to bitcoin address" type of output (see https://en.bitcoin.it/wiki/Transactions#Transfer_to_Bitcoin_address).

It looks like nelisky may be running his own separate P2Pool with the donation address changed to his own address.  Each of the blocks he found (see http://blockexplorer.com/address/1MgH8iNkdZrDW8rsfzifDzsqJnCjBVyRBV) have very few recipients of the block reward meaning he's not part of the original P2Pool.

Yes, thanks.

I was thrown off by the phrase "weekend balance".  The word 'balance' made me think of account balance, not weekend profit/loss.

Can you please clarify what this means exactly?


Does it mean that if I hit $25 at any point during the weekend I qualify?
Or that I must be over $25 all weekend, and if I drop below then I'm disqualified?
Or that I must start the weekend with at least $25?
Or end it with at least $25?
Or something else?

Will the leaderboard show players who aren't qualified such that I may find myself trying to catch someone who isn't really even in the game?

So many questions! 

Very nice, thanks.

If you're aiming for minimal, none of set_privkey(), set_pubkey(), get_privkey() seem to be used at all.

I received a PM from a WBX user:


I hadn't thought of these kinds of liabilities.  So not only to we have BTC and AUD balances on the site, we also have withdrawals that were never really paid out, and now deposits which were never credited to the WBX accounts.

I've no idea how it's possible to get an idea of the total liabilities when I don't have any handle on the total amount of unpaid withdrawals or uncredited deposits.

Andre, you really need to step up and provide this information at the very least.

I just followed your instructions and it worked for me.  I had made a mess of my wallet and failed to tidy it up using bitcointools.  Following your instructions I was able to remove the double-spend transaction and get the wallet back to normal.

I already had Python installed, and am running Linux, so I've not tested it on Windows.

I've been unable to get bitcointools to do what I wanted.

https://bitcointalk.org/index.php?topic=29765.0 seems to be a similar issue to Jeremy's, but the solution that was found there is on an old git branch that never made it back to the master branch.

Yes, that's incorrect.

See http://bitcoin.stackexchange.com/questions/3173/what-information-does-a-wallet-contain

It is rebroadcasting the transactions over and over.  From the log Jeremy put on dropbox:


I'm wondering if the transactions are invalid for some reason, and so his peers aren't accepting the transactions.  I don't know why that would be though.

I think that using Gavin's bitcointools to mark the 5 rogue transactions as 'unspent' should fix your problem.

Get it from here:
  https://github.com/gavinandresen/bitcointools

Then something like:
  python fixwallet.py --tweakspent ccf585bcfed1283d11b5012c5f3f3429c5695a8d2280287c44e605223b6318b5

I'll experiment here to see if it works, and report back.

I think the bigger concern, that I've not seen addressed, is that MtGox may be insolvent.

What is the cause of the problem affecting these "few people".  Is it that you don't have enough US dollars to pay them?



Well, is he?

Ha!

That's the first thing I thought of when I saw this thread, but forgot it in all the technical details.

I sent you a PM.

I've since discovered that it's possible to test whether I've cracked your password without having any of your encrypted keys.

If you can just give us your encrypted master key, the salt, the number of iterations, and your best guess at what you think the passphrase is, we can attempt to find a passphrase that decrypts the master key and know when we've done it, but without getting access to any of your funds.

To get the information we need, get a copy of Gavin's bitcointools: https://github.com/gavinandresen/bitcointools and make the following change in wallet.py:


Then close bitcoin-qt and run:


from the bitcointools directory.

It should print a single line, like this:


Using this, and bitcoin-qt's crypter.cpp, anyone can verify that the passphrase is 'abc'.  But since nobody has a copy of the wallet I encrypted with 'abc', nobody has access to the coins in the wallet.