back to article Boffins build lie detector for crooked CEOs

Researchers have developed a method of rooting out fraudulent financial statements based on the statements CEOs and CFOs make during quarterly earnings calls. The system was developed by researchers at the Stanford Graduate School of Business by analyzing the conference-call transcripts of companies that went on to report a …

COMMENTS

This topic is closed for new posts.
  1. This post has been deleted by its author

  2. Youngone Silver badge
    FAIL

    Stanford Graduate School of Business

    Load of Not-Science. See the bit at the end. How accurate exactly? What does anyone learn at business school? How to lie, that's what.

  3. JaitcH
    Happy

    Very interesting technique, now PR departments we have to learn the patter

    Since most companies and government ministers have professional hired mouthpieces that either prepare statements or even taken on the presentation in person, they will have to learn the new speech techniques for otherwise their falsehoods will be too apparent.

    Possibly Plod could use these analytical tools, as well.

  4. Anonymous Coward
    FAIL

    Hmm,,,i call bollocks.

    "with the overall accuracy of 50% - 65%".

    So, about the same statistical probabilty then as predicting the outcome of a flipped coin.

    Not much better than guessing. Bet it wont stand up in court. Which is good.

    Nervous people will trigger it even though they are telling the truth.

    A non solution to a problem which has been going on since the year dot.

    People lie. Human nature.

    1. Anonymous Coward
      Thumb Down

      Barking up the wrong tree.

      "Nervous people will trigger it even though they are telling the truth."

      Nervous people tend to cock things up, make mistakes, revise their statements and so on. People who know they're telling a lie are more likely to stick to a prepared version of events, make fewer revisions, and so on. Thus nervous people are less likely to trigger the mechanism, release the poison gas, or whatever.

      The difficulty - for this kind of research - is that if the CEO / CFO don't know they're providing dodgy accounts (because their senior management haven't told them, for instance) - their manner of speech will be truthful, meanwhile (and unfortunately) the accounts won't be.

      Surely it would have been better to spend time (and money) creating software to analyse the accounts instead of analysing presentations about accounts?

  5. David 45

    CEO's?

    Sounds like it would be of more use on politicians.

    1. Anonymous Coward
      Unhappy

      Politicians?

      Actually no, its not necessary since you can count on the rate approaching 100%.

  6. Bucky 2
    Pint

    Kitchen Nightmares

    I expect this means that Gordon Ramsey's likelihood of being crooked approaches unity.

  7. K. Adams
    Go

    Great! Now we just need to use this method...

    ... on our politicians.

    Probably blow the stats curve though...

  8. Anonymous Coward
    Joke

    CEO lie detector

    surely it is very simple logic

    CEO talking = CEO lying *

    CEO quiet = trying to misdirect you

    * If there appears to be some truth/logic to what they are saying - that is just to camouflage the hidden lie

    Finally think how many third world graduate you could have for a CEO's pay - would produce much better running of a company

    1. BristolBachelor Gold badge

      Seriously though

      Why the Joke icon?

    2. Kay Burley ate my hamster
      Alert

      The truth

      CEO talking = CEO lying

      CEO quiet = Waiting to steal your ideas

      ---

      Also the science behind this story is common sense stuff. If they try to distract you with tangents and hesitate they are selling something they know is bad.

  9. Bluenose
    Thumb Up

    "Execs who use above-average amounts of expletives are also more likely to be crooked, they said"

    I can think of quite a few people and some on TV who must be wondering what their auditors will be looking for this Christmas

  10. Anonymous Coward
    WTF?

    Oh

    fuck

  11. JasonW
    Coat

    Can they...

    ... run it over Phorm's top brass?

    1. hplasm
      Thumb Up

      Rather...

      Can they run over Phorm's top brass.

  12. Adrian Esdaile

    Going forward on this...

    we can see how shareholder value and corporate governance are basically emboldened by this new paradigm of technologically-based solution-driven performance analysis software. Given the complexities and opportunities in today's basically dynamic global marketplace, one cannot afford to basically allow the shareholders to perceive that corporate planning and control may not basically be in accordance with the publicly-defined financial goals vis-a-vis those put forward basically as a component of investment advice in the form of IPO-oriented prospectii; hence from a position of marketplace assurances a firm policy of financial appeasement should basically appear to be the dominant market capitalization strategy. Basically going forward this synergistic stratagem consists of quantumized packets of financial data bursts routed according to corporate capital expenditure requirements, as determined by the market-based goals and notions of public expectation of corporate-control executive reimbursement initiatives. Utilization of the trickle-down financial model as a prime initiative in this process, combined with best-practice just-in-time process-oriented management decision-making pathways, working synergistically with market-leader green economy standards and ISO certification at all major corporate unit and inter-unit edge-scenarios will basically provide continuing flexibility at the margins of the currency-control sphere envelope.

    There, I didn't swear once, and I expect this paragraph to appear in an end-of-financial-year shareholder report in about 30 seconds.

    Can I have my house in Spain and Ferrari now plz? KTHNXBAI

    1. Fred Flintstone Gold badge

      What a sterling..

      ..load of bollocks. You must be a consultant.

      :)

      1. Vladimir Plouzhnikov

        Re: Going forward on this...

        Fail. You did not say "granularity" and "compartmentalisation" even once. You're a fake.

