Hacking Platform Business Models
I’m currently teaching a course on platform economy at the Aalto University. At the same time, the think tank I work for, Demos Helsinki, has been doing groundbreaking work on the redefining progress. This has lead me to think about the wider implications of platforms.
In this text, I explain what makes platforms click and operate in such a elegance and power that they are becoming the most important context of organising and collaborating in our society. I use business models as a tool to dissect and understand platforms, and systems thinking to grasp different approaches one can take to change platform behaviours.
So, platforms are more than just a set of business models: they might completely redefine our societies and our conceptualisations of time, participation and progress.
Nevertheless, looking at platforms from the perspective of the business models is useful, in part because they focus on value creation instead of value capture, and in part because business models offer a useful theoretical framework in modelling, understanding and conceptualising platforms and choices that are made on platforms.
Most importantly, if they are considered as a way of organising and collaborating, platforms are not very well explained by the current tools of the social science.
Let’s get organised
When Alexis de Tocqueville travelled around northern America in mid 19th century he realised that he lacked tools to investigate and understand the apparent dynamism in the American civil society, which he understood to be a cornerstone of the booming American democracy. His quest for better understanding the processes he saw led eventually to a new field of research, sociology.
Tocqueville was an optimist, who believed that democracy should balance liberty and equality. However, during the next hundred years it became apparent that such freedom would be illusionary. When the individuals gained more and more power, new mechanisms emerged to keep the elite in charge. Finally, Lippman’s media research lead to propaganda — a perfect tool to keep masses under control while also letting them to have an illusion of control. Although it was often used to create direct policy interventions as well, propaganda was not a strictly political tool: it really found its epicenter in marketing. While economy originated from social interactions such as the gift as described by Marcel Mauss, the media control combined with modern consumer economics separeted the man from the community, turning communities from collaboration to reasons to consume. Today, everyone needs to have a yoyo, tomorrow something else.
This was where we were still very recently. However, it’s possible to identify three major shifts in the current status quo. First, political campaigns by Obama and Macron have revealed new partly untapped modes of democratic participation. Second, Trump’s and Putin’s fake news have annihilated the last remains of Habermasian legiticimacy from the media. And thirds, financial crises in the macro level and platform economy in the micro level have shattered the basic assumptions of industrial era economics.
Currently, social sciences are not up to the task to explain these changes, because they all derive not from the world of towns, nation states or groups, but from the behavior of multisided platforms. Platforms have at least four characteristics to which current explanations and tools in social sciences are inadequate.
First, social sciences cannot explain how the experience of time is formulated and shaped in communities. Time, especially delays, is a crucial characteristic in platform design. With time, I refer to the question of how the time is experienced in different moments and how that experience, shared or not, shapes the human interaction and participation. Cognitive science and psychology approach this question from intrapersonal angle, and social psychology and behavior economics by identifying biases. Sociology and anthropology only provides very vague conceptualisations of cyclical and linear perception of time in different cultures, which is not helpful.
Second, social sciences have not lived up to the challenge to explain the postmodern perception of fragmentation. When Baudrillard describes the American condition in his book “America”, he is precise in his descriptions of perception of freedom and lack of history but completely unable to explain the consequences of his perceptions. He has no tools or vocabulary. Clearly, the theory is broken, and when the theory is broken, the postmodern trolls can’t get serious about it. The perceived fragmentation is not caused by postmodernism as such, nor is the politics broken directly because of populist uprising. Both are rather explained by lack of tools, which leads to perceived randomness — not chaos. Right now, people are not voting wrong, but they might vote random. Randomness disappears when we investigate these social phenomena from reasonable, yet undiscovered perspectives.
Thirds, social sciences lack proper explanation to why individuals and communities cannot observe the nature of exponential growth, a typical mode of growth in platforms but not in previous context of human collaboration.
And fourth: social sciences don’t have anything to say to the altering relationship of markets and companies (especially in relation to the platform economy) although these are few of the most powerful actors in the society. Nobody is even pointing out that there’s a need to get rid of the limited liability corporation — an obvious point if one takes into account the changes in risk cumulation, superabundance of capital and planetary boundaries.
Clearly, as pointed out to me by Juha Leppänen, there’s again a need for a new social science.
There’s a need for a new social science
To understand the key elements of this new social science, let’s investigate its primary focus of research: the platform.
Platforms are the new mode for organizing. They born from the trinity of
- algorithms and machine learning,
- abundant sensors and data, and
- novel economic approach of value creation by combining and self-regulating multiple markets.
Platforms demonstrate a future which is both utopian and dystopian. The current media control in democracy is shifting from content to shape, directing media consumption with filters and constructing realities by efficient algorithms that lock groups in their respective filter bubbles. Instead of amusing ourselves to death alone, a platform man does it while being constantly connected: he sends ironic images of himself bowling alone to Snapchat.
Instead of amusing ourselves to death alone, a platform man does it while being constantly connected: he sends ironic images of himself bowling alone to Snapchat.
Work becomes simple. Platforms dissect service work to Fordist levels by creating instructions, service standards and algorithms that do not allow for improvisation. The level of automation of this algorithmic work is uncertain, but the part left to man will mostly be dull, repetive and uninspiring.
Naturally, both of this sides of life described here — the work and the leisure time — could be described also in more utopian terms: there’s less work and it’s more equal; there’s more human connections and opportunities for good times.
Nevertheless, it seems obvious that the time is ripe for a more thorough understanding of the mechanisms to design, change, disrupt and construct platforms. Thus, I’ll move onwards to dissect the platforms and their operational (more typically, “business”) models.
What is a platform, really?
Traditionally, there are three definitions of a platform in business literature. Firstly, it refers to a product platform which is a basis for product differentation inside a company. Secondly, the word means a strategic technology that defines success of companies and their complementators sharing it. Microsoft Windows is the most typical example given about this type of a platform. Thirdly, platform refers to an organisation that creates value primarily by enabling direct interactions between several distinct types of “affiliated customers”.
