Non-Verbal Analysis of statements by Mark and Adam (MtGox)

[bitsalame]

(This post is a spin-off of the reply to the post: http://forum.bitcoin.org/index.php?topic=20501.msg256086#msg256086)
After watching this video:

http://www.youtube.com/watch?v=-0XvP841jaM

I promised I would make a non-verbal analysis.
For those who are late to the party: I am a psychology major who has been studying non-verbal communication for a while.
For the fans of the TV series "Lie To Me": the science actually exists in the real world.
I studied the researches of the "real" Lightman, Paul Ekman, who is the pioneer in proving that facial expressions are universal (which was heavily contended by anthropologists before the revealing results of Ekman's research).
I am studying to certify myself in FACS which stands for "Facial Action Coding System" which is the first effort in the world to codify the facial expressions, also created by Paul Ekman (who is the direct inspiration for the creation of the TV series)
With this knowledge I decided to evaluate the video and analyze Dave's and Mark's facial expressions and body language.

Disclaimer:
The following analysis is purely based on behavioral and expression cues in the video.
The Truth/Lie should be read as "high probability of truth", and it is not a categorical affirmation.
Some signs are ambiguous and need verbal confirmation (basically lie/truth analysis consists on coherence of verbal and non-verbal expressions)
Some special gestures can by themselves tell with high confidence if it is a deception, depending on the context.
I will separate both with + and - signs as level of confidence.
ie.: "+Truth" would be pretty high confidence of being truth. "-Truth" probably truth but lower confidence signs. -Lie: probable lie, but low confidence. +Lie: high confidence that it is a probable lie.
Ambiguous but worth "note taking" signs will be tagged with "?"
Also, something important I must mention is that not detecting signs of deception might mean that they are telling the truth, but it doesn't guarentee it.

I will not detail the rationale behind the analysis was performed as it would take me a whole essay to explain each assessment.
These assessments are backed up by current academic research in non-verbal communication.
I reiterate, this is backed up by science.
This analysis IS NOT based on Neuro-linguistic programming (NLP is Pseudoscience!)
If it is relevant, I will also list the emotional states revealed by gestures and facial expressions, some of them might be obvious to you, some others might not.
Also, understand that this analysis is not exhaustive. Every time I rewatch it I notice something new, but I am time constrained so I'll be focused mainly to their statement about Mt.Gox Account. (it is really time consuming, since I need to watch the whole hour video to see the baseline expressions to detect the "anomalous" expressions)
An accurate assessment requires a team of two or three persons doing it independently to confirm the observations.

Now lets begin:

Mark Karpeles: Owner of MtGox
Adam Barr: Mark's Partner (He's an open book, he should never play poker lol)
Bruce Wagner: Interviewer (Check how many times he says "MHM-MMM Okay" when doesn't give a shit/doesn't understand lol)

4:40
-Truth: Mark isn't Satoshi.
-?: he might had been in contact with Satoshi

07:40
+Adam doesn't believe that the leaked accounts are "wide open". (Probably Adam is thinking that FreeBSD MD5 is not "precisely" plaintext)

07:50
+?: They reported it to the japanese police.

09:55
+No signs of deception detected: regarding to the new server, more security, from scratch and new "state of the art technique".
+?: The main reason why they keep the old system is NOT "precisely" for investigative purposes only.

10:50
+Truth: new system built in-house

11:55
+Truth: They do keep transaction logs

15:31
-Truth: One Account compromised
++"Hacked" Account belonged to Mt.Gox?
Very Probably.
Detected disconformity microexpression in Adam. Also significant macroexpressions in Adam, who presses his lips for the first time; it might be interpreted as "I can't talk about it" and/or imply frustration/repressing emotions.
With what we have so far, I am inclined to believe that it was either Adam's account or the account was under Adam's responsability. (this might be a long shot... but I have a good scope! LOL)
This idea is supported when Mark starts his answer by laughing and looking at Adam out of the corner of his eye. Adam suddenly gets his eyes teary, glances away and bursts several microexpressions, repeating once more time a disconformity microexpression.
Probably Mark bitched a lot about it.
Update - Jackpot: I just found a particular gesture that is equivalent to a non-verbal confession (Thanks Adam! ^^ ).
Now I am "90%" (read as extremely confident) sure that the account was theirs.


