Alex Steer

369 words | ~2 min

I spend a lot of time listening to pitches by media companies and tech companies. These pitches are directed at marketers, and they normally go a bit like this:

1. Consumers are spending more and more time online.
2. Media spend is following them.
3. Consumers are demanding more and more from brands, faster and faster.
4. And marketers are under more and more pressure to demonstrate ROI.
5. You need a technology platform that lets you build rich experiences fast - and measure the results.
6. Our platform lets you build fabulous interactive brand experiences.

So what's wrong with this picture? It sounds great, but there's a big unanswered question.

Who controls it?

Before you sign on the line, bear in mind that when a lot of these companies put your digital campaigns or products on their platforms, they own the data that comes out of those platforms. That often means you can only get access to standard reports, containing the metrics the platform providers want to give you. And, of course, if you're running multiple campaigns on different platforms, you end up with ten different reports, using ten different sets of metrics, that you can't necessarily compare. Not to mention the fact that the platform owners can change the rules of the game and really stitch you up.

I'm pretty sure the arguments new media companies are using to pitch their platforms are the same as were used to pitch TV in the 50s. And we know what happened there: agencies selling clients a hell of a lot of TV work because they were being paid a fixed share of the media billings - rather than the channel mix that will get the best results.

It's that, but far more fragmented these days. Many platforms, many tech standards, lots of isolated data pools, and a lot of competing sales pitches. So before you rush into a platform tie-up, make sure you're not giving your data away or locking yourself into a world of proprietary pain. And makes sure the people doing your data strategy and digital comms planning aren't making money selling certain platforms or media packages.

898 words | ~4 min

This is Gareth Kay's recent presentation to the ICA, and it's very good.

Five things in it that I loved:

1. 'Low degree of difficulty'
2. 'Muscle memory'
3. 'Ingenious and effective'
4. 'People at the heart'
5. '100 small fires'

Read it and you'll understand.

Now, I love this because I work in part of the industry that has a reputation as the enemy of creativity. The part that helps clients use hard data to make marketing decisions. Wherever that line is crossed between maths and magic - whether it's copy testing or targeting or decision analytics - you'll hear the same grumble. That the suits are taking over, that logic is killing creativity, and that it leads to what Jim Collins and John Hegarty brilliantly call 'wind-tunnel marketing' - where everything looks the same. When Douglas Bowman quit Google in 2009, he hit out hard at its culture of testing.

When a company is filled with engineers, it turns to engineering to solve problems. Reduce each decision to a simple logic problem. Remove all subjectivity and just look at the data. Data in your favor? Ok, launch it. Data shows negative effects? Back to the drawing board. And that data eventually becomes a crutch for every decision, paralyzing the company and preventing it from making any daring design decisions.

So, okay, here I am, writing from the dark side. The side that wants to take all your great ideas and put them through the maths-mangle. Is there anything we can add to the effort to bring back some radicalism in advertising? I think so. Here are my three tough lessons from the marketing technologists to the ad agencies. In the spirit of radicalism, they're pretty blunt.

1. Grow up

If you work in advertising, you work in marketing. While your job's to challenge lazy thinking and inject human truths, you need to be able to see the world from your client's point of view. And the reality is that clients are under more and more pressure to deliver growth, with fewer and fewer resources. If you work in a developed market, chances are you're going to be generating that growth by stealing share for the next few years, not from growing your market. We're talking small, grinding, hard-fought gains. Your clients are going to want to measure the impact of every little thing they're doing so they can keep their jobs and their budgets, and lots of companies are stepping up to help them. The way things are, your clients are easily swayed by arguments made on the basis of numbers - even bad ones. So you can complain about measurement paralysis, and be ignored. Or you can start advising your clients on identifying the right opportunities and finding the right measures rather than just the easy ones, and be the agency that keeps doing daring, different work (that works) when everyone else is churning out the same timid rubbish.