Following the last definition and according to the books such as “Platform revolution” and “Machine, Platform, Crowd”, platforms are defined here as infrastructures that enable two or more groups to interact. Thus, they are intermediaries that help various groups (Platform revolution lists “customers, advertisers, service providers, producers, suppliers, and even physical objects”) to interact, for example by helping some participating group to build their own services or marketplaces. Successful platforms also tend to operate in demand-side economies of scale: they create network effects and are, as said, multisided, coordinating different rules (governance conditions) for several different groups that participate to value creation on them. While markets are a context where individuals can meet and collaborate, platforms are a context where different groups or groups that have different roles can meet and collaborate.
Platforms are intermediaries that help various groups to interact. While markets are a context where individuals can meet and collaborate, platforms are a context where different groups or groups that have different roles can meet and collaborate.
The main characteristics of platform economy, by which I refer to the economic development that is shaped by platforms, include new rules that are dictated by these platforms, and near zero marginal costs of access, reproduction and distribution.
Why business models are relevant methodological approach to platforms
In short, business model is a story of how a company creates its value. It’s a model that shows the logic with which a particular business creates and delivers value to customers (HBR), outlining also structure of revenues, costs and profits of the company in question. A business model is a conceptual model of a business, not a financial one (pdf).
In more general terms, platforms are important to understand for people working with business models and strategy, because,
“The failure to transition to a new approach explains the precarious situation that traditional businesses — from hotels to health care providers to taxis — find themselves in. For pipeline firms, the writing is on the wall: Learn the new rules of strategy for a platform world, or begin planning your exit.” (HBR)
Specifically, business models are relevant to platforms because similarly to platforms, business models focus on value creation instead of value capture. They are thus the natural approach to take when creating or understanding platform based business. However, the tools presented here are also not limited to business context. I have used the word “business” to tempt people who have money to read this text. A worrisome, anticapitalist and/or poor reader can freely change the word “business” in here to the word “operational”, as in “operational model”.
In the next chapters I’ll first go through some of the common characteristics and features of business models. Then we spend time investigating different ways to ‘hack’ — improve, change, fine-tune, renew, and innovate — platform business models. Finally, I summarize the findings on how to change platforms to a more general social vision on how to have an impact on the society that is moving to platforms.
There are some features that all the business models possess. First, choices and consequences are the building blocks of any business model. Business model is a collection of choices that create a desired set of consequences. (HBR)
Second, it is beneficial if the choices and consequences formulate self-reinforcing choice-consequence-choice-cycles. This is a crucial aspect of platform design the importance of which cannot be emphasised enough. These cycles can be modelled using the tools of business dynamics. Business dynamics is a specific field of system dynamics that concentrates on modelling feedback cycles in business. However, the contribution of business dynamics to the business model discussion has so far been limited: platform business models still lack a standardised, communicative syntax.
Choices and their consequences formulate self-reinforcing choice-consequence-choice-cycles. This is a crucial aspect of platform design the importance of which cannot be emphasised enough.
Third, these systems dynamics loops demonstrate that platform business models typically contain significant intrinsic inertias or delays. It is crucial to take the time-dimension into account in platform business model design.
Fourth, albeit business models are situational, it is also typical for platform business models that they consist of multiple different kinds of potential revenue streams. This feature is especially crucial in multisided platforms.
Fifth, platform business models and business models in general are conceptualized always to representations of reality, not to reality itself. Due to the latter, it is possible to design business models to different kinds of environmental representations. Moreover, the credibility of the representation of the business environment has to be considered separately from the business model.
We’ll discuss these and other platform business model characteristics in the following chapters.
Choices, consequences and theories
While business models design aims at creating systems that provide value for all participating groups, in the unit of business level they are built of concrete choices about how to operate and the consequences of these choices. The choices are linked to consequences by causal theories.
A music artist can make a choice not to charge for her records (a choice) which she thinks (a theory) will boost her audience during tours (a consequence).
[A choice] — [a theory]–> [A consequence]
Choices can be further separated to three different types of choices: policies, assets and governance. Policies refer to courses of actions in all aspects of a company’s operation. Assets refer to tangible resources such as facilities or physical systems. Governance of assets and policies refer to agreements that confer the decision rights regarding policies or assets. A typical platform has a wide variety of potential governance rules, and most of the platform design is actually figuring out the right set of rules for each participating group.
Consequences of these choices made about policies, assets and governance can be divided into flexible and rigid consequences. (pdf)
Traditionally, the set of causal interactions is selected so that it supports the core competencies of a company in order to achieve competitive advantage. But multiple different revenue streams are possible from the same set of core competences.
Choices and consequences form causal loops. In the case of Uber,
a) Decrease in prices increases demand
b) Increase in demand increases the perceived desirability to drive Uber, because there are more rides to be made
c) Increased in perceived desirability to drive increases number of drivers,
d) Increased number of drivers increases the geographic coverage and saturation of drivers
e) Increased geographic coverage increases the speed of pickup
f) Increased speed of pickup increases the perceived value for user
g) Increased perceived value increases demand
and we are back at b.
Uber has several reinforcing loops similar to this example in their business model.
So, to make platform to click, one needs to come up with several self-reinforcing loops that are launched by a specific strategic choice, and that have consequences that support each others.
Business System and Profit Model
The business models shape the core business processes and organisational systems such as work tasks and roles and competences of the personnel. Further, business model innovation is an important driving force in strategy renewal processes, especially when making the paradigm shift to platform business model, so to understand the functionalities of an organisation, managers should understand how business models work.
Itami and Nishino (pdf) approach business models by dividing them into two elements one of which is a business system and the other one a profit model. Profit model is intended method of increasing profits and reducing costs in the company’s business. A business system then is the designed production and delivery system within and beyond the firm’s boundaries: the realisation of the profit model intention.
Profit model is usually more visible side of the business model. For Google (the search engine), it’s collecting and selling user data to advertisers. In platforms, the profit model is the main mechanism of balancing the platform by creating complementary and often unbalanced offerings. Many platforms, including Google search, offer their services to users for free, or even pay users to use them, because of complementary effects in their profit model. McAfee and Brynjolfsson have good points about these offerings in Machine, Platform, Crowd.
Many platforms offer their services to users for free, or even pay users to use them, because of complementary effects in their profit model.