20:40
++Truth: Not prosecuting Kevin (until they find any solid proof)
+Truth: they don't want to compensate Kevin (no shit, right? lol)

21:00
++Feelings: Whoa, Mark flashes a beauuutiful partial microexpression that reveals concealed DISGUST towards the idea of compensating Kevin. (LOL)

22:31
?: Bruce Wagner asks: The hacker had complete access to the system for 3 days, then he adds "That's gonna come out later". Both Mark and Dave basically say with their heads: "mmm... I don't think so"

25:25
+Truth: they can guarantee that the bitcoins and USD will be in the accounts when the system comes back and they are able to log in.

28:55
++Truth: Adam's facial expression allows us to predict Mark's answer. When Bruce asks: "...or are you gonna ask them to return them..." Dave makes a quick and short nod saying yes.
Mark ends up explaining that those who don't return the withdrawn bitcoins will get a negative balance in MtGox. (nice, lol)

38:32
+Truth: About not being contacted by any legal authority.

41:30
?Feelings: Mark seems to flash sadness when Dave explains about the people losing trust in MtGox. Weird eyebrows movement though, can't confirm.

52:17
+Lie!: When Bruce asks about "if it is a good idea locking the market price" and forcing price of the market. When Mark responds that they have no authority of manipulating the market which seem to be true, but when Bruce adds "well... that makes sense it's a free market, that's the whole idea", both Mark and Dave are unconvinced about it, nevertheless they nod saying yes.
Probably they think the free market is not really a good idea, but they are forced to do it anyway.

54:58
??: There is a LOT going on there but I need more time to dissect it carefully and interpret it properly. I don't see any clear sign of anything, but there is something doesn't quite feel right.
When Dave says: "We've been advised not to really mention any names right now" seems to be bullshit.
When Mark says that they want to be sure this never happens again seems to be honest.
I'll re-watch this part later.
(sorry guys, I am exhausted)

57:50
-Truth: probably truth that they don't know much about bitcoin7

My preliminary veredict: they have definitely something to hide (probably they directly fucked up and did something really stupid/embarrasing) and they don't want us to know, but they have good intentions, they seem to really have put measures to prevent the mistakes and toughen the system, and they are pretty much honest about the business.

This is all by now, I might review it later.
Cheers,
-bitsalame

PS1: My nick is not "bits-a-lame", it is "bit-salame", which in Spanish and Portuguese means "Bit Salami".
PS2: My general feeling is that I might trust in MtGox again, but they really need to get someone to take care of their PR, seriously.
UPDATE: I found another very supporting sign related to it the MtGox Account, it made it earn a second "+"

[1714924507]

1714924507

[1714924507]

1714924507

[1714924507]

1714924507

[1714924507]

1714924507

[bitdragon]

interesting... I had taken an interest some years ago in reversespeech (http://www.reversespeech.com/words_of_creation.htm)

Do you have any opinion/experience with that?

[bitrebel]

I agree 100%, but i'm not a Phych major. I just read people real well, from my experience.

This I agree with 100% especially. I noticed it too, and watched carefully for the reaction.
15:31
-Truth: One Account compromised
+Hacked Account belonged to Mt.Gox? Very Probably.
Detected disconformity microexpression in Adam. Also significant macroexpressions in Adam, who presses his lips for the first time; it might be interpreted as "I can't talk about it" and/or imply frustration/repressing emotions.
With what we have so far, I am inclined to believe (and this a long shot) that it was Adam's account or it was under Adam's responsability.
This idea is supported when Mark starts his answer by laughing and looking at Adam out of the corner of his eye. Adam suddenly gets his eyes teary, glances away and bursts several microexpressions, repeating once more time a disconformity microexpression.
Probably Mark bitched a lot about it.

They would have stated the account was not compromised, were it NOT compromised. It was most likely Mt Gox's account.

[bitsalame]

interesting... I had taken an interest some years ago in reversespeech (http://www.reversespeech.com/words_of_creation.htm)

Do you have any opinion/experience with that?

No, unfortunately Reverse Speech and any kind of Subliminal Messages are pseudoscientific.

