Neil Perkin kindly asked me to join in with a bunch of far smarter folks to talk about innovation in agencies at the Firestarters event he curates for Google. As I’ve never been an Innovation Officer or suchlike, I focused on how planners can inspire innovative behaviour in agencies rather than how agencies can innovate their business models (which probably took me a little way off brief, but there were 7 other presentations that did stick to the subject)…..

My slides are below, with a bit of commentary, but the general gist of it is that planners are by definition the people who cram their heads full of knowledge about people, culture, technology and change, and in presenting this to colleagues and clients, have a commercial imperative to make ourselves look smart. This can (looking at some posts not far beneath this one….) lead to wildly mashing up analogies and theories, which take some explaining. As a lot of what I do focuses on making media for clients, I used an example of the work that I do at LBi to show how to intermediate knowledge and turn it into something that is useful to people whether of not they have read the books/understood the theories. The grids at the end come from the Network Planning approach that I’ve been developing here, also heavily based on the ideas of Mark Earls, Alex Bentley and Daniel Kahnemann

With apologies for the overly militaristic use of analogies…..

So I’ve been thinking about Staging on client projects over the last few months, and explained if in a few presentations using this image of how a fission bomb works:

which is basically along the lines of: two sub-critical masses of fissile material are imploded by conventional explosive in a device lined with a neutron-reflective material, causing the the fissile material to go prompt-critical – ie create an immediate self-sustaining chain reaction, with explosive power orders of magnitude greater than the conventional explosive used to trigger it.

And this gives us a fairly forced analogy to explain how a great idea (fissile material) that crossed over into popular culture has an impact exponentially greater than the conventional explosive (advertising) that had catalysed it. As analogies go, it’s probably better for shock value than, say, trying to explain that if you rate ads on a scale of 1-10, then it only works if that scale is logarithmic (a Richter scale rather than a Millward Brown scale).

When we think about it through the lens of planning for a predictable series of unpredictable events though, it becomes more interesting. The Manhattan Project, the original fission bomb designers, knew that nuclear weapons were theoretically possible before the US even entered World War 2. It took the combined brains of most of the world’s most brilliant physicists and mathematicians to calculate the high energy hydrodynamics equations to work out what materials to use and in what order (the linear part of this process happening in a couple of nano-seconds). Or, the combined brains of all but one – Edward Teller decided at an early stage of the project that fission bombs were simply an engineering challenge, of manufacturing pure enough (‘weapons grade’) fissionable material, and focused his time on the potential to create weapons powered by nuclear fusion. Given the challenges that marketers have experienced trying to create even the ‘fission’ concept – the non-linear cultural explosion of a brand idea through a population – it is worth thinking about what he found.

Fission weapons were outdated within six or seven years of their invention, as Teller and Stanislaw Ulam developed the concept of Staging  - the Staged Radiation Implosion model, or thermonuclear weapon. When i started looking for examples of staging non-linear impacts over a compressed period of time, there seemed to be some analogies relevant to us, particularly as I’ll come on to, in the development of ad-funded programming or similar programme of branded content with high upfront development costs. Which is a roundabout way of saying, bear with the science geekery…..

The basic principles of thermonuclear weapons are as follows:

Stage 1 – Implosion of sub-critical quantities of plutonium 239 to trigger a fission reaction. The principles of the fission bomb are simply the first stage to catalyse an exponentially larger reaction

Boost 1 - Deuterium gas in the centre of the core is compressed and heated by the fission reaction to induce fusion, which increases the speed and efficiency of Stage 1.

Stage 2 - X-ray radiation from the initial reaction is focused by beryllium lenses  to compress the Lithium Deuteride fusion fuel to a fraction of its original density – causing deuterium nucleii to fuse, creating a fusion reaction. As the X-ray radiation from stage 1 travels at speeds orders of magnitude higher than the kinetic energy released by the Stage 1 fission explosion, the structure is still in its original form by this point.