Business system, nevertheless, is sometimes more important to consider as it is the implementation of the strategic intent and also because it is within the business system where the organisational learning and significant network effects take place. Learning creates company’s dynamic capabilities, making long term success possible. Google has experimented with various business system featuring, including the famous “do what ever project you want 20% of the time”.
On the business systems perspective, the firm typically has to determine at least the level of outsourcing, how to organise internal working system and how to control the external activities of its trading partners. These are aspects that change significantly in platform economy. Alstyne and others even write in the Platform revolution that platforms invert the company, meaning, for the most part, the inversion of crucial aspects of the business system. The examples they give about this inversion is marketing (from marketing department to word of mouth and communities), logistics, finance, IT and HR.
Competitive advantage via platform business model
Even though it is not clear in the academia what the distinctive metrics of superior business models are precisely, there are some important characteristics in the design of a business model. This kind of business model 1) focuses on customers, costs and competitors’ capabilities; 2) is difficult to copy; and 3) builds dynamic capabilities.
Customers’ needs, costs and competitors’ capabilities
Profitable business model takes into account customers’ needs, costs and competitors’ capabilities. The bottom line in business model design is (1) to provide value to the customer and (2) capture value while doing so. In platforms, the value capture does not have to happen in every interaction.
Difficult to copy
A good business model is difficult to copy. Different barriers of model copying exist.
Copying business models is usually very simple and thus the competitive advantage they provide is typically not remarkable. However, there are at least four reasons that make the copying of a business model difficult (pdf). First, there might be hard-to-replicate systems, processes or assets, such as in the Walmart business model. Second, some level of opacity may be present. It is not always clear to competitors which part of the business model really builds up the customer acceptability. Third, there might be reluctance in cannibalising existing sales and profits. Further, rigid consequences will slow down any imitation attempt. Also, high reinforcement of the business model will make it more difficult to copy because the competitor must copy the business model features simultaneously. (pdf)
Fourth, distinctively a characteristic of a platform business models, it’s possible to create different mechanisms for customer lock-down towards any participating user group. For example, Facebook practically holds every users’ friendship information and personal data as a hostage, effectively rising the barrier of switching the service very high. On the contrary, Uber has not been very good at forcing a lock-down of its drivers to it’s platforms. Although they have seriously tried to do just that, many drivers use also competing services. Also, the switching costs for passengers is very low.
If any of these are present in the business model or in the business environment, a business model design might provide competitive advantage.
Traditional market research is often not enough to identify emerging trends or unarticulated needs. Distilling fundamental truths about customer desires, customer assessments, the nature and behaviour of costs, and competitor capabilities need to be done when designing a successful business model. (pdf)
Because a good business model is the one that allows a company to pursue its goals in a sustainable way (pdf) truly sustainable competitive advantage can only be guaranteed by developing dynamic capabilities. A part of developing dynamic capabilities is constant business model reconfiguration and innovation via the learning system. Every time an existing business model has been copied and improved by competitor or made irrelevant by environmental events, new business models have the opportunity to prosper. To compete more effectively or to enhance its range, platform companies have to constantly undertake new business systems which often involve elements outside the company’s current core capabilities. This is challenging, often due to intrinsic delays in the business models that create the opacity within the business models not only towards competitors but also to the company itself.
Platform business models and the market
As McAfee and Brynjolfsson describe in Machine, platform, crowd, platform economy changes the relation between companies and the market. I find this an exciting discovery. Obviously, the competitiveness of a company is partly determined by how their business models interact with the environment to create value. Further, the value that a business model creates is dependent on the environment where the business model operates. Exogenous environmental shifts and strategic and tactical moves by other industry players affect the capacity of the firm to continue creating value. (pdf)
Environmental shifts are the driving force especially in platform business model design, because of many participating groups that operate in fine-tuned balance. While there can be innumerable amount of environmental changes, they typically affect the business via changes in customer needs, costs and profits or competitor actions.
In the shift to platform economy, however, it is not only the actors such as competitors and customers that change place, but instead the very strategic landscape is not stable but changes. Although there are events that happen in orderly fashion, there are many events that are sudden, surprising and disordered. What follows is that the single most relevant factor in platform business model design is the company understanding about relevant facets of change in operational environment. Every context is different, so companies must have strong environmental sensing capabilities.
However, sensing is not enough, as we learned in the previous chapter. While technologies and legal structures dictate or allow changes in business models, inertia in changing to platform models in the organisation is usually very strong. Furthermore, newly created platform business models cannot be static. Instead they must consist of dynamic capabilities that allow change over time as the market changes.
As an example of this, Casadesus-Masanell and Ricart studied Catalan companies during significant environmental shift in 1970s and 1980s, finding that many companies did not adapt to new situation. In Catalonia, low cost labor, technology advantage, and large Spanish market protected from outside competition used to create the business model advantage of Catalan companies. For example, high prices did not affect volume in negative way until foreign competition emerged. After the environmental shift, previously victorious business models turned to vicious ones, accelerating the downfall of the companies that were not able to adapt. The companies who were able to adapt were also able to build sustainable competitive advantage. (pdf)
Platform business model is a story
One more thing before we can move to actually understanding the change mechanisms in platform business models.
I mentioned in the first chapter that a business model is a story of how a company creates its value.
A business model is a story of how a company creates its value.
The fact that business model is a story about a platform value creation is not just a nice way to define it. The definition carries an ontology that is very useful.
A business model is a social construct. Business models are conceptualized always to representations of reality, not to reality itself. So it’s possible to design platform business models to different kinds of environmental representations.
We are not limited to what we think exists. We are operating in the world of ideas, constructs. This is important, as it opens a door to business model conceptualization to alternative stories, as the environment where the business model operates is also a socially constructed story about the reality.
So, it’s possible to design platform business models to scenarios and then test scenario hypotheses.
Furthermore, it’s possible to construct models with sensitivity analysis to speculate different outcomes of different models.
The reflection of strategic landscape and timeframe is socially and contextually constructed. Thus, it depends on perceptions about the situation, not about the situation itself. And it’s easier to change the perceptions than the situations. The true value of scenarios and models lie in their ability to change perceptions.