2. Know your enemy

I've said this before on this blog, but here goes. You need to advise your clients when they're picking their data analysts and marketing tech companies. On this side of the logic/magic line, we're as nervous as you are about surging demand for 'big data', because we know the scene's going to get packed out with charlatans. So make sure you're there, helping out evaluating proposals and meeting the analysts and tech strategists. Make sure they're people who get marketing, who have experience agency-side or client-side as well as in IT. Make sure they're people you can work with, who care about using data to help clients be creatively brilliant.

3. Streamline the right way

Advertising research done well doesn't kill creativity - it helps justify its breathing space. That's true of all the new tricks digital marketing's got at its disposal too - data joins, rapid segmentation, A/B testing, viral reach metrics, the works. The stuff that saves marketers money saves advertisers time by quickly getting to problems that need solving. And those tend to be the problems that connect most closely to the sort of truths that lead to really radical advertising.

The right problem doesn't just mean the most specific problem. Yes, a lot of new tools let you split problems into their smallest possible moving parts. Used unwisely that leads to pedantry. But used well it can help you get to work faster, run tests in parallel, light your 100 small fires quicker. So find people who are going to use it wisely - and who care about keeping a broad perspective and hunting for the less obvious solutions.

We're here. We count stuff. We're helpful. We're argumentative. We're nice. We're radical. Now let's get on with it.

465 words | ~2 min

The Atlantic asks whether Facebook is making us lonely. About two-thirds of the way through an astonishingly long article it concludes that, no, it isn't.

Loneliness is certainly not something that Facebook or Twitter or any of the lesser forms of social media is doing to us. We are doing it to ourselves. Casting technology as some vague, impersonal spirit of history forcing our actions is a weak excuse. We make decisions about how we use our machines, not the other way around. Every time I shop at my local grocery store, I am faced with a choice. I can buy my groceries from a human being or from a machine. I always, without exception, choose the machine.

Putting aside this horrible example of the Paragraph 19 Problem, it seems there's no end to demand for these 'is social media making us lonely/stupid/evil/dead/etc?' op-eds - from Sherry Turkle's Bowling Alone (mentioned here) to Nicholas Carr's The Shallows to the, erm, unique work of Baroness Susan Greenfield.

The logic normally comes down to one of those syllogistic fallacies that we use to teach kids about logical error.

1. We are lonelier than we were twenty years ago.
2. We use online social networks now and we didn't twenty years ago.
3. Social networks are making us lonely.

Except of course it's never put so confidently (unless you're Nicholas Carr). It's always phrased as a 'may be', or a threatening 'could be', or a question ('is it?'). Like when expert forecasters tell us confidently that a policy 'may lead to social collapse' (which isn't much of a prediction, now is it?).

All of which is a shame, because this is very simple.

Online networks are a form of social tool: they enable certain forms of social interactions between people - just like telephones, writing, marketplaces, wheels and smoke signals - and they inhibit others.

We create the social tools we want. And they, to some extent, create us - or at least confirm our social interactions. If we are lonely, it is as a side-effect of various social interaction choices that have also led to the appearance of online social networks in the forms they exist.

Loneliness is not born of social networks, any more than people are born from apes, or English is descended from German. They share common ancestors. The social tools will change (slowly, grudgingly) as society changes, and social change is neither fast nor easy. Pretending otherwise is just headline-grabbing.

793 words | ~4 min

I don't find the Guardian's argument that Amazon aren't destroying publishing, they're reshaping it wholly convincing. Or, rather, I think they are reshaping the industry, but not for the reasons given.

They argue that there's a power-shift happening from publishing houses to retailing gatekeepers:

So Amazon, Google and Apple are gatekeepers. They have realised that the key to profitability is not investing in risky startups, but owning the marketplace where those startups stand and (mostly) fall. So for the rest of us, outside the walls, what's the point of taking sides in a battle between "legacy" gatekeepers and new ones?

Well, yes and no. But as the article points out, there's scarcely a battle at all - except the battle that's always raged between producers and retailers. Amazon, like Apple, have switched from a 'wholesale' model of ebook retailing (where they have full control over pricing) to an 'agency' model where prices are set by the publishers. (This is explained better in this Time article.) So this scarcely counts as a disruption at all.