Boost 2 –   A core of plutonium inside the fusion fuel is also compressed to critical density, and the ensuing secondary fission reaction heats and speeds up the Stage 2 reaction

Stage 3 – a blanket of (relatively) inert and cheap to produce Uranium 238 wrapped around the Lithium Deuteride core gains neutrons from the fusion reaction, and is heated  by the previous reactions to several million degrees, unstabilising it enough to trigger a further fission reaction. Average explosive power of a thermonuclear weapon tends to be 50% derived from this final stage – the result of multiple previous staged implosions to create the esoteric conditions under which a cheap and easy to refine isotope can provide extreme energy.

So the ‘value’ in this chain is not in any single part, it is in their sum. More importantly, it is in the understanding of high energy hydrodynamics of a complex system to ensure that the conditions are created to allow reactions to happen. And even more importantly, that they happen at the precise fraction of a nanosecond before the next stage is engulfed in an exponentially expanding ball of plasma hotter than the centre of the sun.

So anyway, I’ve been at a couple of branded content events recently, and the question of predicting success and ROI from heavily front-loaded investment in programme production came up. As they should do, when the actual content a brand can feature in a programme under OFCOM regulation make it difficult to deliver value on air. This leaves a brand investing in creating something editorially impressive enough to be commissioned, with very little predictable success to show for before it airs. Kind of like the idea of using a first stage explosive powerful enough to destroy a city simply to compress three strips of metal down to a smaller size.

In this analogy, the actual AFP is Stage 2. Without stages 1 and 3, the AFP is simply the creation of the ideal programme to sponsor, with the brand existing primarily within bumpers. The Stage 1 energy is derived from social channels’ involvement in the creation of the programme itself, catalysed by conventional paid media to provide initial impetus to the reaction. This focuses attention on a pre-existing cultural object which already has brand ownership through its association with brand fans’ participation in the creation. Linear transfer of attention from stage 1 to stage 2 is focused though linear channels – bought media to amplify fan reaction as part of the tune-in campaign to Stage 2.

But so far this is only playing to fans – Stage 3, over 50% of the value – involves the rapid transfer of attention from the on air idea to a wider audience than actually watch the show – it relies on social transmission of cultural interest from fans and from viewers to catalyse the attention of the wider population – to create a third stage critical mass.

Clearly this is all wildly ambitious for an advertiser funded programme, but the key in thinking about value in a staged model is in thinking about the hydrodynamics of mass attention – thinking about how non-linear effects are achieved, but more importantly how they are compressed in time in such a way as to catalyse the next stage. This is fundamental in thinking about WHAT the roles of owned social channels are in inspiring participation publicly, and what those of paid media are in providing the lenses to focus catalysts at the speed and at the level of compression required to ignite the following stage.

In Part one, I looked at the correlation between bought, owned and earned media and linear, dynamic, and complex systems. Since the bought/owned/earned terminology become popular a few years back,  we have increasingly thought of media as an ecosystem, highlighting the importance of connections between nodes in building success.  However, in successful media ecosystems in the real world there is usually a linear narrative connecting non-linear effects – there are a number of Stages which each achieve an impact orders of magnitude greater than their input from the preceding stage.

A Linear Progression of Non-Linear Events
The Arab Spring revolutions of 2011 are a great example of staged media. They were seen in the West as Facebook revolutions, events that couldn’t have happened without social media. This is probably true, in the same way that the hurricane in Part 1 couldn’t have happened without the butterfly flapping its wings. But also couldn’t have happened without all the onward Stages. Ethan Zuckerman, who runs the Center for Civic Media at MIT MediaLab, summarises the linear progression of non-linear events that lead to the Tunisian Revolution in this video. It is well worth watching, but at 18 minutes long, I’ll also pick out the key Staging principles.

Catalyst 1
Mohammed Bouazizi, an unlicensed street trader in Sidi Bouzid, set himself on fire to protest at the confiscation of his goods for his inability to pay police bribes. This brought protests throughout the town at his funeral. Similar localised protests happened frequently in Ben Ali’s Tunisia. Government response tended to tread a fine line by quelling with force without raising wider interest through massacring civilians. The threat of disproportionate response was maintained by the government’s tight control of the press and strict bans on international TV reporters.