One extremely relevant question remains unanswered. The question is:
what is the value the business model creates, delivers and captures?
Generally there is a difference between moral goods (a conduct of persons) and natural goods (characteristics of objects). Ethics is more concerned about moral goods while economics is more concerned about natural goods. However, both moral and natural goods are equally relevant to goodness and value theory.
Creating economic value
Companies use resources to create products and services. These products and services are accessible to consumers in the marketplace, and if the company has managed to combine resources in the way that the product has more value than the combined resources alone, they can capture this surplus to produce profits. McAfee and Brynjolfsson explain the reason behind this nicely by pointing out that due to the nature of its platform, Uber’s value for passengers is bigger when there are more cars on the streets. It makes sense for Uber to aim for lowest possible price to increase passengers, and thus drivers, to improve geographic coverage and reduce waiting times. Thus, it’s demand curve is not a straight line but a curve, where in the most optimal value capture position for Uber, most users get a very large value surplus.
Economic worth comes from scarcity and necessity. Scarcity implies that choices must be made between different categories; necessity implies that there is no choice of category, just (with luck) a choice of a product in the category.
Value for what tells about the market value of a product. It describes the product characteristics and the price of the product in competitive markets. The decision to buy at market price depends on consumers’ perception of value for money and non-economic value perceptions. Perception is the key term here. It’s central in answering the question “value for whom.” Indeed, price does not equal value, and this is especially true in platforms, where participants often get much more value than they have to pay.
Hacking platform business models with leverage points
Earlier, I have demonstrated how systems dynamics can help in structuring, contextualising and simulating business models. Systems thinking is an abstraction of the real world, a tool. For example, we are able to think of at least 12 different ways to change any system to work for our benefit.
These ways are called leverage points (pdf). Leverage points are points in which the system can be changed.
If Uber would like to increase their hold of their drivers, they could:
Constants, parameters, numbers
Give more money to drivers
The size of buffers
Keep nonactive drivers in the system for a longer time. I have no idea how this particular aspect of their platform actually works.
The structure of material stocks and flows
Open a new inflow, e.g. a recruiting channel for delivery truck drivers.
The length of delays
Increase training time. This improves the metric but doesn’t really help.
The strength of negative [balancing] feedback loops
One of the negative loops comes from the fact that increase of drivers increases driver downtime: this effect can be lessened by guiding drivers to scatter more (which Uber does already).
Power to add, change, evolve, or self-organize system structure
Make it more difficult to practically leave the platform.
The gain around driving positive feedback loops
We could increase the speed of pickups by alarming nearby nonactive drivers if a passenger is waiting close by. No idea if they do this already.
The structure of information flows
Drivers could be made aware of how many people are currently applying to become Uber drivers, which could hypothetically either increase of decrease their willingness to drive Uber.
The rules of the system
Uber could establish a reward mechanism for drivers that stay
The goals of the system
Uber can start delivering packages, thus increasing the desirability to remain as a driver for those who don’t enjoy conversations.
The paradigm out of which the system arises
“Instead of commuting, you as a driver can now communicate via Uber app.”
The power to transcend paradigms
“Why move, when you can stay?”
It is not enough to just to identify leverage points. Often people grasp what the leverage point is, but do not understand how to change the behavior of the system by using it. Using the leverage points is often not intuitive.
Typically, most effort is used to hack constants, parameters and numbers
When we think about making a decision, changing something, we typically think about changing parameters, or values of constants. When the European Central bank changes the interest rate, they change a parameter. When I decide to jog 15 kilometres instead of 12, I manipulate a parameter in my exercise routines. When a company changes its CEO, it changes some parameters. But the underlying structure remains the same.
Because we are used to think that its only the parameters that we can change, we fight about the parameter values and use 99 percent of our time diddling with them, “arranging the deck chairs on the Titanic”. But there is often not a lot of leverage in them. When the interest rate changes, the system self-corrects to adapt to it and same problems arise. When I jog more, I’m in a bit better condition, but there’s diminishing returns and perhaps I jog less often because I’m more tired after the run. The CEO might have a different attitude, but if she is not able to alter the power structures that limited the actions of the previous CEO, there is not much that she can change.
Because changing parameters rarely change the system behavior, changing parameters has the least leverage compared to all intervention mechanisms. Most of our time is used to hack constants, parameters and numbers. We think about them, fight over them, change our minds about them, and most often they don’t really matter that much.
Hacking the Buffers
Donella Meadows gives a nice example about sizes of stock relative to their flows: How often do you hear about lake floods? Rivers flood much more often than lakes, because their flow is much larger relative to the size of the stock than in lakes. Large stocks are more stable than small ones.
Growing the market size (a stock) instead of competing over the existing market sector is almost always the most effective strategy to increase profits. However, large stocks have pitfalls as well. Anyone who deals with shell space gets the next point with ease: system can be stabilized with a big buffer, but with a too large buffer system becomes inflexible. Big buffers also cost a lot to maintain.
In regards to platforms, buffers matter. As written in the Platform revolution, the best case for Hertz, the car company, would be to deliver car to the airport “just in time”, as the phrase goes. But instead of the just in time approach, a platform competitor can utilize a “not-even-mine” strategy by letting the passenger to borrow someone else’s car while in town. Naturally, there’s another stock in place here: the car’s that are accessible. But precisely for that reason picking the right buffer is important. Furthermore, it is crucial for platforms that they create incentives with their profit models so that these buffers are optimal or extensive. Otherwise the passenger will travel happily with her Hertz rental.
However, we cannot fix floods just by increasing the size of flood barriers, because the biggest flood is the one we haven’t seen before. Stock size is a tool to give us time to fix problems in ways I will demonstrate later. Hacking the buffer size rarely actually fixes anything, because in the real world it’s possible to control inflow and outflow only to a point.
Steering platforms by building dynamic capabilities
Building on the previous chapter about constants, we are able to look closer to the budgets of different companies to find out, how companies manipulate constants to build dynamic capabilities.
