Talking To Strangers, Being “Disagreeable” and Thinking Better

In Talking to Strangers, Malcolm Gladwell (who btw is a proud U of Toronto alumni) explores how we make mistakes all the time, some of them jarring or risky ones, about the people we meet face to face – regardless what we knew about them beforehand. Face-to-face interaction is supposed to critically help but instead too often critically hinders us – that means job interviews, first dates, business or diplomatic meetings… but then the book also devotes quite a few pages to suicide and sexual predators and I couldn’t read some of it without flinching, so BIG CAVEAT.  

In 2017, economist Sendhil Mullainathan went through 554,689 New York City defendant arraignment hearings between 2008 and 2013. Records indicated that human judges, evaluating the unique circumstances of each defendant, had released on bail some 400,000 people who appeared to have committed criminal offences and were pending trial. (Roughly, the judge looked at the defendant and formed an opinion – bolstered by professional experience and an ability to size up another human being – as to whether the persons released on bail posed an immediate threat to society.)

Mullainathan then built an AI system, fed it the same data, and had the computer pick 400,000 people from the same list to release. Obviously, the computer didn’t look those people in the eye, or look at their mums sitting in the back of the room (something human judges have said makes things particularly difficult when dealing with say, a 16 year old with a history of violence.)

(Lemme add my own two cents – long ago as a relative market freshie cutting my teeth on default probability software, toggling between the “private” model – where a company is evaluated for risk of demise (probability of default, insolvency) solely based on its reported financial statements – and the “public” one – where market-traded share price is taken into account – the difference in results can be freaking trippy. The only additional data inputted had been a publicly traded share price. In other words, market sentiment, human opinion – as influenced by news reports, rumours, mood swings, advertisements.)

So back to Mullainathan’s study – guess who won (ie put fewer people back on the streets pending trial, who then went on to repeat their crimes), AI Judge or Human Judges?

The difference in correct predictions re who would not repeat the offence while on bail was 25%. AI Judge released 25% fewer people who went right back to crime pending their trials. That’s a difference of 138,672 people over 5 years. 76 people released every day on bail in NYC who would go right back to their criminal offences. 

That’s not all. Human Judges missed 48.5% of the High Risk Group flagged by AI Judge. According to psychologist Tim Levine, human judges on average correctly identify liars 54% of the time. That’s only slightly better than pure chance. 

You know what that means, right? No, not NYC Issa Crime City. Not even Human Judges, Robots Are Coming For Your Jobs. It means the same thing as Whether AI Can Replace Human Chess Champions, whether Deep Blue puts us in Deep Sh*t: We need to go ask Sendhil Mullainathan what exactly he told AI Judge to do. Behind AI Judge was still a human economist. Presumably human economist consulted human judges before building his AI (y’know, since he is not himself trained in Matters Judiciary.) In other words, you still need humans to build and work with AI. You just don’t need humans to be the AI. Winning chess teams were not ones exclusively manned by computers or humans, they were the teams of humans with AI. More humans need to learn more ways of using computers. (No surprises that among the string of illustrious awards he has received is one from Infosys in 2018 for elevating the prestige of scientific research in India and getting more young people to choose the field.)

“We tend to think the problem is solved when we solve the technology problem… but the human problem still remains… This isn’t about the biology of people, it’s about the psychology of people.”

– Sendhil Mullainathan in Solving Social Problems With A Nudge

Here’s an old TED Talk by Mullainathan, that I still like because he talks about how we come up with many solutions for problems, yet across the board is the Last Mile Problem “because our brain is really strange” and as a consequence “people are really weird”)

While the first 999 miles are about science, there’s this “final mile” that involves people intuitively making the wrong decisions when executing the existing technology. Mullainathan highlights that the next step is Psychology, Marketing, Art and Scientific Method (ie Operational Capacity) in getting people to change their habits. He found for eg, that sending letters telling people say, how much energy their households were saving compared with their neighbours’ households (better OR worse than them) provided a boost in energy saving motivations. The letter. Not Yet Another New Technology. It points to (relatively far less glamorous) need for good execution of existing ideas, not just cutting-edge innovation. AND the money argument: The science costs a bomb. Massive research funding. The last mile costs… postage stamps.

Here’s another (true) story:

Practical, plainspoken former businessman-turned-widely-regarded masterly politician then-British Prime Minister Neville Chamberlain once described another man he met in politics thus:

“…you would never notice him in a crowd.. and take him for the house painter he was…”

“…in spite of the.. ruthlessness.. I saw in his face, I got the impression here was a man who could be relied upon when he had given his word..”