For me, the real disruption is the shift of the delivery technology for books, from the hands of publishers to the hands of retailers. The production of books has always lain with publishers - but now retailers (Amazon, Apple, Barnes & Noble) have control of the major ebook devices. Imagine that we only had paper books, and that suddenly it was bookshops doing the printing, not publishers, and you see what a shift that is.

This is disruptive because retailers, unlike publishers, benefit hugely from long-tail effects. As long as the production cost per unit remains low (and with ebooks, even more than paper, it does), it's in the interests of retailers to make it easy for people to produce and sell as many ebooks as possible. Even the really bad ones that only sell a handful of copies will contribute to the profit margin, because there are so many of them. By contrast, it's in a publisher's interest not to produce thousands of third-rate books, because publishers have high non-production overheads (editing, typesetting, publicity, etc.).

But why is this? It's to do with branding. Retailers are what you might call long-tail brands. As long as their catalogues are easy to search, retailers are thanked for carrying a lot of different products (paradox-of-choice arguments aside, for now). Publishers are short-tail brands: guarantors of quality who are expected not to churn out lots and lots of dross, but only to bring the best products to market. (Self-publishing firms are the exception to this rule.)

The reason retailers have invested so heavily in ebook devices is because it makes sense for them to give the many publishers of bad books a cheap and easy way to circumvent the publishing process. We sometimes call this 'democratisation', though that seems a slightly optimistic term for opening up the long-tail market and monetizing it. ('Whatever you produce, we'll try to sell' is the motto of the retailers.)

But retailers also profit hugely from the short tail - the few very good, very popular books - and these are controlled by the publishers. (This makes sense. If you write good, popular books, you want them to be well edited, well promoted, and distinguishable from all the dross on the market.) So having opened up the market to the long tail, retailers now have to make friends with the gatekeepers of the short tail too.

So it's not true that the old laws of publishing have been changed forever (they haven't), but nor is it true that this is just 'same old, same old'. Producers and retailers will always fight for share. They will have to come to a settlement: otherwise we'll see publishers being excluded from popular new platforms, or new platforms becoming less popular because they don't stock good books. This is not new - it happens everywhere from grocery stores to app stores.

What's new is the  shift in production control from publishers (paper) to retailers (devices). Publishers have let this happen; now they are clawing back some control over pricing, but only some. Retailers have annoyed publishers, and now some will suffer from being closed out of the short tail.

For me, the real disruption opportunity isn't on either side of this cat-fight, or in making unlikely claims about falling book prices. It's in helping people decide what they might enjoy reading, in a much-enlarged market where the current model - restricted supply through publishing houses, recommendations made through marketing and press reviews - is suddenly not wholly adequate.

1164 words | ~6 min

Last week we launched BrandEdge, Infosys's digital marketing platform developed in partnership with WPP. With it, we introduced the industry to Fabric, my company, which is handling the WPP side of things. It's been a busy, exciting few days, and now lots more people know what marketing technology is, and what it is that I do for a living.

So this seems like a good time to talk about what's happening in digital marketing at the moment - specifically, the land-rush around 'big data' (high-volume data analysis used for targeting, behavioural analytics and real-time decision-making), and the appearance of digital marketing platforms - web-based management system designed to help large organisations run digital campaigns more simply and effectively.

Not surprisingly, there's been a big push by the large IT and digital media companies, who are used to helping businesses use technology more efficiently, and have spotted the opportunity in marketing. IBM has a 'suite' of digital marketing tools; so do Accenture and Adobe; and Google and Facebook have been strengthening their offers to advertisers for several years. Digital marketing platforms tend to consist of some combination of the following:

• A data management platform that brings all your data together from different websites, devices, channels, markets, search, social media monitoring, etc.
• Analytics tools that let you manipulate the data to find insights and opportunities
• Workflow tools that let you share data, set alerts, etc.
• Asset management tools that let you store and share creative so you're not re-making what's already been made
• Delivery tools that help you create websites, campaigns, apps, etc.
• Campaign and media management tools that let you launch and monitor online campaigns

In theory this is all good news for marketers who want to be able to act faster and measure better. But this is a classic market for lemons. Since the buyers often know less than the sellers about the technology and analytics they're buying, how do they know which providers are good and which ones are charlatans?

