There main difference between this case and Madoff, is that the trades actually happened, and people were making their own trading decisions.
I don't know what the court
will say, but they
could say that the people were trading non-existent coins, so the balances are meaningless.
Actual deposits and withdrawals are numbers that the court can verify with the banks and in the blockchain. The trades inside MtGOX are recorded only in Mark's database, and any information that comes from him must be treated as a mere hint that needs independent evidence to be trusted. The database may have been doctored to give large bogus balances to some clients.
Nope, trades were published in real time, and more or less complete records can be retrieved from many locations. The only unknown is which accounts the trades belonged to. Faking account balances afterwards by switching which account did the trade will require a lot of work to make everything match up, and even more to seem credible. Not sure if it will be possible at all if everyone will be allowed to verify their own data. People have already been able to verify their balances for a while, and I haven't heard of an complaints. An audit will trivially reveal if the trade data matches the balances caused by deposits and withdrawals.
That sounds like a good way to reach a binary decision about out whether the database is correct or incorrect; You could look at the feed, the account history, the blockchain and the account balance and see if everything matches. If it does, problem solved.
But what if it doesn't? If any data is missing or unreliable, eg the database has been hacked at unknown times, the bitcoins have been sent to unknown addresses and not properly logged, the code sometimes put the wrong information in the database or the feed, customer support made transactions that didn't show up in the feed, etc etc, it may not be possible to work out _what_ about the database is incorrect. Once you have any condition where money fails to get logged or money gets logged but fails to get spent as described, it becomes very hard to work out what may have fallen through which hole. It takes reasonably competent design to create good audit logs, and we're talking about an operation that didn't even use source control, so not only was their design probably incompetent and possibly deliberately incompetent, it won't even be possible to track which mistakes the system was making at what times.