Earlier this week, I gave a talk at the excellent State of the Net event in Milan about organisations in the age of algorithms. I have embedded the slides at the bottom of this post, but here is an outline of what I had to say.
This is an exciting time for organisational development, especially working at the intersection between new ways of working and emerging technologies.
Most of today’s organisations and institutions were designed and optimised for a past that is disappearing: companies lasted a hundred years; products lasted years, sometimes decades, and people had jobs for life once they emerged from the education system. What used to be a reliable system of management coordination to control distributed workers and their output is now incapable of the speed and complexity that modern organisations and markets require.
As the pace of change increases, the failure of our organisational models to respond to the challenges of the new connected world is putting companies and institutions in danger. Startups, although they tend to begin with leaner, more effective ways of working, are also not immune from falling into this trap as they grow, as they are often guided towards conventional management roles and org charts by their investors or advisers.
Assuming we still need organisations and institutions to carry forward long term goals and to coordinate and add value to the work of individuals, which I believe we do, then how might they change in the age of algorithms?
Current structures are neither sentient nor adaptable.
Current corporate structures for coordinating work, and the culture of management they are based on, despite the intelligence of the people who make up the structure, are not what you might call sentient. They do not adapt or learn, have little in the way of distributed intelligence, and in some respects are closer to systems of political control than organisational systems.
We have argued for some time that human, networked organisations support more effective ways of working, and the rise of social technologies inside organisations has helped us understand how this can be achieved even in very traditional firms. There are also plenty of examples of sentient structures to study, such as Morning Star, where colleagues negotiate peer to peer delivery agreements, and Valve, where employees are encouraged to flock towards work they think is important. Both are examples of how more human structures can create a more intelligent and adaptable system, but they remain exceptions rather than common practice.
At the core of Post*Shift’s mission is to find ways to allow all organisations to put these ideas into practice – not just the exceptions.
But now the game is changing in more dramatic ways.
Products are no longer fixed objects to be designed and mass produced over a comfortably long lifecycle. Data + services + experience are becoming more important than the underlying hardware in many product categories, and we are seeing the rise of products as platforms, which in some respects requires companies to also become platforms. And the internet is no longer just an external distribution channel, but a new connected culture that reaches inside organisations. An organisation’s people are connected whether you like it or not.
Connected products demand a connected company to develop them. Daimler announced this week that it is running an autonomous truck on the autobahn in Germany. Bosch and Mercedes already have test version of autonomous cars on the road in Europe, just as Google and others have in the USA, and as commentators have noted in relation to the launch of Tesla’s impressive Model X vehicle, some of its key features seem to be intended for autonomous use as a taxi, rather than just being there for the enjoyment of early adopters. Outside of connected vehicles, we are also seeing rapid progress in the connected home and even the connected factory under the rubric of so-called Industry 4.0.
Across many sectors and product and service categories, the frightening gap between the organisational capabilities needed to respond to these changes and the current way even some quite innovative firms are organised all adds up to a ‘burning platform’ moment. Change is no longer a whimsical idea but a clear and present need if traditional organisations are to endure in this new world. It is not just the emergence of small, specialised startups that threatens traditional firms; there is already a clear trend pointing to a reduction in average lifespan for large firms, which suggests they also have internal challenges to their survival. Current research predicts that the average lifespan of S&P 500 companies will fall to around 15 years in the next decade.
In many cases, the first instinct of organisations is to simply graft on new data science and software capabilities to old structures, perhaps by buying some startups, but without changing those structures or the culture of work that drives them. This carries huge risks of a disconnect within the firm. Volkswagen is a good example of what can happen when a command and control management culture (even in a company seen as having first-class operational capabilities) does not engage with the detail of how software is impacting in its products and its business. They will not be the last such example, because the generation of managers who hold power in these organisations is ill-equipped to understand these changes on their own.
Towards the post-human organisation?
This is the organisational context in which machine intelligence is starting to enter the workplace. Organisations of the future will probably combine human workers, algorithms and AI, and in some industries also physical robots. But this will require more sentient systems for the coordination of work, and also ways of orchestrating data flows, capture and analysis that go beyond what most companies can imagine today.
