Tim Harford's Blog, page 134

Beware online confessions

Why do we reveal so much about ourselves on the web, especially since we also claim to be worried about privacy?

In the fledgling days of Facebook, when it had just a few thousand users and its founder, Mark Zuckerberg, was a student, he boasted to a friend over an instant message service about all the personal information he'd acquired.

"what!? how'd you manage that one?" was the response.

In reply, Zuckerberg typed: "people just submitted it / i don't know why / they "trust me" / dumb f[****]". (It is ironic that a private conversation about privacy was leaked, to the Silicon Alley Insider website.)

In fairness to Facebook, these are different times and it is a different organisation. And yet the question Mr Zuckerberg's friend asked remains fascinating. Why do we reveal so much about ourselves online, especially since we also claim to be worried about privacy?

Three behavioural researchers, Leslie John of Harvard Business School and Alessandro Acquisti and George Loewenstein of Carnegie Mellon University, have been trying to figure out the answer.

John and her colleagues wanted to explore the kind of contextual cues that persuade us to bare our souls. In a series of experiments, they used online surveys to ask intrusive questions, exploring what sort of cues might provoke revealing answers.

In a survey posted on a New York Times blog, billed as "Test Your Ethics", almost a thousand respondents answered questions about whether they had cheated on their taxes, omitted to tell a partner about a sexually transmitted disease, or filed a false insurance claim. For some participants, the questions were asked directly – "have you ever done this?" with a yes/no answer. A follow-up question then asked them to rate the morality of the practice. But for other participants, the questions were indirect: asking people "if you have EVER done this behaviour, how unethical do you think it was" and similarly, "if you have NEVER done this behaviour…." The questions demand the same information, but more people admitted to the dubious behaviour when the questions were asked indirectly. Intriguingly, 88 per cent of respondents also supplied their email address.

Later experiments tested the kind of web design that would provoke candour. Students were invited to answer questions on a laptop, with different students being exposed to identical questions underneath three different headers. One looked very official, with the Carnegie Mellon crest on the top, entitled "Carnegie Mellon University Executive Council Survey on Ethical Behaviors". Another was neutral. The third asked, "How BAD are U?", with a perky red font and a logo showing a little devil.

Objectively, it would not seem very smart to divulge information to the little devil site. "There is a wealth of research showing it is more dangerous to divulge on these websites," says Ms John. Yet that is what happened. "I find the result quite alarming," she told me."

There are two obvious alternative interpretations of the results. One is that students, who were from CMU, didn't care to confess their sins to an official-looking site. (HM Revenue and Customs may be trustworthy, but they are the last people I'd tell if I was cheating on my tax return.) Yet the neutral website drew similar results to the official-looking one, which suggests that the results were not down to a fear of punishment. Nor is there any indication that the little devil website encouraged students to think it was cool to have unprotected sex or take drugs (the researchers checked this). Instead, it somehow persuaded them to banish privacy concerns from their minds.

Online, there are many more influences guiding our disclosure than the simple appearance of a site. But this research is an important reminder that we are easily nudged into revealing more than, on careful reflection, we might wish to.

Also published at ft.com.

Taming the patent troll

Could a market for intellectual property ever work smoothly?

Trolls, as any three-year-old can tell you, lurk underneath structures they didn't build and pop up at unexpected moments to get in the way and make unreasonable demands. And "patent trolls"? They do much the same thing, nosing around patent systems and buying up intellectual property, often of questionable quality, and using it to extort money from genuinely innovative companies by threatening protracted and expensive legal action.

I am sympathetic to the general point that many patents, and their potential for abuse, actively discourage innovation. But if we're to solve the problem, it's worth pinpointing where it lies – and the rise of the trolls is a symptom, not the cause.

The three pillars that enable patent trolling are: the existence of absurd patents; the forbidding cost of the legal process; and the business model of buying up patents as assets in their own right, rather than building blocks for innovation. National Public Radio's This American Life recently discussed all three elements but focused on the last: the story is more compelling with a bad guy, after all. One contributor even compared patent trolls to a mafia collecting protection money.