Staged Amplification 1
Smartphone videos taken by protesters were put on Facebook (the only unblocked social network available to them). This DID NOT result in Tunisian citizens seeing and copying protests, as the government had phished passwords for the majority of the country’s Facebook accounts, and even slightly activist-minded Tunisians knew that viewing dissident material on Facebook would be monitored by government intelligence

Catalyst 2
The people who DID see the Facebook videos were the exiled dissident bloggers running Nawaat, who redubbed them in Arabic, French and English and passed them to Al Jazeera

Staged Amplification 2
Al Jazeera was banned from reporting in Tunisia, but was watched all over the country. The on the ground reporting distributed by Nawaat was seen throughout Tunisia

Catalyst 3
The ability to protest against the feared Ben Ali regime without being indiscriminately slaughtered inspired immediate copycat protests to spring up all over the country

Staged Amplification 3
As fear of using social networks diminished, increasing amounts of on the ground content found its way directly to Al Jazeera, who focused the eyes of global media on a significant national event

Catalyst 4
Copycat protests sprang up around the Middle East, supported by global media coverage and leading to the overthrow of dictatorships in Tunisia, Libya and Eygpt

The key point for thinking about complex systems in all of this, is that the individual progression from stage to stage is a linear and predictable one. Protests hadn’t previously spread out of confined towns because no media could get in or out. Smartphones and social networks change this. Social networks were of no use in spreading media internally in such a climate of fear, but could transmit information immediately to international sources. Al Jazeera were banned from reporting, so knew that something was worth reporting on. Millions of Tunisians watched Al Jazeera.

So although each stage delivered a non-linear impact in media amplification, it was also relatively predictable. Underpinning this was a huge national appetite for change, which meant that the Stages were all in place and primed.

In thinking about what this means in the world of marketing, we need to understand not just the potential non-linear impact of Stages, but also the underlying motivations at population level that will provide the speed and scale for activity. To use the old Forest Fire analogy, it isn’t enough to drop lots of matches, you also need the right conditions for fire.

Please ignore the word ‘prediction’ in the title. This isn’t going to be ‘planner buzz words for next year’. More about the general unpredictability of prediction, and how we might learn from it. I’ve been reading Nate Silver’s very smart “The Signal and the Noise”, which takes the methodologies he has learnt from poker, baseball and most famously politics, and looks at how they are applied in economics, geology and climate science. For all the promise in the 1970s that computing power would help us predict storms and earthquakes, science’s greatest success has been to give a couple of days’ more notice of hurricane landfall. Although we can predict to a high degree of accuracy how often a Richter magnitude 7 earthquake will occur for a given faultline (an order of magnitude less frequently than one measuring 6 on the scale), this is of no use in predicting whether one will next strike San Francisco - all we know is that one will, and it will happen at some point in the next 50 years, give or take a few.

Where this starts to become relevant to marketing science is in Silver’s succinct explanation of why we can’t predict – the complexity of non-linear dynamic systems. The classic example of complexity is the butterfly flapping its wings in the Pacific and causing a hurricane in the Atlantic – while this is a neat illustration, it does play to our natural post-rationalisation bias, in the sense that a butterfly flapping its wings would have been one of many millions of factors contributing to the hurricane and without it the hurricane would not have formed, but could not be said to be directly causual.

Silver’s explanation considers two factors – non-linear systems and dynamic systems. When we set out to predict something, we invariably make inaccurate assumptions at the outset. Where non-linear relationships are involved, an initial misplaced digit can result in huge variation. For example, a linear calculation looks like this:

5+5 = 10

5+6 = 11

the single digit variation between the two calculations results in 10% difference in the results. However, the same figures in non-linear form look like this:

5 to the power 5 = 3,215

5 to the power 6= 15,625

Leaving a 480% difference. So in one step, we could be several hundred percent off course in our predictions. In years gone by this was of relatively limited concern, as in most cases the success of marketing ideas was directly linked to the amount of money spent on ensuring people saw them. In a few cases the ideas themselves were so strong that they crossed into popular culture, crossed from linear to non-linear dynamics, but in most cases there was a linear relationship.