So in the interests of making marketing technology a bit less of a black-box mystery for marketing directors, here are my three top questions you should ask anyone who's trying to sell you their platform.

1. Who owns the data?

This is vital. Most digital analytics involves dropping a piece of code (known as a tag) onto your sites, social channels, digital media, etc. But a lot of the time, that tag is a third-party tag: it's owned and operated by the provider, and the data it gathers is collected by the provider and sold back to you. If you've ever used Google Analytics or Facebook Insights, you're accessing third-party data. (Google and Facebook gather the data, then give it back to you.) The problem is that you never truly control your data. There are often limits to how you can download it, cut it, and join it up to other data. So you end up with more data, but often unable to bring it all together to tailor exactly the view of behaviour that you need.  So you've still got lots of different blocks of data (media, social, web, etc.) that don't speak the same language and can't be used together. Along with all that, there's a lot of pressure from lawmakers in Europe and the US to crack down on third-party data gathering.

What to ask for: a first-party data management platform and tagging service - a secure data store owned and operated on your behalf. This is very different from just buying data from a third party.

2. Where is data stored?

Having full access to your data isn't the only reason to want a first-party data management platform. You also need confidence that your data isn't going places it shouldn't. If you're a consumer business, you have a duty of care over the data you gather about your online customers. Do you know where it's going? Just as importantly, are you sure that your data providers aren't aggregating and selling your data to your competitors? This happens a lot.

What to ask for: your data management platform to be provided on a software-as-a-service basis, kept separate from data from your provider's other clients. And a contractual guarantee that data won't be aggregated and re-sold.

3. Who are you hiring?

This is my favourite, and the one most likely to make a lot of providers sweat nervously. Most tech companies hire like tech companies: they've got database experts, product developers, maths guys, and salespeople. This is great, of course, but start grilling some of these companies on marketing, and they come up short.

For my money, I'd want to know that my digital marketing firm gets marketing. And I mean really gets it. I don't just want some software consultants who have read about the Four Ps. I want people who can work with me and my agencies to understand business problems, formulate marketing and communications strategy and get things done.

What to ask for: treat your marketing technology company like they're an agency. Okay, don't put them through a hoop-jumping pitch process for the sake of it. (The good ones definitely shouldn't want to steal work from your creative agencies.) But look for a place that hires people with a solid marketing and media background - including strategists, designers and creatives. These people should be building products that marketers want to use, that make life easier; and they should be helping clients use data to understand their customers better, and act on opportunities faster.

Why this matters right now

Digital marketing companies are promising the moon at the moment. A lot of them can't deliver, because they don't get marketing. That's going to have some terrible consequences over the next few years. We'll see businesses who have invested in locked-down third-party data that they can't do much with; clunky user-unfriendly data dashboards that nobody ever bothers logging into; and insecure systems that lead to a privacy backlash from their customers. It's going to be messy. So ask for lots of control over your data. And look for smart people that you actually want to work with, not just ones who blind you with science or talk a good game.

So in the end, the best question to ask is: Can I work with this?

664 words | ~3 min

This is an interesting one.

[Roberts'] knowledge of the posts only becomes relevant if the court finds the activity of liking a Facebook page to be constitutionally protected. It is the court’s conclusion that merely "liking" a Facebook page is insufficient speech to merit constitutional protection. In cases where courts have found that constitutional speech protections extended to Facebook posts, actual statements existed within the record.

I don't agree that Facebook 'likes' aren't a form of 'speech' (at least where speech is defined as an expressive act). Nor does Ars Technica. But the point about 'insufficient speech' is interesting, because Facebook 'likes' suffer from various problems of expression.

