Jordan Peterson en het Hitler-citaat

In Machine, Platform, Crowd, authors McAfee and Brynjolfsson describe three major developments that led to the enormous economic change we have seen over the past decades. The rapid developments in technology (machine) led to possibilities of the forming of powerful new layers that bring consumers and producers closer together (platforms), and how these platform thrive through direct involvement of the consumer in the production and dissemination of the product and services provided through the platforms.

How can companies like Uber, Facebook, Amazon have become so big and influential, considering they are only thin layers? These platforms do not produce goods, and have no or little assets (at least at the outset).

In the book many aspects around these developments are brought together. The authors contrast the old world and the new world: machines versus human intelligence, platform versus product, crowd versus core (core meaning something driven by an organisational structure).

Machines: Why Computers Make Better Decisions

Machines have developed that can crunch the new large volumes of data that the Internet era has enabled. Here we see that technological developments create their own new opportunities. The authors go into why these things are so hard to predict, and have no good answer. New technology enables things we can not foresee. We can dream, but technology continues to surprise us.

The developments of AI have been an important factor. But why computers are better than humans at making (some) decisions. The book draws on the work of Kahneman and others. Kahneman has learned us that our decision making is highly subjective and prone to errors. Fast decision making is done by our System 1 thinking, which is impulsive and subjective. Our System 2 is more thoughtful and slow, but tends not to correct System 1 decisions but rather justify those decisions. Our biases make us poor decision-makers. And computers can ignore all the subjective crap that clutters our decision making. And of course they can very fast go through last piles of data.

Though McAfee also shows that if the AI is fed with “biased” data, the computer will also make biased decisions. But, the computer can be easily corrected, while for humans that is a lot more difficult.

In the end, the computer is better at doing specific things. (The worst are Hippo based decisions: Highest Paid Person’s Personal View. A problem common in organisations with narcissistic leaders.) AI is increasingly efficient at making decisions for “narrow” problems.Scientists however indicate that Artificial General Intelligence (AGI) – is a stage we now even getting close to.

The authors do not go into the hypes that are created around AGI. People like Harari in Homo Deus offer extensive and interesting perspectives on what the world may become when AI takes over. But these are, I believe, not based on realistic views on the state on AI, or even on what AI might brings us in the future.McAfee and Brynjolfsson do not elaborate on this humbling perspective. They even ignore it later on, where the describe their believe that when given enough data, engineering knowledge, and requirements, computers will be able figure out novel ways to do things. This statement remains unsubstantiated and even contradicts their earlier statements about AGI from an MIT scientist.It is also contradictory to the Polanyi paradox: we do not know what we know. So that engineering knowledge may very well remain buried in human brain mass.Finally, to end this tangent, the claim itself seems somewhat circular. If I rephrase the statement: if we know what to do, how to do it, and have the right inputs, we can program a computer to do it. Well, of course, I would say, because that is as much as the definition of automation.

So how come we see this rise of AI technology now? McAfee and Brynjolfsson summarize:

• The availability of computing power. The power of CPUs and specially GPUs has reached a level that enabled and boosted the usability of neural network performance.
• The drastically decreased cost of computing.
• The availability of large amounts of data.

When will robots be used and when humans? Robots for Dull, Dirty, Dangerous work (DDD) and/or where Dear/Expensive resources are used.But coordination, teamwork, problem solving and very fine hand/foot/senses work is needed. These are all things computer and robot are not good at.Creative and social jobs are safe from robotisation.

Platforms: The New Economic Layers

Platforms have appeared that killed or diminished existing often large industries. Where products become digital, the fact that these are free (zero cost to copy) and perfect (no loss off quality when copying), economies have radically changed.Two ways are left to make money with these products:

• Unbundle products – like iTunes sells songs instead of albums.
• Rebundle products – like Spotify creates subscriptions instead of selling albums/songs.

Complements increase the sales of goods. Like apps increase the sales of iPhones. Free products can be bundled to make money out of them:

• Freemium products
• Put ads in free products
• Add customer service (open source products)
• Provide a public service (for public organisations)
• Pairing with products

For platforms, curation of products and reputation systems become crucial to filter and make products find-able to clients. Characteristics of successful platforms:

• Early – attract a crowd before others do
• Use economy of complementary products
• Open up the platforms
• Guarantee experience through curation/reputation like mechanisms.

Online-to-Offline platforms have emerged. These bring together products and consumers for a market that optimises asset utilisation. In a 2-sided market, demand is for low prices from multiple suppliers, and suppliers want their products in as many consumers as possible. Both sides wish to achieve economies of scale. If it is a product in an undifferentiated market, prices will come down. Such products are vulnerable to platform destruction. Which is less vulnerable: complex services or markets with few participants?

Crowd: Why the Many Beat the Few

How to make successfully use of crowd-sourced information?

• Make information findable and organise it
• Curate bad content

Crowd sourced platforms can only be successful when

• They are open
• Everyone can contribute (no credentials needed)
• Contributions can be verified and reversed (prevent destruction of the asset)
• They are self organising (distributed trust)
• They have a geeky leadership

The volume of the crowd knows more than a few experts. Crowd beats core.The core nowadays uses the crowd:

• To get things done (upwork)
• For finding a resource
• For market research
• To acquire new customers
• For acquiring innovation

The Future of Firms

Distrust in organisations leads to a wish for Decentralization of Everything. But “The Nature of the Firm” describes why organisations exist and why their is always a place for them.

The cost of linking parts of the supply chain in more expensive when it needs to be done with different players all the time. In an organisation that handles a large part of the supply chain, cheap communication drives down costs. More importantly, contracts are never complete.

There is always a thing called Residual Rights of Control over assets. The concept is not further elaborated. But in a distributed model the ownership of the produced assets poses problem: who owns the right over the assets.The problem seems incomplete and drives construction of firms.

Firms drive group work and management:

• To coordinate more complex work: transmission belts for coordination and organisational problem solving
• Human/social skills
• People want to work together
• Best way to get things done

They end with the question: what will we do with all that technology – that is the question we should answer, not: what will technology do with us.

Apply technology to solve real-word problems – in a combination of technology, humans, and other resources.