03 November 2023
Lies aren't OK just because they turn out to be true
Michael Sanders
The prescription in science is as clear as it is in the law – if you did the thing, you're guilty
Behavioural scientists and people interested in cryptocurrency do not typically have a lot in common, but both are having a pretty bad couple of years. 2022 saw a crash in the value of most cryptocurrencies, most prominently in the form of the total collapse of the crypto exchange FTX and its sister company Alameda Research, a high-frequency crypto trading firm. Sam Bankman-Fried, the founder of both firms, was convicted yesterday of using billions of dollars of FTX customers’ money to gamble, make political donations, and secure the support and sponsorship of sportspeople. He intends to appeal.
Over roughly the same period, the behavioural science world has been rocked by allegations against some of its most prominent figures, including two professors who are also authors of New York Times bestsellers, suggesting that they may have altered Excel spreadsheets of data from experiments to make them support their hypothesis, or even have fabricated parts of datasets for the same purpose. These two professors are accused of independently of having modified the data associated with two studies within the same paper. Both deny the claims made against them.
In the background, but no less interesting, there is a question about whether or not, in some deeper sense, any crime had been committed, if the effects of that crime subsequently disappear. If I lie about a tree falling in the forest, and that tree subsequently falls down, was I really lying?
In crypto, something like $8 billion dollars of FTX customer money was thought to be missing at the time of their bankruptcy last November. Since then, the firm handling the bankruptcy has said it has been able to get almost all of it back. They’ve done this by suing people who had the money and shouldn’t, uncovering pots that had simply been mislaid rather than lost, and by cashing out some investments Bankman-Fried had made which did well. Bankman-Fried may go to prison for the rest of his life (his sentencing will not be until March next year), but there’s a good chance that all or most of his customers will be made whole at the end of the day. The lie he is accused of telling – that depositors’ money was safe – might be true in the end.
In behavioural science, we can turn to the political scientist Michael LaCour, who published a paper in 2014 that purported to show that a short encounter with someone gay significantly increased people’s support for same-sex marriage (which was at that stage yet to be made legal by the Supreme Court), and that this effect was highly durable. The only problem was, as was uncovered thanks to some excellent data detective work, the entire study seems to have been made up, with LaCour allegedly manipulating an unrelated survey dataset to give the appearance of a study that never actually happened. (He denies these claims.) Later, the academics who looked into his data managed to actually run the experiment, and found that a conversation with a trans person did actually significantly increase people’s support for trans rights.
In both of these cases, the claims have turned out to be true – at least to some extent – and an argument can be made that perhaps that should be taken into account when, for example, a judge is deciding how long to send Bankman-Fried to prison for (the maximum sentence for his crimes is thought to be somewhere to the north of 100 years). Should it be?
I think not. If you invested money you shouldn’t have in a company you shouldn’t have invested it in, and that investment pays off so you can afford to pay everyone back the money you stole from them, it doesn’t alter the fact that you lied and stole. Intuition and knowledge play a role in investing, so it’s not quite the same as putting everything on black on a roulette wheel, but there is some probabilistic element, such that if you continue to behave in this way, at some point your luck will run out – which is usually how Ponzi schemes end up being found out.
Similarly, the findings that people choose to fake in behavioural science are not done at random – they believe that the finding is true, and often the people committing the most compelling frauds are smart, hardworking, and have great intuition. But this intuition isn’t infallible, and that’s the reason we have a scientific process. Unlike a generic motivational speaker, the case we’re making isn’t purely that we have the experience or the intuition to make great decisions. If you’re claiming to be a scientist, political, behavioural, or otherwise, then a huge part of the reason people believe you is because of your claims to science. If your intuition is right even 75% of the time, then one in every four papers where you commit fraud are going to turn out not to be true, allowing the field to be built on fraudulent foundations.
The prescription in science is therefore as clear as it is in the law. If you did the thing, you’re guilty – regardless of the outcome. Now that Sam Bankman-Fried has been convicted of fraud, he should go to prison even if nobody was made financially worse off by that fraud. If someone committed research fraud, they should be cast out of the scientific community and their work excised. If the things they faked end up being true, well, they just got lucky.
Michael Sanders is Professor of Public Policy and Director of the Experimental Government Team at the Policy Institute, King’s College London.