As we focus more as an industry on the world of social networks, search optimisation and  piggybacking on popular culture to transmit ideas, we not only become less linear in our planning, but also start it to run into the second amplifier in Silver’s models: the dynamic system. Dynamic systems take the output of the first stage as the input for the second stage. In simple terms, this is the reason that pollutants in the food chain impact higher predators disproportionately to the herbivores that initially come into contact with the chemical. It is the reason why today’s (and realistically tomorrow’s and the day after that’s) computers can’t predict the weather with any accuracy more than three days from now. And as we increasingly see marketing as spreading ideas through networks it makes predicting success, and therefore justifying investment, increasingly difficult.

This is a challenge when we think about the most effective reception and transmission of ideas, particularly in small world networks. Small world network structure has little nodal centrality – there are no experts or influencers, and ideas are received based on personal relevance and transmitted based on reinterpreting them for a similarly small number of people. To succeed at scale, we therefore need many interpretations of them, the majority of which will not be created by a brand: ie. we rely on non-linear media structure. As each transmission can be seen as a separate stage, networked transmission is by nature dynamic. So while we can predict success, we are doing so in the sense of earthquake prediction: we know the probability of success at a magnitude 7, but we don’t know which idea will deliver it. Non-linear success is logarithmic, so magnitude 7 is 1000 times greater than magnitude 6. Forecasting results becomes inherently unstable, as although we might be able to predict results over a year, we can’t divide the target by the number of weeks of a campaign to set shorter term targets: 90% of the success might well happen in 10% of the period. To someone whose experience is based on linear results, and whose investment decisions are guided by the principle of not failing, this makes setting a course for non-linear dynamic success very difficult.

The first challenge to overcome is around dealing with uncertainty. I’ve illustrated the uncertainty of prediction, which can be seen to correlate with the principles of Bought, Owned and Earned media. Excluding for a moment the obvious irrelevance of treating any one of those things as distinct or separate from either of the others – we’ll come on to that in Part 2 - Bought media works as a linear relationship. You get what you pay for

Owned systems rely on a multi-stage approach: the output of search optimisation is the input into your site, and the output of your conversion optimisation is the input into your content or e-commerce journey. In years gone by owned Facebook pages worked in this way also – the output of your Facebook content strategy in month 1 interactions was your available newsfeed audience in month 2.

Once we move into networked distribution, the potential for both success and failure become amplified exponentially – whether in embedding ideas into popular culture (the tried, tested and inherently unpredictable approach that can result in a Levi’s Launderette, an Old Spice’s Man Your Man Could Smell Like, or a Compare the Meerkat) or in stimulating the reception and remixing of ideas in small world networks (arguably the same thing, but using the internet as a proxy or a vehicle for actual transmission)

So in setting ambitions and predicting success, we cannot rely on non-linear dynamic systems in isolation. But the potential for success in this space is clearly greater, so we need to focus on the dynamic system first and foremost, an then understand how we can apply linear mechanics to unlock non-linear results – so effectively how we move from thinking about an ecosystem of bought owned and earned, to a complex system of staged amplification.

There are plenty of examples of staged use of media within complex systems. In Part 2, I’m going to look at the Tunisian revolution as an unpredictable but post-rationalised use of staged media. In Part 3, I’m going to delve back into marketing’s fetish for military analogies and think about the Teller-Ulam design of Staged Radiation Implosion – the thermonuclear bomb – and what it can teach us about predicting staged outcomes in complex systems

It seems that barely a week goes by at the moment without another agency group launching another real time data planning unit. The opportunity to focus ad targeting and messaging down to an individual level is being snapped up by all the main holding companies, offering a potential to craft campaigns focusing on exactly the discrete audiences of the right people at exactly the right time.