The extreme version of this vision for post-human organisations is the Distributed Autonomous Organisation (DAO), which is managed and run entirely by algorithms based on the assurance provided by the distributed ledger of the blockchain. This is interesting, of course, but I suspect will remain an edge case at best. The really important work will be in equipping mainstream organisations, firms and institutions to embrace non-human factors and combine them intelligently with their workplace systems. We still have a lot of work to do to bring even basic machine intelligence into the workplace of most existing companies.
There are, of course, many good reasons to worry about automation, and no doubt that many jobs will disappear; but as with all such technological shifts, it will also produce new jobs and make many existing jobs more interesting and productive.
In manufacturing, we have lived with robot workers for many years. Interestingly, we are now seeing the return of humans to the factory floor working alongside small, nimble robots. Toyota, a pioneer of lean working in the automotive sector, has recently begun doing this to take advantage of the craft and mastery of highly skilled workers:
“We cannot simply depend on the machines that only repeat the same task over and over again,” Kawai said. “To be the master of the machine, you have to have the knowledge and the skills to teach the machine.”
This points to what I think is the most exciting potential for algorithms in the workplace: not replacing humans with artificial intelligence, but enhancing their work by pursuing the goal of augmenting human intelligence. As Andrew McAfee wrote in his book about the second machine age: “the best chess player on the planet today is not a computer. Nor is it a human. The best chess player is a team of humans using computers.”
Just as we are getting used to asking Siri for help with simple tasks, so too will we begin to see technologies like Siri and IBM’s Watson quietly assisting us to get things done. We already have examples of Watson’s AI capabilities being made available via text interfaces, and it is likely we will see many more ways in which AI can help us in our day-to-day work, such as intelligent agents that listen in to our email and activity streams or help us find legal opinions or relevant documentation to the work we are doing. Bots will probably be more polite and more helpful than those who worry about a robot apocalypse realise. This is part of what IBM refer to as Cognitive Computing.
But what does this mean for organisations and how we design them?
First, although there are legitimate fears for many low-end jobs at risk from automation, in fact, there will also be a number of management tasks that can also be automated, such as reporting and sharing information between teams. It might be the case that the middle managers who populate bureaucratic structures are most at risk, and it is hard to say if this would be a good thing or a bad thing for the performance of organisations. But we need to be careful with aspects of the workplace that probably should not be automated, such as mediating social trust between people, HR performance management and the growing need for high-touch service experiences, among others. These are areas where the human touch and human insight are still needed.
Second, it means we need to combine streams of machine and human outputs within the workplace. We have activity streams of human chatter in tools like Slack, Yammer or WhatsApp, and we can connect our machines and algorithms into these social streams to increase awareness of what is going on. Partly this means sharing information like their status or operating condition, but it probably also means intelligent agents listening for events they can help with, or patterns they can help us understand.
Third, it means we need to create platforms to surface and socialise data so that everyone in the organisation can help interpret it and therefore use it more effectively. This is something that requires a major culture change in organisations from ‘Need to Know’ to openness towards operating data.
In short: we need a new operating system for the modern organisation.
In some ways, the changes we have seen in how software works over the past decade or so provides a helpful analogy for what needs to change in an organisation. Software used to be very rigid. Database and proprietary code vertically integrated, and programmes were optimised for a single purpose or use case. Code was hard to manage and to change, and difficult to integrate with other tools.
But now software is a layer cake of platform, services, APIs, data and experiences, and each layer can integrate with other tools and other code. We tend to build the services first and then the apps and interfaces that can use them rather than try to vertically integrate everything.
In the old days, corporate IT departments would mandate everyone use the same tools from a single vendor as interoperability was hard, but they are now starting to become comfortable supporting different apps as long as they follow standards for data and security etc., and can work together. The existence of modern service-oriented platforms and interfaces has enabled a new and better way of working, much like the way iOS and Android platforms on our phones have taught us how a tightly controlled, a modern platform can support a wide array of different apps.
The organisation as a platform for people and code.