Lots of users know the quandary you get into if you want to respond to a piece of ambiguous news. For example, if I post, 'Got burgled, but thank God my insurance covered it', or 'Turns out I need a new liver, thankfully the waiting list is short', is it appropriate to like that? There's a problem of 'insufficient speech' there. Likes on page posts appear to be binary - you like, or you don't.

In fact, likes have high pragmatic variance. As the giver or receiver of a like, you're as dependent on intepretation as you are on fact. If I tell a hilarious story and you like it, I infer that you found it funny. If I tell you about my struggles against adversity and you like it, I infer that it's the equivalent of a pat on the back, a little note of support. If I post a photograph of my dinner and you like it, I infer that you think it looks delicious. But if I post a photograph of a sunset or a close-up of a beetle, and you like it, I infer that you're giving props to my photography skills. The same is true of not liking. If you don't like my post, I normally don't infer that you dislike it - I'm likelier to assume that you just haven't seen it, though either is possible. So likes are binary in form but not in usage.

Likes of pages are a slightly different story. The range of options is more restricted. When you like a single post, you are making a speech-act (like) in response to a speech-act (post). When you like a page you are making a speech-act in response to a whole class of speech-acts: everything the owner of that page posts now and in future. In that sense it's a speech-act that's also a bet. You're betting that the future posts from that page will be in some way of interest.

But what kind of interest? I 'like' several Facebook pages - but not all of those likes are equal endorsements. There are pages for organisations I support; pages for clients I work for; pages I actually administer; and of course pages of clients' competitors. Though Facebook says I like these, from my point of view I am merely following them. I use liking for competitive intelligence.

So it's no surprise to me that some staffers 'liked' their boss's rival's Facebook page. I'm not saying this is what happened here - but it should be considered. The problem is not that likes constitute 'insufficient speech' to warrant protection. On that count, I entirely disagree. The problem is that likes are ambiguous speech. They provide an insufficient basis for interpretation.

In the old joke about the perils of binary speech, there's no good 'yes/no' answer to the question, 'Do you still beat your wife?' On Facebook, you either like something, or you don't. Until they introduce 'laugh', 'support' and 'stalk' buttons, we can't make out-of-context judgements about the meaning of a like. And even if we can, they are still speech-acts within the network, and merit protection.

371 words | ~2 min

On the subject of wallpaper advertising, Doug Nichol's short film Sunshine, about the experience of shooting two McDonald's adverts in China, makes for fairly bleak viewing. It's an interesting take on the tension between creativity and commerce.

I was struck by this line:

A movie can't just be a bunch of images strung together. But in advertising you can kind of get away with that.

And also by the sense you get from the film's subject, John Benet, that there's something rather accidental, or incidental, about advertising - that it's a kind of strange offshoot of consumer capitalism, disguising itself as a confused form of art. Benet seems at once a lot more cynical and a lot less realistic than a lot of people in the profession, which is necessary to reinforce the strangeness of what television advertising is, especially in markets where it's fairly new.

All that said, there's a risk of making advertising sound either too vacuous or too sinister. I think advertising's at its best when it's at its most self-aware, and when it treats its audiences as people who are smart enough to know they're being advertised to. The result tends to be better, less insulting work, that delivers better results and leaves the people making it happier than if we're all trying to pretend it's art. Our viewers are unlikely to be convinced.

314 words | ~2 min

M&S's new brand ad is out. Here it is.

Much as I quite like Gary Barlow's rendition of Here Comes The Sun (take aim, hipsters), I have a problem with this ad. And with ads like it, in fact - nothing particular against M&S as a brand, I just saw it and the thought occurred to me.

The ad is wallpaper. Nothing really happens in it that resembles a narrative; there's no joke that builds to a punchline; no puzzle that needs solving; no challenge to your thinking; nothing, really, except a bunch of people milling around in the sunshine, having a picnic, looking happy. I suppose in theory that's a plot, but it's a bit of a thin one.

If the thought here is that it's enough to create something evocative and beautiful that associates the brand with the idea of summer fun... well, I don't think it is enough. I ended up watching it, wondering what I was supposed to do with it. As a viewer, I've got no right of reply, and no reason to talk. All I can do is watch it, and go, 'Oh.'