“..(his manner) appeared to show the storm signals were up… (but) he gave me the double handshake he reserves for specially friendly demonstrations…”

Aristocrat and worldly-seasoned top student from Eton and Oxford Lord Halifax who became Chamberlain’s Foreign Secretary watched the same individual rail about for 5 hours at their first meeting after mistaking him for a foot servant, then spent 5 days with him, then concluded he was “open to negotiating peace.” 

Dapper British diplomat Nevile Henderson met this man repeatedly as well, attending his rallies, concluding he “may have crossed the borderline into insanity… but he hates war as much as anyone.”

The person described above was Adolf Hitler.

The war all 3 experienced politicians didn’t see coming following their meetings with Hitler was the Second World War. 

Chamberlain’s successor Sir Winston Churchill on the other hand held the longstanding notion that Hitler was a “duplicitous thug.” Unlike Chamberlain, Lord Halifax and Henderson, Sir Churchill never actually met Hitler. 

In his book, Malcolm Gladwell observes that all 3 British politicians who encountered Hitler had had peace and productive discussion in mind (as Viceroy of India in the pre-war years, Lord Halifax had indeed negotiated brilliantly with Mahatma Gandhi.) Adolf Hitler however, had entered the same interactions from the very start with the intention to deceive. Against an “opponent” of that sort, you are better off not meeting them face-to-face. Faced with someone who never intended good, it makes human you far more susceptible to being fooled, than an AI.

When he looked him in the eye, Neville Chamberlain, arguably a man of honour, wanted to believe Hitler played by the same rules. And so when Hitler signed an agreement not to go to war, Chamberlain took him at his word. For someone who never intended to keep it, giving their word is the easiest thing to do. It took just 6 months for Hitler to renege on the agreement signed between him and Chamberlain. 

“If real life were like Friends (the sitcom), human judges would outperform AI…”

As part of their craft, Jennifer Aniston, Matthew Perry and the rest of the stars who flesh out the main characters clearly display on their faces the emotions being experienced (and therefore the thoughts that go with), as scripted by the writers of the show. Friends is transparent. In real life however, people with the intention to deceive portray on their faces and in their mannerisms what they want  judges, diplomats, investors etc to see, and there is a reason predatory people succeed: They’re good at deception. 

This also happens unintentionally:

“The phenomena that is at the heart of all modern controversies is people of different perspectives meet each other for the first time, and proceed to profoundly misinterpret each others’ intentions.” – Malcolm Gladwell. 

The Trobriands is a cluster of islands in the middle of the Solomon Sea, home to an isolated community of 40,000. Psychologist Carlos Crivelli and anthropologist Sergio Jarillo showed 6 photos of different facial expressions to urban primary school children, then showed the same selection to the Trobrianders. The kids interpreted all the facial expressions in line with how most of us urban adults would have, but to Trobrianders say, an “angry” face was thought to mean happiness 20% of the time, while 30% thought it signified fear. It’s not just one group of isolated islanders, the same huge discrepancies in interpretation of facial expressions happened with a group of fishermen known as the Mwani, and classicist Mary Beard explores this in Laughter in Ancient Rome.

If Friends was shown to Trobrianders, Mwani or Romans, they would have taken away something very different from the show. I’m just sayin’ – imagine what happens when say, Trobrianders, Mwani or Romans get into a staring match with… Angry Drunk Urban Person Inna Bar. Gladwell in Talking to Strangers addresses the terrible effects of too much alcohol on misinterpreted intentions in bars, it gets graphic, so I’m only going to say it’s amazing how much alcohol those promising students in top Ivies consumed, and then proceeded to place themselves in situations they could terribly regret. Like, make like Odysseus and tie yourself to a mast if you want to go near the sirens. Have a think beforehand (long. beforehand.) what your boundary is and never, never, never move that once you’ve started drinking.

And then I’ll tell you instead about the time I had a former boss from over a decade ago who was fascinated by Chinatowns and more… “obscure” bars in Asian towns. He told of one hapless person he read about who got into a staring/glaring match with someone who, unbeknownst to him, was a triad member. As he was leaving the bar late in the night he had his hand chopped off. The part that really got my former boss was when a local paper called the perpetrators “unprofessional” because they didn’t leave the severed limb behind. As in, apparently the “appropriate” or “proportional” response is to leave the hand after you chop it off so it can be re-attached. NO I HAVE NO FURTHER KNOWLEDGE ABOUT THIS beyond that my former boss (who is from Brisbane) was amazed the local news article had tut-tutted the hand’s disappearance. Not that it was chopped off to begin with.

Here’s a more easily-Google-able one about interactions that went terribly wrong. Amanda Knox is not a murderer, “just a little bit weird.” But “just a little bit weird” led to a murder conviction and 8 years in an Italian prison for the murder of her room mate, because her “strange” behaviours were interpreted by police officers as signs of guilt or lack of remorse, despite lack of any other evidence linking her to the murder AND in spite of substantial evidence including crime scene DNA incriminating another person. If you read some of their testimonial, just as in the case of the British Politicians Who Wanted To Believe No One Could Ever Choose A Terrible War, the people involved in the Amanda Knox Case Saw What They Wanted To See.