Which is all well and good, and represents the logical extreme of what media agencies have always tried to do. However, the social sciences, who already know a fair bit about how people make decisions, would suggest that it may not be the right way to apply the technology that we have at our disposal. Daniel Kahnemann in psychology and Richard Thaler in economics have shown that rational thinking plays very little role in our decision making; we naturally default to habits and rules of thumb. Robin Dunbar’s work in evolutionary psychology and Nicolas Christakis and Paul Omerod in public policy highlight how much of our habitual behaviour is derived from our interactions with other people. Peter Field and Les Binet’s work on the IPA Databank show that advertising has the greatest chance of success if it sets Fame as an objective – this makes sense in the context of our predisposition to learn socially, as it is a substitute for other visible cues of popularity: it gives us something to copy. However, micro-segmentation as a basis for targeting individuals comes with no social context, no understanding of people who live in networks not as individual rational units.

That isn’t to say that data isn’t hugely important, rather that we need to think wider about how it is applied and for what purposes. Demographics have always been used to average out difference, and create target audiences of individuals grouped together based on similarity. Real people on the other hand are diverse, passionate and socially adept at living in networks of other people. Demographic data made sense in the making of television (whether ads or programmes), as TV needs to tell a story universally relevant to a large number of people in a short space of time. TV injects cultural energy into human networks. However, time is short because costs are high, so brands have to ensure that messages are simple and have broad appeal. When creating ideas that live in the infinite number and variety of free media that are now at our disposal, we believe that the most powerful source of data is still untapped; the ‘data’ held about us by our friends. Those people that Mark Granovetter refers to as our ‘strong tie networks’, the people closest to us (and those that Facebook’s Edgerank algorithm for example is designed to approximate) know far more about each of us than any data source available to marketers. However, for a brand with big ambitions, this data is hard to scale. So how do we make use of ‘Small Data” to create ideas for millions of micro networks?

I believe that the value in Big Data is not in the data itself, but in the processing power that we apply to it. This allows us to set objectives based on patterns in data. Understanding which behaviours may be changed through social learning, from our closest ties, which are learnt through wider observation of people around us, who form part of the bigger communities we are part of, and which are learnt from experts, is of fundamental importance in setting the direction for creative development of ideas. Particularly if those ideas are going to be inclusive enough to spread through communities. It also allows us to think more accurately about the transmission of ideas. As Mark Earls and Alex Bentley outline in I’ll Have What She’s Having, social learning is often about choosing to receive an idea, not choosing to transmit it. Thinking about the role of ‘influence’ in marketing is impossible without this framework. After all, Oasis were influenced by The Beatles, but that was not the objective that The Beatles set out with.

Using Big Data to map network structures has far greater value in showing how ideas spread at an aggregate level than in understanding the individual. A small world network (ie one that has no central nodes, or ‘influencers, as classical marketing would call them) requires a very different approach to one that is scale free (and so has central hubs that are disproportionately connected). And this in turn has a major impact on the type of idea that will affect behaviour at a mass level. The standard approach to a small world structure (where one has been identified) is to create many small ideas that are set loose into the world on the basis that prediction of success is impossible – you might just as well try and predict stock market – so you see which take off and get behind them. However this isn’t an approach that marketing directors are often comfortable with: it sounds expensive.

Instead, I see the power of Big Data as providing the direction in which you can pursue something of far greater value: Small Data. Using small data means creating things that people personalize to make relevant for their friendship groups – from turning a real life experience into shareable media, like Luna Park in Sydney to selling named bottles of Coke across Australia to create opportunities to buy presents and take personalized product photos, to Smirnoff tapping into intercity football rivalry to vote for and against the UK’s top clubs, to the many remixes of Mastercard’s ‘Priceless’ meme/ad. It is about remixing ideas because that has value to the individual and their friends, rather than because a brand thinks that lots of people want to re-edit their ad. It isn’t about people participating in a brand’s idea. It is about brands participating in people’s ideas.