By contrast, here are three ads that look and feel very similar, but that in some small way engage the brain.

With a slight puzzle (Why are they all hunting for chairs?) - KFC, 'Spare Chairs', 2011:

With some insight into how British people actually behave during summer - Bulmer's, Great British Summer (2008):

And, simply, with a plot - Sainsbury's, 'Bare Necessities', 2012:

1896 words | ~9 min

It's a hundred years since the Titanic sank. With the centenary, a lot of the questions about why the ship went down are being raised again. For me, the Titanic case is an important reminder about the relationship between risk and uncertainty, and how we react to extreme events.

Fates and flaws

Thomas Hardy, in his poem The Convergence of the Twain, saw the loss of the Titanic rather typically as an act of God or Fate against human vanity:

The Immanent Will that stirs and urges everything
Prepared a sinister mate
For her — so gaily great —
A Shape of Ice, for the time far and dissociate.

And as the smart ship grew
In stature, grace, and hue,<
In shadowy silent distance grew the Iceberg too.

Alien they seemed to be;
No mortal eye could see
The intimate welding of their later history,

Or sign that they were bent
By paths coincident
On being anon twin halves of one august event.

There is a touch of this, too, in the Spectator's reaction, from the 20th April 1912, which  it has re-published on its blog this week.

The destruction of the largest ship afloat on her maiden voyage, of a ship reputed to be unsinkable, of a ship followed everywhere with admiring thoughts as the last word in ingenuity, in luxury, and in the impressive accomplishments of science, brings to every thoughtful person a deep sense of powerlessness, of smallness, and humility. Even in these moments of crushing personal sorrow one is conscious — perhaps only to deepen the sorrow — of the overwhelming reverses of human confidence.

Happily, though, the article changes course fast, and shows an early public interest in the forensics of the disaster:

Although we do not know as we write these lines the details of the wreck, it is certain that the 'Titanic' struck an iceberg, and that there were not enough boats to take off more than about one in every three of the passengers. All the lifeboats carried by the 'Titanic' were picked up by the ‘Carpathia,' and the passengers in them — chiefly women and children — were saved. Although there was a slight swell there was no wind, and one cannot possibly escape from the conclusion that if there had been enough boats all the passengers might have been saved. Most people have learned with astonishment that it is possible for a ship like the ‘Titanic' to pass the Board of Trade tests with an insufficient number of boats.

I'd like to stick with the Spectator piece because it's so unusual. Looking back at archive material from the time, it's clear that in the days and weeks after the sinking the analysis of causes became increasingly sentimental. Yes, valuable criticisms like the one above, or the fault with the watertight bulkheads, were identified and raised. But we have also ended up with a long tradition of thought about the Titanic which tries to identify 'human confidence' or other moral flaws as the real reason the ship went down. From cowardly owners, to unwise claims about unsinkability, to a rash captain's decision to sail too close to the ice.

Here's how the Baltimore Morning Sun put the matter, just three days after the event:

Image from the Library of Virginia, used with thanks.

The Titanic sinking was seen as a tragedy, and thinking about tragic events in the early decades of the 20th century centred largely on the idea of the 'tragic flaw' in human character that inevitably leads to disaster. For a scholarly exposition of this, see for example Oxford Professor of Poetry A.C. Bradley's Shakespearean Tragedy, published in 1904. The Titanic seemed to have no shortage of Bradleian tragic figures. That same mode of thinking can be seen in James Cameron's 1997 film Titanic, where you can pretty much pick which combination of character flaws (by Captain Smith, Mr Ismay, Mr Andrews, etc.) led to the disaster.

What's important about this line of analysis is the uniqueness. We try to see what was uniquely wrong with the Titanic that led to its demise - so of course it's easy to look at the unique combination of people who were on it, or the unique marketing claims that were made about the unsinkability of the ship. With the centenary we see a lot of this 'unique tragic flaw' theory turning up again in new clothes - such as Francis Wilson's Wall Street Journal piece, 'How bad management helped sink the Titanic', which is really just a Bradleian tragic analysis dressed up in MBA-speak.