[bitrebel]

interesting... I had taken an interest some years ago in reversespeech (http://www.reversespeech.com/words_of_creation.htm)

Do you have any opinion/experience with that?

No, unfortunately Reverse Speech and any kind of Subliminal Messages are pseudoscientific.

Also, I have heard David Oats, i think his name is...I do not believe he is really onto anything significant. He is on Rense Radio once a month. Sometimes it's clear, but other times I hear him reach and when he "reaches" it's for a definitive conclusion, he rarely plays anything and wonders what it is. He usually has his mind made up first. It's not scientific or even intuitive enough for my liking.

I give tremendous credence to the OP theory and study.

When someone lies, they tend to look away. That's the most obvious one people know intuitively. When you are ashamed, you bow your head or look down. When you are hiding something you make fidgety moves or nervous gestures, when confronted with uncomfortable truth, you get upset. That's the obvious stuff. It's subconscious too, but much more revealing than reverse speech, IMO

[davout]

-Truth: One Account compromised
+Hacked Account belonged to Mt.Gox? Very Probably.
Detected disconformity microexpression in Adam. Also significant macroexpressions in Adam, who presses his lips for the first time; it might be interpreted as "I can't talk about it" and/or imply frustration/repressing emotions.
With what we have so far, I am inclined to believe (and this a long shot) that it was Adam's account or it was under Adam's responsability.
This idea is supported when Mark starts his answer by laughing and looking at Adam out of the corner of his eye. Adam suddenly gets his eyes teary, glances away and bursts several microexpressions, repeating once more time a disconformity microexpression.
Probably Mark bitched a lot about it

Interesting, I'm inclined to think that if anyone owned half a million coins, then he wouldn't have stored them on mtgox, unless it felt like home

[bitrebel]

You know, come to think of it...and it occurred to me earlier....

What if Mt Gox, as an exchange, waited and waited for each good opportunity to buy bitcoins himself at the lowest price he could, and sell them off through his own exchange each time, using the earnings to buy up more bitcoins at the cheapest price, thus acquiring 500,000 bitcoins in a very short period of time?

He would not have done anything wrong, really, legally, but his advantage as an exchange would be considered something of a "conflict of interest", yes?

I'm pretty sure the 500,000 bitcoins were theirs. I never had much doubt about that.

[bitsalame]

You know, come to think of it...and it occurred to me earlier....

What if Mt Gox, as an exchange, waited and waited for each good opportunity to buy bitcoins himself at the lowest price he could, and sell them off through his own exchange each time, using the earnings to buy up more bitcoins at the cheapest price, thus acquiring 500,000 bitcoins in a very short period of time?

He would not have done anything wrong, really, legally, but his advantage as an exchange would be considered something of a "conflict of interest", yes?

Uhm, that is too much conjecture. If they make a video responding to that question, let me know

I'm pretty sure the 500,000 bitcoins were theirs. I never had much doubt about that.

Yeah, that was also my main hypothesis, but this is the first time I have some supporting evidence.
That's why I was so excited on watching this video

[davout]

He would not have done anything wrong, really, legally, but his advantage as an exchange would be considered something of a "conflict of interest", yes?

It would have been a violation of their own terms : "mtgox does not act as a counter party to any trades".

[bitrebel]

He would not have done anything wrong, really, legally, but his advantage as an exchange would be considered something of a "conflict of interest", yes?

It would have been a violation of their own terms : "mtgox does not act as a counter party to any trades".

What if Mark thought nobody would ever know because the bitcoin accounts are held by him, and bitcoins are an anonymous currency?

[nathanrees19]

the science actually exists in the real world.

The chance of a false positive is too high for it to be useful.

[bitrebel]

the science actually exists in the real world.

The chance of a false positive is too high for it to be useful.

That's why many things are taken into consideration all at once. 

[bitsalame]

the science actually exists in the real world.

The chance of a false positive is too high for it to be useful.

Are you making an informed opinion or just making a supposition from hunch?
All sciences have margin of errors, and there are many ways of minimizing false positives (and false negatives).
It is a very serious area of research, and there is strong empirical evidence too that supports non-verbal communication.
What it is known in academia it is usually mixed with crap in the books available to the public... (and definitely it is almost never like what they show on TV lol)

That's why many things are taken into consideration all at once.  