    2. Kay Burley ate my hamster
      Thumb Up

      I feel sick after reading the first few lines

      If I read the rest my 3 day weekend could be a 4, hmmmm...

    3. Jim Benn
      Unhappy

      Not quite CEO material.

      NICE try, but it is quantized (quantised to brits) and the 'Basically' should be removed from the start of that one sentence. Otherwise, full marks !! Assistant to CFO at least !.

  13. Fatman
    FAIL

    How to catch a lying CEO

    Hook him up to a lie detector, and if it indicates deception, administer a taser shot to the nads.

    Then tell that CEO, the next time he lies, it will be for twice the voltage and twice the time duration.

    Continued lying will result in a further doubling of voltage and duration.

    I bet you will get the truth after two or three shocks.

    1. Anonymous Coward
      Go

      Nah

      I say just tazer the crap out of his nads every time he opens his mouth. Problem solved.

    2. Kay Burley ate my hamster
      Thumb Up

      No no no

      Time duration is no match for multiple jolts, esp with varying time gaps.

      1. Anonymous Coward
        Thumb Up

        Please tell me

        We can do the same for politicians. A few (thousand) shots to the nads would make my day.

  14. jonathan keith

    @ JasonW

    You mean "run over Phorm's top brass", shurely*?

    (*cont. p94 for spelling pedants)

  15. Dr_Jim

    And now, the politicians

    This is very perceptive, and really useful conceptually. One implication is that the integrity of a response to a question can actually be measured by the semantic values in that response. Perhaps if this were added to voice stress analysis, we would have a relatively high accuracy of lie detection in real time.

    That would really impact the moral quality of those CEO briefings, but what if the same method is applied to political speeches? I think we all intuit some of the mis-statements pols make, but the structured CEO-tester approach would improve the quality of our legislators no end. Maybe the research team should apply for CIA venture money for that one???

    1. David Evans

      "This is very perceptive, and really useful conceptually"

      ...except it isn't is it? 50-65% success rate makes it basically a guess. I'd also suggest the patterns its trying to measure aren't culturally fixed; British bosses are way more likely to swear than Americans for instance. Load of psychobabble bullshit basically - and that's the truth.

  16. Anonymous Coward
    Anonymous Coward

    Where on earth

    Did they find the *honest* ones?

  17. John Tserkezis
    Jobs Horns

    Do I have to be the one?

    "You're fucking holding it wrong".

  18. Anonymous Coward
    Stop

    Wrong

    There is no such thing as a 'lie detector', merely equipment that can detect the phenomena that accompany lies. The 'lie detector' is a misnomer, generated by film studios, the press and other ill informed individual. I know this because I had to do a lot of research in the subject and it is a topic in my field.

    This is not mere pedantry. The previous government wanted to reduce all complexities to a technical solution, and then became mired down in expensive projects, many of which were founded on myths and misnomers. Their predecessors rejected the use of polygraphs, in the wake of the Geoffrey Prime spy scandal, following a long and balanced assessment by a highly respected scientist. He did the right thing.

    During the last government's tenure the polygraph was introduced to detect dissimulation/recidivism in paedophiles. There are many technical reasons why this is dangerous, the most prominent of which are the 'false positive' and the 'false negative', i.e., where the offender is thought not to have offend but has not, or is thought not to have offended but has. The latter, although occurring at a lower rate than the former, has obvious implications, as does the former.

    For some things great care is needed.

  19. hugo tyson
    Coat

    overall accuracy of 50%

    Saying it's the same as a coin-toss presumes that 50% of the input data are lying and 50% telling the truth. Isn't it more likely that with input data of 90% truth, 10% deceptive, they flag it as follows:

    85% : actually true, labelled true

    5% : actually true, mislabelled lying

    5% : actually lying, mislabelled true

    5% : actually lying, labelled lying.

    One might argue that's a 50% accurate positive detection rate, though its output overall is 90% accurate (only 10% mislabelled). Whereas a coin-toss would mislabel 45% of truthsayers as liars!

  20. The Other Steve
    Thumb Down

    Just Three Words

    Fuck right off

  21. Hollerith 1

    Assuming they have normal brains

    Many high level managers are to some degree sociopaths, that is, someone who has no empathy or understanding of anyone else, thinks they themselves are superior, and everyone else is there to be snookered. These people are the human equivalent of sharks. Some hide their natures better than others. The mark of a sociopath is charm, energy and arrogance. These can be appealing: Alpha Male and all of that, and they are often effective if what they want for themselves is achieved by the work they are supposed to do. But if their personal needs require them to lie, fabricate, and embezzle, then they will do so without a flicker of conscience, as that's what they don't have.

    So how do these tests catch these people?

  22. Jim Benn
    Alert

    Boffins win !

    These blokes snookered a grant from someone to do this study, thereby increasing Stanford's take for the year. They then wrote a lengthy report, saying nothing, but saying it in such a way that it is difficult to figure out they are saying nothing. I would bet the "heads" of this research team are destined for CEO positions in the future ! Or they will windup working for the PR department of some government agency.

  23. Aaron Em

    SVM applied to excessively vague problem...

    ...SVM produces effectively meaningless results. Film at eleven.

  24. Anonymous Coward
    Boffin

    expletives?

    "Execs who use above-average amounts of expletives are also more likely to be crooked, they said."

    What a crock of shit!

This topic is closed for new posts.