Burying the survivors

Here's the problem with the 'unique tragic flaw' way of looking at extreme events. You ignore all the ships that didn't sink in 1912.

With a grim irony, the cognitive error that leads us to focus on the Titanic is called 'survivorship bias'. I don't have figures to hand, but imagine that in 1911-12 a hundred large cruise ships set off across the Atlantic, all similarly built and provisioned. Only one of them hits an  iceberg, and sinks, and causes the deaths 1,500 people.

We tend to look at the one that sank and ask, 'What was different?' That's a natural question - but it leads us to attribute all the causes of the disaster to things that were different about Titanic. And that leads us to assume, without questioning the evidence, that all the causes of the sinking were aberrations - things about Titanic, or its crew, that were unique. Specifically, we end up with the idea that the Titanic took unique risks, and paid the price.

Gary Cooper puts this superbly in an article on the ship's captain, Edward Smith, on the BBC News site:

Though Smith was undoubtedly a forceful sailor who pushed his ships hard in conditions that may have daunted other captains, it is a fact of history that providing the weather was calm and clear - as it was that night - it was not unusual for any captain to sail ships into ice regions at speed and several captains from other shipping companies testified to this at the disaster inquiries. [my emphasis]

Our survivorship bias tells us that Titanic must have done something abnormal - from offending the fates to sailing too fast, too close to the ice. In other words, the problem was one of risk.

In fact, it's just as likely that the problem was not risk. To use Frank Knight's model, risk exists within a model where the potential probability of outcomes is known. If you and I flip a coin to decide who wins it, you know there is a 50% risk of loss. In scenarios where you can't calculate all the factors, we're not talking about risk - we're talking about uncertainty. In a highly uncertain situation, you can do all the things you normally do, and have a completely unpredictable outcome.

It seems the builders and owners and operators of the Titanic did the same as the builders, owners and operators of other ships. It's just that they happened to hit an iceberg. Of course, they had watertight bulkheads and lifeboats designed to cope with certain levels of damage. In other words, they treated icebergs and other events as calculable risks. Like the risk-management models that failed during the financial crisis, they assumed that the chances of a catastrophic event were low enough not to be worth insuring against. And in most cases, they were right - but in the Titanic's case, they were unprepared.

Futurists tell you that trying to calculate the risk from extreme events is pointless. Extreme events operate in the domain of uncertainty, and there's just not enough data (because they're so rare) to model the probability accurately. All you can do, in the end, is try to be resilient against the impact of those extreme events that you can imagine. And if you're a shipping company whose goal is to travel at high speed through the icy waters of the North Atlantic, you can definitely imagine a collision with an Iceberg. By getting hung up on likelihood, the White Star Line seem to have been blinded to possibility - the idea that a big enough Iceberg could destroy a whole ship, not just disable part of it. By this reckoning, the loss of the Titanic wasn't anything unique to do with the Titanic - it was a failure of resilience throughout the entire fleet, that just happened to make itself known one April night.

I'm going to end with a long quote from the 1912 Spectator piece. More than anything I've read on the subject from the last hundred years, this captures the difference between risk and uncertainty, the need for resilience, but also the inevitability of surprise.

The truth is, as we said the other day in writing about the Oceans, that every passenger ship puts to sea with the assumption that disaster will not happen. Far too much is taken for granted. The whole subject ought now to be threshed out and the proper provision of boats or rafts insisted on for ships of whatsoever size they may be. Even if seamen declare it impossible to manage more than an inadequate number of life boats, we cannot see why other means of life saving should not be perfected, such, for example, as having the top deck entirely detachable, so that in the last resort it could be floated off in sections, each of which would be a duly equipped raft. As for increasing the buoyancy of the ship herself we do not know whether experiments which are being tried with submarines would be possible with ships of great size. Perhaps not; the inflation of special floats in an emergency would probably not be possible on a large scale. But, however that may be, it is necessary seriously to meet the tragically established fact that all the bulkheads of the largest ship in the world may be so damaged, strained, or knocked away that she will sink just like a cheaply built tramp steamer.