But I'd look first at the patents themselves, many of which are for ideas that seem vague, broad and unoriginal. There are jokes, such as the "swinging on a swing" (US patent 6368227). But more serious are patents on, for instance, running an auction over the internet (US patent 7702540) that could surely have been dreamed up without the incentive of an intellectual property deed. The laxity of the US patent system is compounded by its sympathy for applicants: they seem to get the benefit of the doubt until a costly legal process says otherwise.

If patents were sensibly scrutinised and legal proceedings cheaper (big "ifs", it's true) there would be nothing wrong with buying up patents from the original inventors and then trying to collect licence fees for their use.

Could a market for intellectual property ever work smoothly? This is the dream of, among other people, Nathan Myhrvold, former chief technology officer of Microsoft. Myhrvold runs Intellectual Ventures, a company which is either the world's worst patent troll or a leader in creating a market for inventions, depending on who you listen to. Myhrvold, with law professor Mark Lemley, has argued that when patents are licensed, the details of the deal should be a matter of public record. This would help provide the kind of transparent information to get a market operating, one which would grease the wheels of innovation rather than rusting up the gears.

But would this be enough? The economists Joshua Gans and Scott Stern argue that there are formidable obstacles in the way of setting up a well-functioning "market for ideas". The basic problem is that patentable ideas are supposed to be unique, and ideas are typically only useful as part of an accumulation of other ideas. As a result, negotiations over patents are vulnerable to "hold-up" as various intellectual assets are acquired. As an analogy, imagine trying to buy land to build a railway line: each property owner has the incentive to hold the entire deal hostage.

So what is the solution? It might look a lot like a patent troll: a company with a large portfolio of patents it doesn't itself use. It would put together package deals. With a reputation to uphold, the patent troll would have an incentive to make things run smoothly, rather than behaving like a cranky old man with a house on the planned railway track.

Is this just a children's fairy tale? It is for now. But with a more sensible patent system, even the trolls may become useful members of society.

Also published at ft.com.

In praise of the nowcast

Technology is making it possible to collect and disseminate – in real time – information flowing through computers

Switch on any business TV channel and you'll be bombarded with scrolling quotes and jagged graphs. It's all very well if you like to watch that kind of thing, but the flow of financial data can lull us into an illusion: that we understand what is happening in the economy at the moment that it happens. We don't.

Take those share prices: each one is a miniature forecast of future profits for the company in question. They may or may not be good forecasts, but what they are not is a measure of real economic activity today: the price of shares in BP reflects today's supply and demand for shares in BP, not today's supply and demand for petrol.

Away from the financial markets things are even more obscure. We have indices of house prices, measuring how they have changed over the past month, quarter or year. But the houses bought and sold last month are not the same as the houses bought and sold the month before: are they more expensive because housing itself is more expensive, or because fewer studio flats and more penthouses were sold?

As for measures of retail sales or unemployment – or, above all, gross domestic product, the summary measure of economic activity – this all arrives with a delay, sometimes of years. We must judge conditions today by looking in the rear-view mirror. Regardless of what Bloomberg TV may tell us, much of today's economic conditions will only be understood in 2012 or 2013.

All this may now be changing, because it is now possible to gather, in real time, information flowing through computers. For instance, a curious piece of research was published last year by three computer scientists, Johan Bollen, Huina Mao and Xiao-Jun Zeng. The paper, "Twitter mood predicts the stock market", has a self-explanatory title. The researchers searched Twitter for expressions of mood such as "I'm feeling happy", and discovered that expressions of calm predicted movements in the Dow a few days in advance.

If this effect is real, it will almost certainly be exploited by automated trading strategies. But other computer-based indicators may prove more useful and more lasting, because they are tied not to mood but to actual activity.