There are at least eight loops that form the learning system of a business to create dynamic capabilities:
- R&D investments to new product/service research
- R&D investments to product/service improvement
- Workforce via personnel education/hiring
- Workforce via perceived opportunities due to company growth
- Workforce via personnel benefits
- Vertical acquisition
- Horizontal acquisition
Platform companies can use their (expected) profits to invest in all or any of these. The described characteristics are relevant to most if not all companies; these loops are a part of the business model of companies. However, there is no reason why other loops could not exist. In addition, emergent characteristics of business models do provide inter-loop benefits that are difficult if not impossible to grasp by reducing the business model to some of its parts. Moreover, some of the mentioned loops are not extensively or purposefully used by most platform companies.
Material stocks and flows
The structure of the stocks and flows has an enormous effect on how the system operates. However, the physical structure is very difficult to change.
For example, the Finnish government struggles with a large post-war generation that causes large healthcare and retirement plan costs (the sustainability gap). Previously, this same large generation caused large education or child care costs, swelling schools and then jobs. There is not much to do about this, because people age (move through the stocks of schools, hospitals and work places) at a constant speed. In platforms, however, these effects can be mitigated with clever pricing decisions.
Role of delays is crucial to establishing a way to describe and understand time in platform design and platform behaviour. I’ll describe their role briefly here, and return to them later on, especially focusing on them from the valuable perspective of belief system and ideology.
From the systems dynamics perspective, delays are closely linked to stocks. Nevertheless, delays are a very specific kinds of stocks. A delay is modeled using a stock value of which affects the flows, divided by a constant time. Thus mathematically the delay is very easy to demonstrate — and even if the formula might sound counterintuitive, it works.
If system delay is too short, it will cause over reactions — oscillations amplified by the jumpiness of the response. Car accidents often happen when the driver overreacts to a sudden change. If system delay is too long, it will cause damped, sustained or exploding oscillations — fragility.
The tricky part with the delays is to change them to a right direction. Often the direction is counterintuitive. For example, the great push to reduce information and money transfer delays in financial markets is just asking for wild flunctuations.
Moreover, it’s difficult to change the system delays. Things take time. But it might be a good idea to consider slowing the the system rate down, so that inevitable feedback delays won’t cause so much trouble.
For example, in the classic Forrester’s world model slowing economic growth is a greater leverage point than fast technological development or freer market prices. The latter two are merely attempts to speed up the rate of adjustment,
“[b]ut the world’s physical capital plant, its factories and boilers, the concrete manifestations of its working technologies, can only change so fast, even in the face of new prices or new ideas–and prices and ideas don’t change instantaneously either, not through a whole global culture.” (pdf)
There’s more leverage in slowing down the growth of the financial system so technologies and prices can keep up with it, than there is in wishing the delays away.
Negative feedback loops
Negative feedback loop, or a balancing loop, keeps the system at a certain state. Any negative feedback loop needs a) a goal, b) a monitoring device to detect excursions, and c) a response mechanism.
Let’s use democracy as an example of a system which is also, arguably, a platform. In this system there clearly is a negative loop where the actions of the representatives are monitored and reflected upon by the voting public.
Again, we are able to change the system in two ways by manipulating the negative feedback loops. First, we can minimize their impact or reduce them completely. Second, we can enhance their effects.
Let’s first investigate the option to minimize the impacts of these loops. Again, this is something that is often done to gain savings or to enable more rapid growth. Negative loops are often hidden in the way that they are not operating constantly (because they only work when the system is not at the goal dictated by the loop). These hidden balancing loops could be called emergency loops, as they are only activated in an emergency situation.
In principle, a way of manipulating a negative feedback loop is to make it stronger. This can be done in four ways:
- manipulating some of its parameters and links
- increasing the the accuracy and rapidity of monitoring
- increasing the quickness and power of response
- changing the directness and size of corrective flows
Let’s consider democracy as a system to illustrate this point. Democracy system was invented to put self-correcting feedback between the people and their government. The people respond by voting the representatives in or out of office. How can we enhance this process?
- To empower more people, we could manipulate the importance of votes to give more representative power to minorities
- As the process depends upon the free, full, unbiased flow of information back and forth between electorate and leaders, we could make the process more transparent using the media system.
- To enhance the power of the feedback loop, we could decide that representatives that make decisions that are collectively decided to be bad ones are not allowed to run again.
- To increase directness, we could vote the representatives to each question independently rather than voting a set of representatives to take care of all the questions at the same time.
It has often bugged me that, when playing Civilization 5, there is a quote when researching Economics that “Compound interest is the most powerful force in the universe”. The quote is fine, but it’s attributed to Albert Einstein, who would have deserved to get his own technology to the tech tree.
Nonetheless, a positive loops can work in two directions: increase a value, and the value will increase exponentially; decrease a value, and you are in a vicious circle.
Thus, we need to be careful when we use the mathematical terms negative and positive feedback loops. A negative loop is often a very nice thing to have. And positive feedback loops are sources of growth, but also sources of explosion, erosion and collapse in systems.
In any case, being the first one to tap to previously untapped reinforcing loop will surely offer great first mover advantages. So let’s start by discussing the first mover advantage and then move to more specific platform advantages regarding customer lock-up, and lastly looking platforms from the perspective of the traditional protection mechanisms of brands and patents.
First mover advantage is one of the core innovation drivers. It is generally assumed that the first one in the market has the strongest position and is able to gain from the lack of competition with higher prices. This is not necessary the case, though. Continuing on the Civilization 5 theme, it’s often the settlers who reap the benefits while scouts get slaughtered.
In general, a first-mover advantage is possible if there is a gain in positive loops. Nevertheless, the function has more than one variable: it’s not only about the time spent at the market but also the temptingness of the offering.
In network theory, a classic business example is the entry of Google to search engine markets. To model the competition of this new business segment in a representative way, each company needs two values: time of entry is not enough, there needs also be a value on how good their offering is. In a network model simulating the market entry, the probability of a customer to use a specific search engine is the number of the customers already using the particular search engine times the goodness. When Google enters the market, it quickly adapts an ever-growing proportion of the customers. In this particular case, the first mover advantage brought some gains to Altavista, askJeeves and what have you. But those gains were relatively small compared to major advances Google made. However, now that the search engine market has more or less stabilized, it is much more difficult to gain the same advantage using a better value proposition. This is because the market is not constantly flooded with such a high number of users, and because Google can use its huge profits to detect and erase threats before they grow large enough (with a notable exception of Facebook).
