Andrew McAfee – Principal Research Scientist, MIT Sloan School of Management and Author of “Machine, Platform, Crowd”
When there is a huge surge in technology, something in management needs to change. We underestimate how much change is coming. Andrew wanted to talk about 3 pieces of business advice that are no longer relevant, at three levels: 1) Process, 2) Core vs Crowd and 3) Products vs Platforms.
The current paradigm is:
“Let machines handle the routine work; let people make the judgement calls”.
CEOs love talking about making decisions “from the gut” – HiPPOs or “highest paid person’s opinion” is common. The nemesis of the HiPPO is the “Geek” – who makes decisions based on the evidence, applies analysis, and doesn’t try to second guess it…
The evidence is mounting up that Geeks make better decisions vs HiPPOs. Grove et al (Clinical versus mechanical prediction: A Meta-analysis) found 136 studies of expert vs algorithmic prediction – 48% the HiPPOs and algorithms were equivalent. 46% of the time, the algorithm was clearly better. Only 6% of the time the experts were clearly better. And these studies were all using “small data”, not big data leveraging massive neural networks.
Conclusion – the HiPPO needs to become an endangered species in the big organisation.
Excellent judgement and machines
Andrew then talked about Google’s AlphaGo victory over Lee Se-dol in 2016 and the “story behind the story” that he found when he interviewed the Deepmind team. In Go, you need to memorise all the combinations of opening moves – and it is more complex than chess. Andrew explained that there are two early strategies to aim for the periphery or centre board – and if you don’t take them, you are obviously not very good.
If you look at move 37 of game 2, AlphaGo made an unusual and unexpected move. The result is that learning in Go has accelerated due to the algorithm – what other areas could this happen? We now have a colleague to accelerate our knowledge…
The Core vs The Crowd
We’re now learning how powerful the crowd is – the wisdom of the crowd, as opposed to the core organisations doing work in isolation.
Karim Lakhani is an expert on this topic and wrote a great article in the Harvard Business review about using the crowd as an innovation partner.
Andrew talked about sequencing the genome of human white cells (infection fighting cells of the immune system) – NIH got performance time of 10000 seconds (s) with 72% accuracy. Harvard sped it up and got 74% accuracy. Karim Lakhani ran a two week challenge to see what the crowd could do by posting it as a challenge on Topcode – the time came down to 10s and with increased accuracy of around 80%…
Interestingly, speaking to some of the individuals in the crowd, none of them identified as geneticists – but they were young, passionate experts in data analysis.
Crowdsourced Quantopian dived into algorithms and are hoping trading strategies coded by members will disrupt the industry – as opposed to having to employ and find good quants, you leverage their tech in a win-win situation.
Platforms – Smartphones vs Taxis vs Group Exercise
For the first year of Appstore, Jobs didn’t allow outside developers to publish apps – as he was concerned about malware. When they experimented with allowing it, as we all well know it had massive take-up. Andrew’s point is thatit is no longer a market of phone sellers and buyers – it is the sale of a platform to allow access to apps. Apple captures a huge amount of the profits just because of their Appstore platform.
Uber shares growing financials to distract from negative publicity – as opposed to the classic opposite behaviour. The lesson is that if you get the economics and technology right in a platform business, that will carry you a long way in terms of revenue…
Even group exercise is being changed by technology – Classpass is a platform with an intriguing proposition. They are a two sided market that offers their members lots of different classes in different places turning your city into a “gym smorgasbord”. Their value prop to class owners is that they will fill up classes at a discount – some revenue for a spot is better than no revenue…
The advantages of platforms are:
- network effects / demand side economies of scale
- control of user interface and user experience – you take an “Uber”, and don’t remember the make and model of the car…
- control of ecosystem
- crowd-based discovery
- Data and algorithms for
- trust – eg if someone had come to you 10 years ago and said I have a business idea where you can travel around the world and stay in a stranger’s guest bedroom… you would have said no way! Airbnb realised that lightweight trust mechanisms like linking social media profiles and ratings are powerful.
How do we prepare for our future generations and their future work
The world is getting better, more quickly, than ever before.
Andrew’s self-therapy is a website called Our World in Data – a website that allows you to explore the ongoing history of human civilization at the broadest level, through research and data visualization. The number who live in poverty is going down, infant mortality is going down, drinking water access is going up. The challenges about what we face are not about whether we are becoming wealthier in aggregate, but more about whether the division of that wealth is fair to everyone. And that everybody feels like they are being brought along with this increasing wealth and ability.
Platform economics determines successful business models…