The most obvious, perhaps, is the information that we give Google every time we make a search. Much of what Google knows about us as individuals – based on our use of Gmail, or the physical locations from which we access the search engine – is information that it will keep to itself. But the search engine giant does release data on search volumes: how many people in the UK are searching for "estate agents", for instance, or for "jobs", "JSA" (Jobseeker's Allowance) or "unemployed"? Two analysts at the Bank of England, Nick McLaren and Rachana Shanbhogue, have been examining whether the volume of these search terms tells us anything useful about economic activity. It seems that it does. For example, searches for "JSA" are correlated with the official unemployment data gathered by the Labour Force Survey – but are available a month earlier. For the Bank's Monetary Policy Committee this is a useful peek at the state of the economy at the moment that it makes its decisions.

The most striking research of all, for my money, was published by Nikolaos Askitas and Klaus Zimmermann as an IZA working paper in February. Askitas and Zimmermann used data from a new system designed to collect road tolls: most heavy vehicles in Germany have GPS-based technology that tracks their movements and levies the right toll. Unsurprisingly, their "toll index" provides a very effective "nowcast" of German industrial production. Perhaps we can look forward to the day when real economic data are collected and disseminated almost as easily as the charts on business television.

Also published at ft.com.

Dubious data cut down to size

A doctoral student found that national income is correlated with the average length of the erect penis in a country

There's no doubting the smash-hit economics research paper of the summer. It is "Male Organ and Economic Growth: Does Size Matter?", a working paper written by Tatu Westling, a doctoral student at the University of Helsinki.

Let's get the conclusions of the article over with first, because there is a serious lesson, somewhere. Westling discovers that national income is correlated with the average length of the erect penis in a country. Specifically, medium-sized erections are correlated with the highest national income (in 1985, as it happens) and small erections are associated with economic growth between 1960 and 1985.

To reach his conclusions, Westling performed a statistical technique that is known in the trade as a "cross-country growth regression". Beloved of (some) development economists over the past couple of decades, it has been made possible partly by computers. But its deployment depends mostly on the careful cultivation of data sets describing GDP, political system, educational attainment and other variables, for large numbers of countries over many years. You throw the numbers into the computer and see which features of a country's economy are correlated with economic growth. Westling simply added some rather unconventional data.

This strategy has produced many answers – rather too many, in fact. In 2002, the development economist Romain Wacziarg wrote, with a touch of acid: "the list of proposed panaceas for growth in per-capita income has included high rates of physical-capital investment … low fertility, being located far from the equator, a low incidence of tropical diseases, access to the sea … and suitably conditioned foreign aid." I have omitted 17 of Wacziarg's "panaceas", and he pointed out that the list is "growing and non-exhaustive". To it, we can now add erect penis length of 13.5cm – what Westling has demonstrated is "the GDP maximising size".

Westling claims that he uses standard statistical methods and that his results are robust; that the correlation is both highly statistically and economically significant – in other words, the result is big enough to matter (please – no giggling) and is unlikely to be a coincidence.

Well, well. What are we to make of this? I asked Westling how he would characterise his research paper, and he suggested the term "sardonic economics" – and, he added, "Scientifically, this paper is probably as worthless as much of contemporary economics."

Westling is a little too harsh, but only a little. Cross-country growth regressions have been used to demonstrate a vast number of statistical relationships, some mutually contradictory. (Particularly notorious is the vast and unfulfilling literature showing that foreign aid increases growth, or does not increase growth, or leads to growth under certain conditions.) Such regressions are not compelling, because the number of different statistical relationships that might be tested is truly vast, and the data themselves are sometimes sketchy. (Although few data sets are as of dubious a provenance as Westling's data on penile length, a fact he cheerfully acknowledges.)

Cross-country growth regressions have their uses, given a sense of perspective about how fragile their results can be, and how difficult it can be to turn a statistical correlation into a workable policy. But if you are wondering how much weight to place on a statistical technique that can produce a robust connection between the size of a penis and the size of an economy, you are right to do so.

Unless, of course, the apparently ludicrous relationship is real? I asked Westling about this. "One biologist suggested to me that testosterone might explain something," he responded. I doubt it.

Also published at ft.com.