The lesson: time of entry matters, but being the first is not necessarily the best of times. Nevertheless, the first in market has a possibility to utilize (at least) customer lock-up and has the best shot at establishing clear brand. Let’s look at these aspects next.
Many platforms manipulate the customers to a lock-up state in brilliant manners. Slack offers a free service that saves the last 10 000 messages, and paid version in which these messages are searchable. This only becomes problematic when the community grows large enough and then switching costs are likely to be bigger than the price of the offering.
Dropbox held a competition between universities by making an offer of extra storage space to those universities that reclaim the most space. While students are in general considered poor and unwilling to pay for such a services, and while they are likely to be volatile and change to another service in a second, they also collaborate in projects, and are likely to have excess cash after they graduate and go to their first job. Dropbox’s offer included a catch: the free space was only offered to the students while they were studying. The time period is, from a students perspective, like forever. But after graduation, the excess space would cost them the normal fee.
Especially the student collaboration is important to Dropbox. When students collaborate, they create shared folders, which makes it many times more difficult and socially questionable to change a service — even after a student project is completed, because the fellow students notice when a student exits from the folder. Dropbox is also cleverly creating gentle routines, for example by placing the Dropbox folder in users sight and teaching the user to use the folder for sharing.
Platform business provides very interesting opportunities for creating customer lock-ups also via manipulating pricing and other governing mechanisms in different segments of the platform. Actually in multisided platforms the switching costs can be bigger than in one-sided markets, as the service is perhaps delivered with significantly lowered price.
Branding platforms — perceived trust and community inversion
I play a board game “Chinatown” with my friends. The game has six rounds. At the beginning of each round, every player gets properties (empty space to build) and businesses (such as a factory and a restaurant). The businesses make profit every turn based on their relative size. For example, a factory needs six properties next to each others and six factory signs to make full profit. But because the new properties and businesses were randomly scattered to every player, players must make deals to change businesses and properties with each others. The nicest thing about this game is that there are no rules for the deal making, except the fact that both (or all) players in a deal must accept it for it to go through.
Intuitively it might seem that the player who makes the best individual deals will eventually win. But that’s not how it goes. Actually, the player who makes the most money to other players is a strong advocate for the win. Counterintuitively, it’s a solid strategy to be nice. Why is that?
Everyone in the game wants to make the deal with the person who seems to be making the fairest deals. And the person who is able to make most deals probably has the least properties and businesses that don’t really make money for him, even if she might loose some money while doing the deals.
“Hyman Roth always makes money for his partners,” goes the quote from Godfather II.
A win-win strategy does not necessarily mean that both parties win in the current deal. Economies and people are more complex than that.
In regards to platform business models, it’s good to understand that the business model constituents can be more than just real, tangible business assets. It’s very useful to consider different perceived aspects of the model as well, asking, for example, how this particular choice affects our perceived trustworthiness.
Because the positive perception is increased when the platform company collaborates with its participants also outside the actual transactions, branding in platforms is more and more about collaboration and community.
As mentioned in the Platform revolution, branding and marketing is one aspect of company core functions inverting in platform economy. Nowadays a lot of money can be saved by being smart: finding the right communities, using word of mouth, and even harnessing the user base in choosing right strategic choices.
It’s curious how well platforms seem to be doing in the branding front, taking in consideration the fact that they rarely if ever actually collaborate with user directly. Many platforms use reputation systems that they themselves guarantee to build trust in the peer to peer interaction, but because of it’s seemingly collaborative nature, the internal crises in Uber has done little damage to the trustworthiness of the Uber’s reputation system used in the app.
The information flows and opening of data
Students of demand supply chain management play a very illustrative board game called “beer game” to understand the complexities caused by a super simple supply chain.
What makes the game interesting is the lack of knowledge about demand throughout the supply chain. Players only know the what is the demand of their buyer, and based on that ask for a supply. If they are unable to deliver the demand, they get a backlog, which they of course put forward in their own orders.
The catch is that the consumer demand is practically constant throughout the game, but the players don’t know that. Because of the delay in the chain, the system goes to a wide oscillation and gridlock (consumer demand is the black line):
The blue line is the demand in the other end of the system — that is what the factory needs to produce. Of course, no factory would be able to deliver orders ten times the demand once or twice a year.
Adding information flows means, structurally, adding a new loop delivering information in places where it wasn’t going before. This is typically much cheaper than building physical infrastructure. Often the goal is to reduce the delay between actual and perceived situation.
I once heard an urban legend about two apartment buildings that were similar except that the other had an energy meter in the hall showing the consumption of the whole building, and the other building didn’t. And according to this legend the energy consumption was 30% lower in the building with the information feedback loop.
Clearly, if the customer demand in the beer game would be visible to the factory, the whole chain could optimize their operations so that the demand does not oscillate (wildly).
Rules of the platform
Changing the rules of the system is a leverage which is used extensively within platforms, because platforms are in essence a collection of rules.
When I was in the first grades at school, we played football between the classes. When there was snow and ice outside it was really hard to play, so we changed the rules. We decided that there was only one rule: no carrying of the ball with hands.
A bunch of creative kids with that “no rules” statement said out loud — you can imagine what happened. Each of us tried to invent the most clever way of getting out of the mindset of football. We carried the ball inside our shirts. We throw snowballs in defence. The big tough kid was impossible to attack against because he could tackle you with hands (and, mind you, the field was not snowy but covered in ice). Once, few kids changed sides in the middle of the attack — it wasn’t a betrayal because there were no rules. Man, that was fun.
The rules of the system define its scope, boundaries and degrees of freedom. Meadows writes “If you want to understand the deepest malfunctions of systems, pay attention to the rules, and to who has power over them”.
One peculiar characteristic of system rules is that they define what are considered system in the first place in each level of perception.