It’s where we need the Sendhil Mullainathans of the world. (Scroll below for a more recent talk of his, titled Think Better.)

Pull the Goalie!

Human personality can be assessed along 5 dimensions, of which disagreeable-ness is one of them. How willing are you to bend over backwards or in some other way tank your own cause, increase your own “chances of losing” something, just to not appear “disagreeable,” ie “silly” or “unconventional”? On a scale of 0 to 10, where 0 is a Golden Retriever and 10 is Mr Spock, how “disagreeable” are you?

Clifford Asness and Aaron Brown, friends of Gladwell’s whom he describes as philosophers with “Disagreeability Factors” of 7 and 8.5, “test their offbeat theories with mathematical precision,” because it’s fun to espouse the absurd with a grain of truth. They found some jarring differences between what made sense mathematically for a coach to do, and what was socially acceptable (Math Is Fun!) and so they published a paper titled Pulling the Goalie: Hockey & Investment Implications.

In the last minute, minute-and-a-half of a hockey match, a coach of a losing team “typically” pulls the goalie and subs an extra player on Offense. The idea is that you increase your risk of losing the game (with a wider margin) in exchange for a chance to even the score, with the extra attacking player. Asness and Brown (who btw are also successful Wall Street money managers) however proved that in order to have a real chance at evening the score, if you are 1 goal down you should be pulling your goalie….. 5 minutes and 40 seconds before the final whistle blows. (And if you’re 2 goals down, you should sub your goalie with a whopping 11 minutes and 40 seconds left in the game). Wait for it –

NO ONE ACTUALLY DOES THAT. It’s too “disagreeable” a move for a coach. Say you’re the coach. If the gamble doesn’t pay off, everyone is going to hate you for making the call. Fans would be mad at you for making the game more lopsided (read: “boring” because of the wider goal margin -it doesn’t provide the optimal amount of entertainment for people watching the game), and players whose respect you need to command would resent your call because it’s easier to lose a close game than a “landslide” one. This greatly increases the “cost” of making that call.   

Now, Aaron Brown is obviously a quant, a maths guy. But here’s more about him – he grew up with a mum who had a Masters in Chemistry and a dad who had a PhD in Physics, and there was this time Dad Brown inherited a “strange contraption” from his Airforce employers. “Something” a brilliant-but-troubled predecessor had received a little bit of funding to build… Predecessor kills himself and no one knows what the many-tubed many-geared thing does, so Dad Brown volunteers to take it up… then puts it in his basement and tells his teenaged son (that’s Aaron Brown), “Hey, take a look if you want. Like its a box of Legos.“

Brown describes the time, “I stared at (mysterious machine) for hours and hours… it did something when you turned it on, but (his dad’s predecessor’s) notes were indecipherable and he’d burned some of them…”

They never figured out what the machine really did, “But this is how you raise a child with a Disagreeability Factor of 8.5.” Um, ok. For the rest of us plebes with no access to Mysterious Airforce Machines can we adopt a rabbit with a Disagreeability Factor of 8.5.

A-nyway. During the Seattle recession when he was growing up, Aaron Brown decided he needed a “marketable skill,” and taught himself poker. He goes on to describe how he discovers a “big money game in the basement of a local tavern” and that it’s not that difficult to be good enough at the games that he can win consistently. He says the “adults” let him win some money and walk back out, “but I was a shy kid, it took every ounce of courage I had to walk in there…. There are people who could have done that and it would not be a big deal, but for me it was the most traumatic event of my life.. I was 14.”

Know what I’m thinking? Seattle Must Be A Really Nice Buncha People! 😀 They’re nice to other people’s kids 🙂 Young Aaron Brown went on to Applied Mathematics at Harvard and an MBA from U of Chicago, becoming an established author on risk management and gambling-related issues, his website receiving several Forbes awards for Theory and Best Practice in Investing.

While the college students and young professionals in Malcolm Gladwell’s Talking to Strangers stories were getting sh*t-faced and putting themselves in questionable situations, 14 year old Aaron Brown began his long and profitable career in probability games…. playing poker. In bars. 

Epilogue: 

 

Sendhil Mullainathan in Think Better describes replacing students’ 3rd period class with an experiment: Each pair of students is given a ball. One student is to take the ball from the other student. Lots of grappling ensues, and after 10 minutes the student trying to get the ball is asked, “Did you ever ask for the ball?” To which they reply “The other student would never have given it to me.”

And so they ask the other student, “Would you have given them the ball if they simply asked?” To which the reply, “Of course, it’s just a f-ing ball.”

Think Better…