Darkest Peru? It's a beacon for business

Between 2009 and 2010, Peru cut the cost of setting up a new business from $685 to $564, the number of separate bureaucratic procedures from nine to six, and the time the whole affair took from 41 to 27 days. It's an impressive (indeed, world-leading) improvement, and one which presumably will help Peruvian entrepreneurs create jobs, fulfil the desires of Peruvian citizens, and make money into the bargain.

These statistics come from the World Bank's Doing Business project, which since 2003 has been publishing data on the ease of doing business around the world, with indicators such as how long it takes to establish a business, how easy it is to buy and sell commercial property, or how time-consuming and expensive it is to clear customs. (I should declare an interest: in 2004 and 2005, I helped with some of the superficial details of the Doing Business project.)

Yet the Peruvian good news story feels oddly precise. The Doing Business methodology is a kind of time-and-motion study of the process of establishing a hypothetical business with some quite specific characteristics. Local experts, typically lawyers, examine the standardised case study and figure out all the papers that must be filed, departments that must be notified and official fees to be paid.

The focus on this case study allows the Doing Business project to produce specific, comparable information but the risk is that the experience of actual companies is rather different. The World Bank being a pluralistic place, it also gathers and publishes an alternative measure of the burden of regulation and bureaucracy on local businesses: surveys asking local businesses about the obstacles they face – bureaucratic or otherwise.

Mary Hallward-Driemeier of the World Bank and Lant Pritchett, an economics professor at Harvard's Kennedy school, have published a discussion paper comparing the results produced by the two methods. Up to a point, they are in accord: countries which look good on paper also perform well in the enterprise surveys – and the more streamlined the rules, the less trouble businesses claim to have.

And yet for most countries (those in the most bureaucratic two-thirds, according to Doing Business) the relationship breaks down: what the case studies show and what businesses actually report bear no resemblance to each other. And there is huge variation within countries: some companies seem to live blessed existences while others are all but under attack from officialdom. Pritchett quotes an Indian saying, "Show me the face and I'll show you the rule." In short, when the rules are bad enough (and they are pretty bad, in a large number of countries), the rules cease to be directly relevant.

"If your Doing Business number says it would take you 350 days unless you pay a bribe, you are in a country where nobody doesn't pay a bribe," says Pritchett.

Quite so, and it seems plausible that a researcher seeking some direct measure of the costs of silly regulations will do better to consult the enterprise surveys than Doing Business's hypothetical case study.

Yet the case study approach has great merits. Partly, it has rhetorical power – a fact appreciated by the Peruvian economist Hernando de Soto when he pioneered the time-and-motion approach to regulatory reform in the 1980s. Also, the case studies of Doing Business go beyond identifying that a problem exists and point to specific laws and regulations that are causing trouble and enabling bribery. For example, Doing Business has established that a cheap, politically uncontroversial way to streamline regulations and climb the World Bank's ranking is to establish an online one-stop shop for registering new businesses. Perhaps it is no coincidence that this is the direction Peru's reforms have taken.

Also published at ft.com.

In defence of PowerPoint

I am about to do something rash, which is to disagree with Lucy Kellaway. Last week, the fearless observer of business follies went too far: she called for PowerPoint to be banned.

The prosecution's argument is simple: many PowerPoint presentations are very bad. This is true but it hardly makes the case for a ban. Serviceable tools can produce awful results in the wrong hands, as anyone who has seen me put up shelves can attest. Banning the screwdriver is not the answer.

So it is with PowerPoint. It's an unromantic, practical piece of kit. It is often used poorly. It is not the most elegant tool, but botched jobs must be blamed on the workman. Many of the bad presentations people deliver with the help of PowerPoint would have been bad presentations in any case. Would it have been better to hear the impromptu ramblings of a nervous speaker in total cognitive meltdown? Or to watch a piece of professionally produced but irrelevant film, in the dark? Many readers will remember corporate life before PowerPoint. It was no lost Eden.

PowerPoint is not the world's most wonderful piece of software. The built-in templates have long been ugly, the clip-art tacky and the animations risible. As if determined to deliver on the name, it inserts bullet points into text with little provocation. It is harder than it should be simply to make all the letters line up. (I am still using PowerPoint 2003. By all means dismiss this column as the ranting of a corporate shill.)