Thinking in systems means that certain subsystems are considered as one system, which interacts with other systems. Defining a system boundary defines what is in the “box” — what is considered a system. Because it’s possible to think systems on every level of abstraction (say, a forest consists of flora and fauna, which consists of e.g. insects, mammals and amphibians, which are frogs, which have legs and torso, which is built from molecules which consists of atoms and so on and so forth. It makes no sense trying to map the behavior of the forest system from the atom level, because the emergent properties of all the other system levels would not be seen. Similarly, it makes sense grouping a system of molecules to a single entity, called the frog, to understand the behavior of the frog system in relation to the environment and other frog systems in the environment.
Changing the perception of what is a system and what is not might have significant impact on the behavior of the systems. An example of this is the way business networks are nowadays often considered as a meaningful entity, when before this conceptualization they were merely thought to be competitors or parts of a value chain system. The new rule that allows a manager to think the network as an operating system creates new business opportunities — for the whole network!
The power to add, change, evolve or self-organise platform structure
Changing the actual system structure is of course a powerful leverage point. This sounds easy, but typically it’s not. Systems, especially social systems such as businesses, lay on a complex power networks and contradictory intentions, which cause significant friction to any system development.
It is not likely that the action which follows a presented system model will be implemented as a whole. Checkland, in Systems Thinking, Systems Practice, writes:
“In general in these more nebulous problem situations, the eventual action is likely to be less than the implementation of a system; it is more likely to be the introduction of a more modest change.”
According to Checkland, there are three kinds of possible changes: those in structure, in procedures and in attitudes. Structural changes alter an aspect of reality that in the short term would not have changed without an action, such as structures of responsibility in an organisation. Procedural changes change the dynamic elements such as the processes of reporting the activities that happen in static structures. Both procedural and structural changes are easy to specify and relatively easy to implement by authorities. However, they might have side effects which were not anticipated. This is not the case with the changes in attitudes. These changes happen steadily as results of shared experiences of people in groups and they are affected by the changes in structures and procedures.
Due to power structures and the fact that a system structure is rarely decided by a single entity, the most important way of system structure to change is via self-organisation.
Self-organisation means adding completely new physical structures to a system, such as new brains (as in recruitment), a new pair of wings (as in evolution), or computers (as in digitalisation). It also means adding completely new negative or positive feedback loops — making new rules.
Self-organization always has system derived rules how it operates. These rules dictate how, where and what the system can add onto or subtract from itself under what conditions. In software development, machine learning provide this kind of behavior.
The key point is this: the power of system self-organization is stronger the more material there are for positive evolution.
The goals of the platform
Earlier we learned how every balancing loop has a goal. From the perspective of the goal of the total system, these loop level goals are microideologies, and their impact on the goal of the whole system is significant.
The way to find a system level goal is not to ask “what is the goal of this system?” That would only provide an answer that is the litany: “The goal of a business is to create value to the shareholders”. The correct way to identify a system level goal is to look at the system and see what it does, really.
If the goal of the corporation is to make profits, that is just a rule to stay in the game. What is the point of the profits? Why that rule is being implied in the actions of the company? When you look at the actual behavior of the system, it seems that the goal is to bring the world more and more under control of the corporations. Companies do this to shield them from uncertainty (however, they are failing badly as the system fragility is increasing due to these “stabilizing” actions).
Survival, resilience, differentiation and evolution are examples of system-level goals.
It’s crucial to identify the actual goal, because what you are trying to archive will dictate the other ways of evolving the system. Understanding the goal makes it possible to change the goal, or accept it and thrive for it.
To rip apart some of the commonly accepted system goals, let’s take a quick look to a futures research method called Causal Layered Analysis (pdf), CLA in short.
In CLA, an object is inspected from four perspectives. The first level is the litany — “the goal of a business is to create profits for the shareholders”. Often, litany level consists of problems and trends that could be used for political purposes. Litany is difficult to question, because it is said to be the truth. When litany consists a problem, such as the inevitable warming of the planet, the end result is typically feeling of helplessness.
The second level is the social causation level. In this level, the litany can be questioned, typically with data. In the case of the climate change, a researcher could cite sources showing the decreasing prices of solar panels and global homeostasis mechanisms.
The third level is the worldview, or the discourse. In this level, the deep, ideological assumptions are unpacked. By looking at the discourses and the words used in the litany, it is possible to investigate in which way the litany and the system is constructed by the stakeholders. In the climate change example, this level could consists of analysis on how the climate change interest groups are presented in the media, or how concepts such as the american dream or right of development in developing countries impact on the discussion.
The fourth level is the level of the myths and metaphors. They are the unconscious emotional dimensions of the topic. The researcher should be able to go beyond the conventional framing of the issue. Is there a battle of good and evil? What is the metaphor of the planet in the climate change discussion — a mutual spaceship or a battleground?
In CLA, the researcher moves up and down these levels to undersand the deep, underlying concepts and assumptions behind different levels. This allows the researcher to create authentically alternative futures and transformations by questioning the litany, social causation, discourse and myths of the current perspectives towards futures.
When we use CLA to discuss about the growth of companies, perhaps we could end up in a following categorization:
Litany: Companies need to lay off people to serve their purpose — to maximize shareholder value.
Social causation: Laying off people reduces the purchasing power in the society, reducing market demand.
Worldview / Discourse: Corporations exist outside of the civil society, or at least they are in a high hierarchy within it. They have they right to utilize people, because people are resources. Growth is maximized by competition.
Myths/Metaphors: Growth is good. Thus, it is the only reasonable goal for a company.
Hacking a paradigm
When Thomas Kuhn adopted the word paradigm to refer to practices in science that dictate what is observed, what kinds of questions are supposed to be asked, how these questions are structured, how the results are interpreted and how an experiment is conducted, he might have had a clue that the word ‘paradigm’ would have a life of its own and soon it would point to any kind of current set of discourses.
In short, a paradigm is a shared social agreement about the nature of reality. Paradigms are the sources of systems: they create and maintain system goals, information flows, feedbacks… Those who change paradigms hit leverage points that completely transform systems.
Paradigms are the sources of systems: they create and maintain system goals, information flows, feedbacks… Those who change paradigms hit leverage points that completely transform systems.