973 words | ~5 min

In the first of these posts we covered mounting concern about behavioural tracking and targeting by digital marketers, and traced some of the history of the 'data panic' that people are experiencing now - the sense that their data is at large, at risk, and for sale. Now it's time to take the temperature of data panic today, see where it might lead, and what marketers need to do.

Short-term vs long-term risks

As I said in the previous post, the current state of data panic is an odd mix of the well-informed and the ill-informed. Let's try to tease out those two strands in order to see what might change, and what might stick, in the current mindset around tracking and targeting. I should say up-front, this requires a bit of honesty.

After all, a lot of the things people are worried about, well... they are a bit true, aren't they? The technology, media and advertising industries have carried on quite happily over the past few years collecting data about their customers' online behaviours, without much scrutiny, and without much sign of wanting to take the conversation public. And that's a problem, because when it does come to light and people get concerned about it, nothing says 'you got me' like an awkward silence.

It doesn't help that the technology and the mathematics involved in targeting are so opaque and often rather hard to explain. As soon as you get into a discussion about cookies or sessions or clustering or segmentation or regression analysis, you're in a sort of voodoo country of the mind that calls up the ghosts of high-speed algorithmic trading and  Long-Term Capital Management and whatever the hell they were doing at Enron. Terrifying, confusing, the-machines-are-taking-over stuff.

Obviously there's a lot of misreporting here, as well as more well-founded concern. I think it's likely that, if privacy stays on the public agenda, it will lose some of its ability to shock.  Specifically, I think it will split out again into two separate discussions: about the merits of tracking and targeting, and about security of personal data. The second of these will remain high-stakes: when data has the power to expose or embarrass us, it matters. Tracking and targeting will probably become more transactional, though, as consumers either accept it as it is, in exchange for increased relevance and free services, or seek to gear the system more in their favour, for example by warehousing their behavioural data and selling it to advertisers. (Though I tend to buy Doc Searls's argument, that the benefits of giving data away are greater than the sale value.) Finally, while I think concerns about 'filter bubble' effects are legitimate, I don't think they're mainstream enough yet to drive the agenda.

Even if the privacy discussion matures, though, it hasn't yet. Let's not kid ourselves: there is public concern about targeting, and there is legislative pressure in the markets where the use of targeting is most advanced.

Why you can't just ride this one out

It's easy for industry insiders to succumb to a kind of disaster myopia around changes to privacy legislation. We tend to see data collection as not only so valuable, but so undeniably useful and beneficial, that we forget what a shock it is to users who barely know what a cookie is, let alone the extent to which their behaviours are being recorded and trafficked between servers for a profit. It's also easy for us to forget that the instincts of policymakers are likely to resemble those of the voting public, rather than those of digital marketers.

So let's assume that, a few months down the line, the US and EU require users to opt in before third-party cookies can be set on their machines. At the moment, when asked 'Do you mind if marketers track your behaviour?', the answer's hardly likely to be yes. If marketers want to avoid losing all their tracking data overnight, there are a couple of things they should be doing now.

The first is to get their data back under their own control. A lot of behavioural tracking is done on behalf of marketers by third parties who install their tracking technology (typically script tags) on marketers' websites. So data collected on domain1.com (marketer's website) is trafficked over to servers owned by domain2.com (tracking provider), and possibly beyond. That makes it hard for marketers to keep track of where they are sending their customers' data, and leaves them open to charges of negligence, not to mention security risks. Since third-party cookies are likely to be hit hardest by new legislation, marketers need to take more responsibility for data management.

The second priority is more basic. Marketers need to start being honest about data - and to take privacy seriously. The case for data collection has to be made by marketers, not just technologists, and it has to be put - genuinely - in terms of benefit to users, not just value to businesses and advertisers. We need to show that a better understanding of people's real behaviours isn't just used to serve up ads like cards from a deck - but to design better products and services, and give your users more relevant information.

We haven't been good at this, and we need to get good at it, fast. Otherwise we'll quickly find ourselves left behind.