Yet for all its flaws, PowerPoint performs two useful tasks well enough. It quickly allows one to compose speaking notes and to create slides showing images and graphs. The trouble starts when people confuse the two jobs.

There is nothing wrong with jotting down speaking notes as a memory aid. PowerPoint is as good a way of doing this as any, especially if you have handwriting like mine. For the vast majority of speakers, such speaking notes are preferable to the alternatives, including memorising, ad-libbing on the spot or writing the whole speech out and reading it in a wooden monotone.

The problem is that for some baffling reason, many speakers decide to project their speaking notes on to a wall rather than printing them out, postcard size, and sticking them on to 3×5 inch cards. I often sketch out my speeches with the help of PowerPoint. I just prefer to keep the slides to myself.

The second use of PowerPoint is to project visual aids on to a screen. This it does perfectly well – and the clichéd clip-art of yesteryear is now almost extinct. These days people "borrow" cartoons from Dilbert, or grab photos from the web. The effect is often pleasing enough.

It would be better if people learnt a bit about fonts, and better still if they learnt that by pressing "B" they could temporarily blank the screen. But one cannot have everything.

Lucy approvingly mentions a famous condemnation of PowerPoint by the brilliant information designer Edward Tufte. Professor Tufte attacks PowerPoint partly for its "relentless sequentiality, one damn slide after another" and partly for the asymmetric relationship between speaker and "followers".

This is odd because Tufte does not acknowledge that he is really assaulting the idea of public speaking itself. What could be more relentlessly sequential than a speech? One damn word in front of another. If you hate the very idea of a speech, fine. But say so.

It would take little to improve greatly the quality of most people's PowerPoint presentations – far less than it would take to improve the quality of corporate Newspeak. So why call for a ban?

The true problem is far more troubling. It is that in a corporate environment, we are asked to read prose by people who cannot write and watch performances given by people with neither the talent nor the training to perform. For some reason these amateurs are better paid than most writers and performers. There is something depressing about all this, but the blame cannot be pinned on PowerPoint.

I cannot finish without confronting the greatest sin in my version of PowerPoint: the "AutoContent" function, which sketches out a speech if you cannot do it yourself. AutoContent, The New Yorker once reported, was named as a joke, in "outright mockery of its target customers". The very idea of the function is pernicious indeed but the real horror is that it was created to satisfy a demand.

Fortunately, that demand may have worked itself out, too: AutoContent was discontinued in 2007.

First published in the Financial Times, 25 July 2011

What now for newspapers?

Competition alone cannot guarantee that readers get access to information they need

With the realisation that Rupert Murdoch's newspaper, the News of the World, had done some loathsome things, MPs realised that they had moved into a looking-glass world. When once it was politically dangerous not to bend the knee to Murdoch, suddenly it was politically dangerous not to stand up to him instead – and standing up to Murdoch is presumably a lot more fun, to boot.

But what now? The presumption seems to be that no future proprietor can ever be allowed to accumulate such a large market share. Newspaper and television markets must be kept competitive for the good of political and cultural discourse. This was the idea behind referring Murdoch's now-withdrawn bid for BSkyB to the Competition Commission.

As an economist, I'm always banging on about the benefits of competition – but in the case of media, they are less clear-cut than one might expect. Competition is usually a great process for giving customers what they want at a price they can afford. (Although let us not forget that key customers in the newspaper business are advertisers – and it was an advertisers' boycott that brought down the News of the World.) But in the modern-day newspaper market, price is not necessarily the key issue. Many consumers are getting their news for free from internet sources, and many advertisers have moved online, too. The question is not whether readers and advertisers are being exploited by monopolistic prices, but whether traditional newspapers can survive.

If price is not the issue, then what we are really concerned about is that newspaper readers get access to informative news about the key issues of the day – the kind of thing upon which our democracy depends. But it is foolish to expect this from competition alone. Sadly, we do not value information about how to vote nearly as much as information about which car to buy. We probably value shots of celebs in swimwear more than either.