Precisely, seeing the whole picture it is essential to a proper analysis.
It is called the baseline. For example, when the question about MtGox's account is popped out triggers a series of gestures and microexpressions that aren't repeated ever again in the whole video.
It indicates that "something is going on", but we can't be certain the real reasons of that feeling. To find out that requires more digging.
There are many types of gestures that reveals different things each of them with different degrees of confidence.
It is mostly a probability game, the more coherent and supporting signs, the higher the confidence.

Cheers,
bitsalame

[LightRider]

Interesting analysis. This is why I'm always gratified to see people's faces recorded on video so that we can have access to this kind of data. I wonder if they will show their faces publicly ever again.

[aral]

It would have been a violation of their own terms : "mtgox does not act as a counter party to any trades".

I don't know what the hell they meant by that as it's not an otc trading platform, facilitating exchange between individuals, like bitmarket.eu is.  You deposit your money in their accounts and they exchange it, transferring funds between multiple individuals.  They are always your counter party.  The BTC and USD traded there were not 'real' but just mtgox credits, which they could have created and manipulated at will.  Some people have been suggesting that the hack was so bad that the 500k BTC perhaps never existed.  

[Horkabork]

While your analysis is interesting, I think that you know better than to attempt to practice a method of lie detection that is both very inaccurate and very misleading if used by someone who does not have a great deal of experience. I'm not saying that the science is inaccurate, but that you should know that YOU don't know enough about this to practice it on strangers for the benefit of the public. You're hurting your reputation as a science major. You're also presenting your "findings" of truth or deception as truth or deception, when, in fact, each gesture, expression, or apparent emotion can point to a multitude of other conditions. If you're going to do this, please at least list them with probabilities along with the dominant/combined findings.

By findings, I don't mean interpretation of gestures, expression, microexpressions, or apparent emotions but only the states that they could point to. You should know better than to attempt to interpret the communicated message behind a perceived behavior. Dr. Ekman found that expressions might be consistent across cultures in the states that they point to, but the communicated nonverbal message is a crapshoot. Expression analysis, for example, is troublesome when used by trained soldiers interviewing potential terrorists (they have to phrase questions very specifically). I mention this because your subject is a French(?) guy speaking English that he obviously can't fully understand, living in Japan, and unable to hear his mumbling, badly-miked interviewer half the time, let alone understand him.

If you really want to do this, run everything you did by a psychology professor and have him post. If you can't get anyone to do this, well, there's a sign. Part of good science is realizing that you have inadequate skills to interpret results properly. Especially if you're essentially diagnosing humans. Especially when crime is involved. Especially if you have a reputation to earn or keep. Especially if you don't have credentials or someone with credentials to provide evidence as to your skill level. Especially when you are making judgements that affect someone's reputation. Especially when you're trying to do something that only a fraction of clinical psychology grad students would attempt to do in a public forum for the sake of, well, entertainment. Dude, this is the psychology equivalent of a pre-med major doing knee surgery. Yeah, he might get the fundamentals taken care of, but it's not good for anyone involved.

And you should note that this criticism has little to do with the science. I likely don't know the science as well as you, but I do know that you don't know it half as well as you think you do, or else you wouldn't be doing it here. Even a well-trained psychology professional with years of experience in lie detection would be very reluctant to write what you wrote about an internet video. Well, at least without lots of prior footage of the subjects and/or the ability to ask the questions themselves.

You admit that you aren't trained, but your idol Dr. Ekman says that something like 0.2% of people have the ability to semi-accurately perceive microexpressions without training. I would wager that an even much smaller fraction can, untrained, translate those perceived microexpressions into a reasonably certain analysis.

Your decision to providing this analysis alone, regardless of the content or quality of the analysis, basically screams "I have poor judgement". Since apparently having a handful of undergraduate psychology classes is enough to allow you to nearly write libel, that's my psychological analysis.

Please for the love of god don't take what I'm saying here as tearing you up for the sake of being a jerk. I would absolutely love it if you could do an analysis like that but could back it up. I also greatly encourage you to continue your studies, but to be very self-critical and skeptical, because yours is a field that has humans as the lab rats, so mistakes are often costly.