Russian formalist whose name I cannot recall had a beautiful explanation about what in each epoch is categorized as (true) literacy and what is not. At any period of time, what is literacy is a small subsegment of what could be literacy. And the sphere of literacy moves around the what-could-be-literacy similarly to the large storm in Jupiter.
As we have been taught be anonymous alcoholics, the first step to solve a problem is to identify one and admit that it plays a role. So the first step in changing paradigms is to have an idea about the current paradigm — the one that needs to be changed. Look. What do you see? Where is the focus now?
When you have an idea of what the paradigm is, you might start to see its pitfalls. You might have a gut feeling that there is another way, and perhaps you are even able to find words to describe your intuition.
Kuhn gives four tips for changing the paradigm. While remembering that he was talking about scientific paradigms, who is he to draw the lines, anyway? Kuhn’s strategies were
- Keep pointing at the anomalies and failings in the old paradigm
- Speak louder and with assurance from the new one
- Insert people with the new paradigm in places of public visibility and power
- Don’t waste time with reactionaries; rather work with the active change agents and with the vast middle ground of people who are open minded
Beyond companies: platforms as an organising principle of society
Legitimacy of platforms
“Legitimacy is a generalized perception or assumption that the actions of an entity are desirable, proper and appropriate within some socially constructed system of values, beliefs and definitions.” (Suchman 1995, 574)
This means that legitimacy is what makes organisations natural. Thus, legitimation and institutionalisation are synonyms: when an individual uses legitimation to increase her abilities to act, the individual becomes an institution.
Let’s consider legitimation from an organisational level. It’s a process by which an organisation seeks approval or avoidance of sanction from groups in society (e.g. Kaplan and Ruland 1991, 370). From this perspective, legitimacy is an operational resource.
Legitimacy theory posits that platforms are bound to social contract in which the firms agree to perform various socially desired actions in return of approval of their objectives. When companies do not perform these actions but instead seek quick wins (the delays in trust based loops are longer), they wreck their changes of successful continuity of their businesses in many ways. On the other hand, if platform companies understand this side of their business model, they are able to utilise it towards various participating user groups.
From place to purpose
It’s impossible to understand the ancient Greek democracy from the perspective of today: this is known for sure. The current individualistic view on voting is completely detached from the democracy of Greece, where the main actor was not a person but a household and a city (oikos, polis).
Many big names of the social sciences, such as Weber and Tönnies, relied their view of the world to an idealised concepts of towns. These kind of towns are not relevant anymore. Our poleis are platforms.
Do the platforms really change out perception from place to purpose? Clearly, the 90s vision of the world without a place was a false utopia. There are many good sources that point out that world is more place oriented than ever. For example, most of the content is created in New York, and most startups head to Silicon Valley. Moreover, platforms such as AirBNB are more often than not place oriented businesses.
Although places matter in the markets and networks, platforms are making it an illegitimate context to do politics or establish and maintain societies, even though it inevitably used to be.
So, the place does have a meaning, perhaps even more meaning than it used to in some respects, but that it’s not what constructs the field of politics and participation anymore. Instead, the new local is the platform where people share purposes, ideas, values and so on. Facebook groups that focus on local city planning are driven together by common purpose, not by common place (as not everybody in the place is involved).
Tönnies used to criticize this view, because according to him, a community is something that shares all the purposes and values, while society (and organizations) is the collective of I’s that only share some purposes. Nevertheless, the local (as in place) values, ideas and concepts are nowadays not shared — instead, the shared concepts are found in platforms.
Platforms as the organizing principle in the society
Platforms ought to be thought from a wider perspective than just one approach to business, and from a wider perspective than just overlapping digital tools. They are, of course, precisely these, organising collaboration between the tension of openness and access and regulation and rules, but their core characteristic is that of the feedback loop and it’s ability to create compound interests.
Progress, previously, was piling of stones. Each new stone was a bit higher than the last one: this was the shoulder of the giant that Newton talked about. But in the platform society, progress is piling of mobile phones: every new phone adds in the pile as many connections as there are in phones in the pile.
In platforms, the focus must shift from units to relationships between units. This is a radical transition.
Feedback loops are a key feature in enabling the scaling, durability, and cumulation of value in platforms. Platforms are built from strategic choices which support the feedback and create cumulative consequences.
So, what then is the new progress about? Previously progress was slow exchange of value from one type to another, for example, from ecological wellbeing to social wellbeing due to new washing machines and cars. Platforms change this: progress in the platform societies is ortogonal to this kind of correlations. The correctly designed platform feedback loops allow for breaking correlations, such as the aforementioned one, or the ones regarding traffic (cars / jams), food (price / ethics), work (too much work / too little work), equality and participation.
A new field of social research
Clearly, to understand platforms, there is a need for a new field of social research. The first element of this field of social research is the feedback loop as a central organising principle of progress. The second element of this field of research is the delay (or “waiting”). Delay bounds the feedback loop and thus the analysis in time. This opens the door for investigating beliefs, desires and microideologies in the context of time.
Platforms are quickly becoming the most important context for human interaction. Unlike markets, in which individuals come together to interact, they allow different groups or groups in different roles to interact and collaborate. Thus, they are more than just a new set of business models. Nevertheless, looking at platforms from the perspective of the business models is useful, in part because they focus on value creation instead of value capture, and in part because business models offer a useful theoretical framework in modelling, understanding and conceptualising platforms and choices that are made on platforms.
Most platform decisions are made by changing the governing rules and policies by which the groups interact. This includes pricing decisions, where oftentimes the demand curves are surprising and support extensive scale.
From changing constants to changing the goal of the platform, there are several different ways to change the way a platform operates. These tools are important, because platforms, constructed from the self-reinforcing loops which all contain a desired future, are microideologies that can have conflicting and undesirable consequences.
A new field for social research is needed to better grasp the perceived time and delays, which have a grave impact on the behavior of the system, the exponential growth in reinforcing loops, the perceived fragmentation and randomness in behaviors on platforms, and in relationship between markets and platform companies. The two main elements of this new social science ought to be the feedback loop and the delay. Within this new social science there needs to be a new conceptualization of progress which is natural to platforms: progress as ortogonal movement away from the unhealthy correlations of the industrial era.