True, competition can encourage certain kinds of quality journalism. For instance, the economists Jesse Shapiro and Matthew Gentzkow have pointed out that certain stories – of abuses by US soldiers at Abu Ghraib, for instance – are promoted by media pluralism. CBS had the story and sat on it at the request of the US government (the story was said to be dangerous to US hostages) before broadcasting after it became clear that the story would emerge in The New Yorker.

But competition also promotes gutter journalism and it probably promotes opinionated journalism, too. Fox News, Rupert Murdoch's hugely influential TV news channel, seems to have become popular by staking out ground as the source of right-wing rabble-rousing, and MSNBC has gained ground as it has moved to the left. Competitive markets give people what they want rather than what is good for them.

Gentzkow and Shapiro have studied this question in the US. Using an objective (if imperfect) measure of bias, they found that newspapers closely match the political biases of their potential readers, as measured by votes cast in the 2004 presidential election, and by the source of campaign contributions to each party. No doubt the causation runs both ways, but one striking result is that the proprietor's identity seems to make no difference to the bias. The media barons tell us what we wish to hear.

The most disturbing aspect of the phone-hacking scandal, it seems to me, is the reluctance of politicians to challenge Murdoch's empire, and in particular its cosy relationship with the police. If more competition dispels that sense of fear in future, it will be all to the good. But don't expect every journalist to suddenly start working on the next Watergate.

Also published at ft.com.

A handbag away from our debt ceiling

"Mum?"

"Yes, dear?"

"What do you and Dad keep arguing about late at night?"

"You should be asleep. When Daddy and I have our little chats, it's past your bedtime."

"Mum, I'm 15 – a bit old for bedtime. Anyway, I usually ignore your arguments but this one seems important."

"Well, it is important. We are debating whether to raise the family's debt ceiling."

"Debt ceiling?"

"It's the limit on what we can borrow."

"Doesn't the bank manager decide that?"

"Yes, ultimately Mr Foss-Smythe can refuse to lend money to us. But your father and I have a different system. For years and years, we've taken a solemn vow never to let our household borrowing exceed £500 without both of us explicitly agreeing to raise the limit."

"So you were arguing about whether to borrow more than £500?"

"Darling, we've already borrowed £500,000. Houses aren't cheap these days, even in Hackney. The point is that your father and I agreed each time we needed to raise the debt ceiling. But now we can't agree. Your father wants me to cut back on my spending, or he won't agree."

"So, what happens if he doesn't agree?"

"Then we have to ensure that our household debt doesn't go over £500,000."

"Fine. £500,000 is a lot of money. Dad has a point. You buy too many handbags."

"But darling, the handbags aren't the point here!"

"Aren't they?"

"Well, perhaps they are a bit. But even if I did stop buying handbags, we'd still go over the limit. Look – ever since I lost my job I've been looking for a new one. I'm sure it's just a matter of time. But until then our income is depleted."

"So – stop buying handbags, Mum!"

"It's not that easy. The percentage of household income spent on handbags has been considerably exaggerated by your weaselly father. Far more important is the mortgage. If we stop the payments, we lose the house."

"But you're not actually borrowing more money to pay the mortgage, are you?"

"Actually, we are. We're taking out loans and using credit cards to pay the mortgage. There are also the payments on the car, your school fees and the money we promised to send every month to Grandma.

"And then there's stuff like food and electricity – we haven't promised anybody that we'll spend that money but life would get pretty difficult if we didn't. Without my salary, our spending is inevitably higher than our income."

"Wow. So Dad wants us to do something really drastic?"

"Well – no, not really. He's just using the debt ceiling as a bargaining chip. We agreed all of this spending and set out the whole budget a few months ago."

"Was that before you lost your job?"

"No, that was after."

"Mum, I'm confused. You and Dad knew what your income was going to be and you agreed all the spending. So, basically, you agreed spending that would guarantee that you'd break through the debt ceiling. So you've agreed to exceed the debt ceiling."

"Yes – we agreed to exceed the debt ceiling. But we haven't agreed to raise it. That's the problem."

"But that doesn't make any sense!"