[tcpip4lyfe]

Guys: GO OUTSIDE

[mouse]

While your analysis is interesting, I think that you know better than to attempt to practice a method of lie detection that is both very inaccurate and very misleading if used by someone who does not have a great deal of experience. I'm not saying that the science is inaccurate, but that you should know that YOU don't know enough about this to practice it on strangers for the benefit of the public. You're hurting your reputation as a science major. You're also presenting your "findings" of truth or deception as truth or deception, when, in fact, each gesture, expression, or apparent emotion can point to a multitude of other conditions. If you're going to do this, please at least list them with probabilities along with the dominant/combined findings.

By findings, I don't mean interpretation of gestures, expression, microexpressions, or apparent emotions but only the states that they could point to. You should know better than to attempt to interpret the communicated message behind a perceived behavior. Dr. Ekman found that expressions might be consistent across cultures in the states that they point to, but the communicated nonverbal message is a crapshoot. Expression analysis, for example, is troublesome when used by trained soldiers interviewing potential terrorists (they have to phrase questions very specifically). I mention this because your subject is a French(?) guy speaking English that he obviously can't fully understand, living in Japan, and unable to hear his mumbling, badly-miked interviewer half the time, let alone understand him.

If you really want to do this, run everything you did by a psychology professor and have him post. If you can't get anyone to do this, well, there's a sign. Part of good science is realizing that you have inadequate skills to interpret results properly. Especially if you're essentially diagnosing humans. Especially when crime is involved. Especially if you have a reputation to earn or keep. Especially if you don't have credentials or someone with credentials to provide evidence as to your skill level. Especially when you are making judgements that affect someone's reputation. Especially when you're trying to do something that only a fraction of clinical psychology grad students would attempt to do in a public forum for the sake of, well, entertainment. Dude, this is the psychology equivalent of a pre-med major doing knee surgery. Yeah, he might get the fundamentals taken care of, but it's not good for anyone involved.

And you should note that this criticism has little to do with the science. I likely don't know the science as well as you, but I do know that you don't know it half as well as you think you do, or else you wouldn't be doing it here. Even a well-trained psychology professional with years of experience in lie detection would be very reluctant to write what you wrote about an internet video. Well, at least without lots of prior footage of the subjects and/or the ability to ask the questions themselves.

You admit that you aren't trained, but your idol Dr. Ekman says that something like 0.2% of people have the ability to semi-accurately perceive microexpressions without training. I would wager that an even much smaller fraction can, untrained, translate those perceived microexpressions into a reasonably certain analysis.

Your decision to providing this analysis alone, regardless of the content or quality of the analysis, basically screams "I have poor judgement". Since apparently having a handful of undergraduate psychology classes is enough to allow you to nearly write libel, that's my psychological analysis.

Please for the love of god don't take what I'm saying here as tearing you up for the sake of being a jerk. I would absolutely love it if you could do an analysis like that but could back it up. I also greatly encourage you to continue your studies, but to be very self-critical and skeptical, because yours is a field that has humans as the lab rats, so mistakes are often costly.

I don't know much about the research, but I just searched our EBSCOHost on the psych literature and found this:

Telling lies: Clues to deceit in the marketplace, politics, and marriage.
Ekman, Paul, University of California, San Francisco, CA, US
Source:
    New York, NY, US: W W Norton & Co, 2009. 402 pp.
ISBN:
    978-0-393-33745-7 (Paperback)

Abstract:
    (from the cover) In this revised edition, Paul Ekman, a renowned expert in emotions research and nonverbal communication, adds a new chapter to present his latest research on his groundbreaking inquiry into lying and the methods for uncovering lies. Ekman has figured out the most important behavioral clues to deceit; he has developed a one-hour self-instructional program that trains people to observe and understand "micro expressions"; and he has done research that identifies the facial expressions that show whether someone is likely to become violent—a self-instructional program to train recognition of these dangerous signals has also been developed. Telling Lies describes how lies vary in form and how they can differ from other types of misinformation that can reveal untruths. It discusses how a person's body language, voice, and facial expressions can give away a lie but still fool professional lie hunters—even judges, police officers, drug enforcement agents, and Secret Service agents. (PsycINFO Database Record (c) 2010 APA, all rights reserved)


Note the abstract:
he has developed a one-hour self-instructional program that trains people to observe and understand "micro expressions";

P.S. one thing I do know, subliminal mesages are complete BS.