Those examples above are about small data, but they are also small in size. What I’m trying to do on this blog is play around with the approaches that make them bigger. Big data is a starting point, but what all those examples use is Regular Data: sales data, visitor data, media data, that has been analysed to focus on WHY they exist, whatever form they take. Too often marketers and agencies who are focusing on Big Data, or on the Small/social sides of data, lose sight of the data that we have always had access to: the bit that explains why we are doing what we are doing, and why it is the most effective route to growth. Insight, rather than data. To paraphrase a line from Rory Sutherland:

Without a theory, data is useless. Without a theory, all Darwin had was “some funny birds and lizards and shit”. Darwin saw insight in the data.

To solve a problem, first we need to define it.

In the MIT Media Lab’s introduction to the idea of what media are, Nicholas Negroponte tells a story of how the Media Lab was named. The university executive told him he could call his new department anything he liked, as long as it didn’t involve the words ‘computers’ or ‘communications’, as both of these departments already existed with in MIT. When Negroponte suggested ‘Media’, the response was “you mean like newspapers and magazines? Yuck, all yours”. 

In the early 1980s, it seemed like a pretty simple question. And today, for most people who define what they do with the word ‘media’, it remains a fairly straightforward definition.

If you work in ‘the media’ you are loosely connected to the production of entertainment or news, transmitted through the mass media

If you work in PR, you deal with elements of these media. As well as an increasing number of other media

If you work in media planning you buy space in these media. As well as doing an increasing number of things in other media

If you are a person connected to other people on the internet, then by communicating  with them you leave a trail of media, and increasingly you may choose to think about how these public media define your projection of yourself to the world.

If you are a teacher, media are the building blocks of learning, vehicles for knowledge.

If you are a biologist, media are what you put use to grow cells on.

If you are an artist, you create things out of media. Media are raw materials for art

Michael Faraday discovered the principles of electromagnetism. His discoveries were of little practical use at the time, as they were just theories. He gave them practical application by building the first electric motor. He intermediated them in metal and wood. This gave people access to his work without having to understand his theories: the metal and wood were media for his knowledge.

In the same sense, we can look at toothpaste as a medium, giving us access to the knowledge of the dental health benefits of the synthesis of Floride and Triclosan without having any personal background in either dentristry or biochemistry.

MIT economist Professor Cesar Hidalgo, from whom i am borrowing these analogies, sees media as the result of a world in which the amount of knowledge required to perform economic tasks is greater than the amount of knowledge able to be held by one person – a Peoplebyte world populated by Personbyte capacity. As a result, we need to collaborate to perform meaningful actions. As digital communication is persistant (ie once created it takes effort to destroy), this increases the overall quantity of communication media at an order of magnitude of the overall rate of collaboration

So ‘media’ in their purest form are simply methods of access – whether access to knowledge, to collaboration, or to cell cultures that grow faster than in nature. So can those of us in the communications industry helpfully refer to ‘communications media’ as a definable thing? Are communications media the connections between people? Does the increasing persistance of communications media mean that we are approaching some sort of infinite media singularity? An infinite quantity of media by definition suggests that media have no monetary value.

Connections between people help us to think about how to attribute value. The resource that all media consume is attention, and attention is finite. To make value judgements about the creation of media, we need to understand more about people and the interactions between them. To apply these value judgements to commercial ventures we need a reliable map to navigate human interaction. And human interaction is a subject that the twentieth century got badly wrong.

This is a successor to what i used to write over at GeekMedia. I started writing that because it helped me practice thinking about the industry i work in, and to come to some conclusions that i was able to talk about more confidently than if i hadn’t had the practice.

Planning for Networks is an approach to thinking about communications based on how people actually make decisions. It is inspired by the work of marketing scientists like Byron Sharp at the Ehrenburg Bass Institute, economists like Cesar Hidalgo and Paul Ormerod, pioneering network scientists like Duncan Watts, Nicholas Christakis and Albert-Laszlo Barabasi, anthropologists like Alex Bentley, Mark Earls and Robin Dunbar, and psychologists like Daniel Kahnemann.

It is a work in progress, and so many of the posts here are likely to be thinking out loud.