[rwallace]

My gut feeling, speaking as an outsider reading the news, is Karpeles was never dishonest, just hopelessly out of his depth and too stubborn to ask for help.

In the normal course of events I also would expect money disappearing over that long a timescale to have to have been embezzlement by an employee, and I would certainly consider that a plausible hypothesis here. On the other hand, reading the IRC chat transcript where Karpeles talks about how he's been under so much stress that it's ruining his health, it's obvious he made the classic mistake of trying to work so many hours he couldn't think straight.

So I'm guessing either he delegated the accounts to someone with no oversight and it was an inside job... or he refused to delegate, which meant the accounts were in the hands of someone too tired to be capable of coherent thought, which amounted to the same result as nobody looking at them, and an outside job would suffice.

Moral of this story for customers: don't risk money you can't afford to lose.

Moral of this story for founders: don't try to be a hero and do more than one job's worth of work, especially in a domain where a screwup may cause significant harm.

[mattm]

I've spent way too much time following this and I agree with your thoughts as well. Karpeles seems to personify the phrase we know in software development - that he knows just enough to be dangerous but not enough to be competent.

I was stunned at this blog post (http://antonopoulos.com/2014/02/25/statement-on-mt-gox/) where Andreas Antonopolous said:

"During this time, Mark Karpeles was active on the forums and developer boards and appeared to be implementing fixes to Gox software to address Tx-Mal."

He really should not have been looking for solutions himself.

[jnbiche]

>He really should not have been looking for solutions himself.

Actually, I think at this point he had waited too late to ask for competent help. Here we have a custom Bitcoin client, written in what is probably a tangled hairball of PHP. The Bitcoin protocol is already fairly complex, and even Bitcoin software written by competent developers using modern software development methods can be tricky to understand (see btcd for an example of a well-engineered client).

How many engineers could fix that in a reasonable period of time?

Add to that the fact that there are only a handful of competent programmers in the world who also are experts at the Bitcoin protocol, and I'm not convinced that he could have convinced any of them to take on the task. Most of them are already wealthy, and they have a lot of reputation to lose by trying (quite possibly in vain) to fix a dying company.

Mark's biggest mistake was in trying to do this all himself once Bitcoin really took off. So yes, a valuable lesson for entrepreneurs.

[jnbiche]

Thank you for one of the few rational comments I've read about the whole ordeal. And yes, as a Bitcoiner since 2011 (who last dealt with Gox that same year), your assessment of Karpeles is spot on. My impression is that he's not dishonest, just the classic Dunning-Kruger programmer we've all worked with.

The hard thing about accounting for these Bitcoins is that we keep track of them via addresses, but Karpeles was robbed at a lower level of abstraction -- via outputs. So he's probably struggling with that.

EDIT: I should have written that my impression is that he's not a thief. I suspect he has been dishonest about what he's known about his company's situation, in the sense that he's probably lied by commission or at least omission. But I don't think he stole anything.