[bitsalame]

While your analysis is interesting, I think that you know better than to attempt to practice a method of lie detection that is both very inaccurate and very misleading if used by someone who does not have a great deal of experience. I'm not saying that the science is inaccurate, but that you should know that YOU don't know enough about this to practice it on strangers for the benefit of the public. You're hurting your reputation as a science major. You're also presenting your "findings" of truth or deception as truth or deception, when, in fact, each gesture, expression, or apparent emotion can point to a multitude of other conditions. If you're going to do this, please at least list them with probabilities along with the dominant/combined findings.

By findings, I don't mean interpretation of gestures, expression, microexpressions, or apparent emotions but only the states that they could point to. You should know better than to attempt to interpret the communicated message behind a perceived behavior. Dr. Ekman found that expressions might be consistent across cultures in the states that they point to, but the communicated nonverbal message is a crapshoot. Expression analysis, for example, is troublesome when used by trained soldiers interviewing potential terrorists (they have to phrase questions very specifically). I mention this because your subject is a French(?) guy speaking English that he obviously can't fully understand, living in Japan, and unable to hear his mumbling, badly-miked interviewer half the time, let alone understand him.

If you really want to do this, run everything you did by a psychology professor and have him post. If you can't get anyone to do this, well, there's a sign. Part of good science is realizing that you have inadequate skills to interpret results properly. Especially if you're essentially diagnosing humans. Especially when crime is involved. Especially if you have a reputation to earn or keep. Especially if you don't have credentials or someone with credentials to provide evidence as to your skill level. Especially when you are making judgements that affect someone's reputation. Especially when you're trying to do something that only a fraction of clinical psychology grad students would attempt to do in a public forum for the sake of, well, entertainment. Dude, this is the psychology equivalent of a pre-med major doing knee surgery. Yeah, he might get the fundamentals taken care of, but it's not good for anyone involved.

And you should note that this criticism has little to do with the science. I likely don't know the science as well as you, but I do know that you don't know it half as well as you think you do, or else you wouldn't be doing it here. Even a well-trained psychology professional with years of experience in lie detection would be very reluctant to write what you wrote about an internet video. Well, at least without lots of prior footage of the subjects and/or the ability to ask the questions themselves.

You admit that you aren't trained, but your idol Dr. Ekman says that something like 0.2% of people have the ability to semi-accurately perceive microexpressions without training. I would wager that an even much smaller fraction can, untrained, translate those perceived microexpressions into a reasonably certain analysis.

Your decision to providing this analysis alone, regardless of the content or quality of the analysis, basically screams "I have poor judgement". Since apparently having a handful of undergraduate psychology classes is enough to allow you to nearly write libel, that's my psychological analysis.

Please for the love of god don't take what I'm saying here as tearing you up for the sake of being a jerk. I would absolutely love it if you could do an analysis like that but could back it up. I also greatly encourage you to continue your studies, but to be very self-critical and skeptical, because yours is a field that has humans as the lab rats, so mistakes are often costly.

First of all, you don't know WHAT I analyzed, and don't know who I've been trained with.
You don't know neither my experience, nor my rate of successful profiling.

Also, I stated very clearly the level of confidence and if it was a unsupported hunch I clearly stated it.
I've been responsible enough to create a disclaimer before making the analysis where I think I've been clear it is not categorical affirmation of anything, you obviously ignored the whole disclaimer in the beginning of the post.

PS: For the record, my successful rate of detection of microexpression without previous training was 85%. After the training I detect 100% consistently. I am also gifted in detecting emotion based deceptions, which means that even before I was aware of the researches in non-verbal communications my hunches were usually correct when something was "off".
With the research and training, now I have an explanation to my accuracy.

[phoebusg]

Although I really like the show "Lie to me" and have read Ekman's books. I must stand firmly behind Horkabork's positions - for they are identical to my own. I'm a biopsychology recent graduate. I've discussed Ekman's techniques with a cognitive neuroscience PhD - and the consensus is - in short. There is not enough evidence to fully support Ekman's theory yet - enough to make it functional. That is, making the predictions actually accurate.

His work in inter-cultural expressions is definitely of interest though.