The primitive vertical integration of the top-down org chart needs to change in much the same way. We need a modern organisational operating system that has a core platform of shared capabilities, with a service-oriented architecture that allows individuals and small teams – perhaps also partners and customers – to plug in and work in the way they want to whilst remaining coordinated and contributing to shared value creation. This is what Dave Gray wrote about in his book The Connected Company – a shared service platform that manages standard, protocols and security to enable more free-form small teams to operate autonomously on top.
Organisational systems need to be evolutionary. Alive. Sentient. To evolve, they need data about how they are doing and the ability to rapidly iterate and try out new ideas and techniques governed by tight feedback loops. A key part of these platforms will relate to how we handle data. Not just mining stored data, as some advanced companies are already doing, but sharing socialising and interpreting real-time data flows from both human and machine workers to create a self-aware organisation.
We need self-aware, adaptive, learning organisations.
The attributes of a modern organisation cannot be created with more top-down management or more processes. They need to be baked into the Operating System (OS) in order to create adaptive organisations.
We need more:
- Autonomy and self-management
- Everything as a Service / interfaces
- Self awareness / sensing
- Connected and data-driven operations
- Agile and adaptive, resilient structures
- Intelligence at the edges
- Social trust and social authority
- ‘Sense and respond’ rather than ‘plan and manage’
- Autonomous small teams that can move quickly
We should not accept the organisational structures bequeathed to us from the past as fixed. We need to bring them to life and expose them to the evolutionary forces that have shaped the internet. The power of competition in the open to create evolutionary improvement is obvious when it comes to free markets, and the Cambrian explosion of startups that has brought so much innovation in recent years. An open, intelligent organisation where data flows as needed, based on a common platform that supports small teams and agile working could bring the same breath of fresh air into the organisation, unleashing the power of evolutionary improvement and adaptation.
Where to begin for an existing organisation?
Our approach, as we have outlined previously, starts by defining what the organisational system is for:
- How is it supposed to add value to the work of the humans and machines that make it up?
- What capabilities does it need to fulfil its evolutionary purpose and how can we measure and test for them?
- How can it build out the features of a shared platform to support them?
Having defined organisational health measures and created tests for the key strategic capabilities the organisation needs to improve, this is where the power of real-time data can be brought to bear on organisational transformation. But we also need to mobilise what Dave Snowden has called the human sensor network, to let people at all levels of the organisation provide feedback about where they think these capabilities are improving and what can be done to accelerate the process.
The point is not that we can predict the future so well that we can design the perfect organisation to match a future scenario. The point is that we need organisations that are self-aware, adaptive and capable of constant transformation in order to cope with the increasing pace of change. At the core of this is the idea of a company as a platform for people, code and systems to create value for customers.
The Quantified Organisation and agile transformation.
Instead of just using big data tools to increasingly monitor and manage individual performance, as the HR Technology world is starting to do, we can also use real-time data to monitor how the organisational system itself is either adding value to the work of its employees or getting in their way.
- Define key capabilities of the organisation
- Identify measures in the data that test for these capabilities
- Ask all teams to try small changes and tweaks that can move these capabilities in the right direction
- Allow employees to share their view on what is working and what is not
- Use this data and human experience to drive small transformation activities at all levels
In contrast to big-bang top-down change programmes, or perhaps as a supplement to increase their impact, this approach to agile transformation can give everybody a role and a sense of ownership of the organisation and its development as it gradually build out the operating system. It means we can be open with them about the current state of the organisation and how it is changing, creating a real-time map of the organisation and its key capabilities along the way.
Connectivity is our new superpower.
Humans have an amazing ability to achieve the impossible when joined together in networks of common purpose. When we give them intelligent tech to support their work and handle some of the basic tasks, they can do even more.
If we can surround people and processes with an organisational operating system that can adapt and respond to change and help coordinate outputs, then we I think we can look forward to a new era of innovation and productivity inside the enterprise. Yes, there will be legitimate fears about the human impact of automation, especially on jobs; but there is also a very positive story to be told about how this can make organisations more fluid, agile and sentient, which mean they would also become better places to work.
There is some great work going on right now to explore the contours of this challenge, but let’s not just leave it to the early adopting experimenters. Any organisation can begin to create an evolutionary improvement cycle using agile transformation methods and smart approaches to digital transformation to develop the capabilities they need to succeed in the age of algorithms.