"It doesn't matter that it doesn't make any sense. The point is he has bargaining power now.

"The next mortgage payment is due on August 2; that's also the same day we send money to Grandma. If he hasn't specifically approved for us to go overdrawn on our joint account, we have to choose: Grandma doesn't get paid, or we go hungry, or we skip the mortgage payment."

"But you have the overdraft limit?"

"Oh yes. In fact we have the overdraft limit agreed with Mr Foss-Smythe. The same bank we have the mortgage with."

"And he's happy to lend you the money?"

"Delighted, darling."

"So he'll be happy if you borrow money from your current account to pay the mortgage, but not if you just skip the mortgage."

"Exactly. He will be very unhappy. I'm surprised he's not more nervous already."

"Perhaps he should be more nervous that you don't have a job and that you and Dad are spending far more than you earn, than about some entirely arbitrary debt ceiling that the two of you can agree to waive at any moment?"

"You may be right."

"Where is Dad, anyway?"

"I think he's buying himself a treat to relieve the stress."

Also published at ft.com.

TEDTalk – Trial, Error and the God Complex

Why social marketing doesn't work

We often overestimate the likelihood of success of viral hits.

In 1948 Harold Lasswell defined the objective of media communications research as discovering "who says what to whom in what channel with what effect". The difficulty for researchers has been that for the first half-century or so after Lasswell set out the aim, it has been largely impossible.

Perhaps that is now changing. Online social networks generate a huge amount of information about who says what to whom. Economists, computer scientists and sociologists are now digging through these social networks for the answers to long-standing questions – and few answers are as eagerly awaited as the secret of producing a sure-fire hit.

So how do you produce the perfect film or write the perfect book – or compose the perfect tweet on Twitter – in a way that will maximise the chances of catching on? Duncan Watts, a mathematical sociologist at Columbia University and Yahoo! Research, has answers – and I'm afraid they're not too encouraging. "I've been using social media to promote my book," he says, "and it's just a waste of time – it has almost no impact at all."

(I'll throw him a bone. His book is called Everything is Obvious (Once You Know the Answer).)

Part of the problem, perhaps, is that our expectations are skewed. If you ride on London buses, you may be astonished to discover that many of them are almost empty. The average London bus, according to the UK's Department for Environment, Food and Rural Affairs, contains only 17 passengers. Clearly most bus-riding people are travelling on the full ones.

It's a similar story with viral media: we notice the successes simply because they are successful, and overestimate the likelihood of success. And there's a survivor bias: in our analysis of what works we ignore what fails. "People think it's all about videos of cats or cute children," says Watts, "But there are millions of videos that have these attributes but haven't spread."

Watts and his colleagues, in a research paper titled Everyone's An Influencer, place numbers on a specific type of media hit called a "Twitter cascade". A Twitter cascade occurs when one person's "tweet" (short message) is repeated by other users ("retweeted"), whose retweets are themselves retweeted further, and so on. Anyone unfamiliar with Twitter can imagine a particularly good joke or piece of gossip spreading.

The first surprise, then, is that the typical Twitter cascade is both rare and tiny. Ninety per cent of tweets are never retweeted, and most of the remainder are retweeted only by a person's immediate followers, not by those at two or three removes.

The second surprise is that beyond the mind-numbingly obvious, it's impossible to predict which tweets will start cascades. Simply knowing that a user has started previous cascades tells Watts and his colleagues almost everything they can divine about the likelihood of future cascades – which is not very much. (It is not especially useful to know how many followers a user has if you know about their previous success in starting cascades, because the two pieces of data overlap.)

Duncan Watts would like to see marketing companies running properly controlled experiments to see which messages carry through social networks such as Facebook. But he's not convinced that they will. "When you do the experiment properly, all the numbers go down," he says. Watts believes that the likely outcome of such experiments would be to demonstrate how difficult it is for social marketing to have any impact.

I can now barely summon the will to beg you to follow me on Twitter (username: @timharford). But, there it is. Most things fail, and as Watts says, "the curse of being able to measure everything is that you get slapped in the face with this reality all the time."

Also published at ft.com.