Thursday, November 17, 2011


Today, I'll be talking about People Will Talk on AirTalk with Larry Mantle, KPCC 89.3. It's an NPR station broadcasting to Pasadena, LA, Orange County and the Web. I'm scheduled to be on at around 11.30 a.m. Pacific Time.

Tuesday, November 15, 2011


I haven't even got copies of People Will Talk yet, and I've found a mistake.

In the last chapter, I talk about reputation on the large scale, between groups and countries. This involves a mention of work by Michael Tomz, an economist at Stanford, on the role of reputation in sovereign debt. Tomz's 2007 book Reputation and International Cooperation looks at several centuries of data from the bond markets, and shows how a country's past behaviour influences how the markets treat it.

All very interesting and relevant stuff. The problem is, I call Tomz 'David' in the text (the reference is correct). Bloody bloody bloody hell. A lesson in the power of embarrassment, something also discussed in the book.

My apologies to Professor Tomz.

Monday, November 14, 2011

People Will Talk in the NY Post

Yesterday's New York Post has an article extracted/adapted from People Will Talk.

It's mostly about gossip, the subject of two of the book's chapters. To go to the source for a lot of what's in the article, check out the work of Ronald Burt at U Chicago. He's found that one's reputation is a product of one's social network, and that the kind of reputation you get depends a lot on the structure of your network - particularly on how closed (cliquey, in which everyone talks to everyone else) or open it is.

In particular, I refer you to these two pdfs:

Gossip and reputation

Closure and Stability: Persistent Reputation and Enduring Relations among Bankers and Analysts

And his 2005 book Brokerage and Closure.

Steering clear of the magic kingdom

In his stylish and thought-provoking book Frozen Desire: An Enquiry into the Meaning of Money, James Buchan argues that one of money's defining properties is that it robs things of their essences by making them comparable.

Pre-money, each thing had a certain thinginess. It was what it was, and could not be otherwise. But once money comes along, a thing can be expressed in terms of its cash value, and so, instead of seeing it as uniquely itself, we see it as more or less valuable than something else, and it loses that essence.

(One reason that men give women chocolates and flowers, Buchan suggests, is that they are perishable, and so, because they cannot be converted back into money, they retain their thinginess, helping to keep money and romance apart.)

Similarly, one feature of social media is that, by making our social lives quantifiable in a way they weren't before, they make them more like money. One can count one's twitter followers and facebook friends, compare them to someone else's, feel inadequate, and change one's behaviour to try and catch up.

Of course, people have always courted popularity, but never before has it been so easy to measure and compare. And, as occurred with money, something is lost in the process.

Or, as Bonnie Stewart puts it: "The peer-to-peer relationality of social media is undermined by the kind of behavior that cultivates status over relationships." (Stewart is writing in Salon, or rather Salon has reproduced one of her blog posts.)

Stewart is discussing Klout, which aims to give you "a single tidy number meant to sum up your influence and engagement in the social media sphere". There have been several attempts in the past to create general online measures of social worth, often with reference to Whuffie, the reputation-based currency imagined by Cory Doctorow in his novel Down and Out in the Magic Kingdom.

In People Will Talk, I'm sceptical about them, for the same reason as Stewart is - because by reducing a narrative to a number, you lose information, but gain a precision and quantifiability that is actually spurious.

But it looks like Klout is taking off. It, Stewart writes: "has established itself as a major arbiter of influence in social media’s already driving hiring in social media spheres. Its algorithm is being taken up as a factual assessment."

One reason for this is probably that employers have no idea how to decide who to hire, and will clutch at any straw, such as credit scores, that seems to differentiate one candidate from another.

That may be bad for certain types of people trying to get certain types of job. On a self-interested note, I wouldn't be surprised if this type of information were also influencing hiring decisions in journalism.

Whether it means anything, I don't know. Is your Klout score a reflection of anything more than your Klout score? I guess the way to judge that is whether it tells you anything you don't already know, or can't deduce another way - simply by counting their friends/followers, for example. (Given that too many can be a bad thing, in terms of likability and credibility, it'd be interesting to know how Klout treats this.)

Klout, for example, hints that Justin Bieber is famous. I am, however, prepared to believe that these things can be useful, and will probably become more so (even though I don't really want to, because I value the thinginess of things, and doubt that I am the kind of person/writer who will prosper in that kind of world - I don't want to know my Klout, and not just because I suspect it's tiny).

More generally, what's pernicious about this type of quantification is that it infects everyone, not just those who play the game professionally. It encourages us to compare and commodify ourselves - again, not a phenomenon confined to online life, and not a new complaint about online society, but something that seems to be becoming easier to do and more widespread.

One of my favourite lines in a novel comes in Jonathan Franzen's the Corrections, when one character says to another, "what's life about?", and she replies "I don't know, but it's not about winning." And yet, we're being given ever more ways to treat life like a high-score table. And, because we are competitive animals, as well as cooperative ones, we lap it up.

In the process, social information becomes like money, and our social lives become monetized. Stewart again: "We are gradually being directed away from sociality and toward businesslike behaviors by the business interests that design and profit from the platforms we use."

Or, to quote Slavoj Zizek (I know!) in the LRB: "We are often told that privacy is disappearing, that the most intimate secrets are open to public probing. But the reality is the opposite: what is effectively disappearing is public space, with its attendant dignity."

Or to put it a third way, the difference between social interaction and social media is like the difference between a park and a theme park. In the latter, your experience is controlled to commercial ends. Theme parks can be fantastic, but (unlike the hero of Down and Out) I wouldn't want to live in one.

Wednesday, November 09, 2011

Itetem ta humgopak...Altogether now!...

If you read the paper on musical diversity within and between cultures in todays Proc R Soc, you will search in vain for any idea of what the music actually sounds like. So I had a google. And then I had a totally unrigorous ponder on what I found (I also looked in vain for any proper coverage of this work).

Tom Rzeszutek and his colleagues analyse a corpus of folk songs from '16 Austronesian-speaking aboriginal populations from Taiwan and the northern Philippines'. These include the Paiwan...

Here's a snippet of a rather nice Paiwan love song

What does the fact that this sits so easily on the western ear say about musical diversity - perhaps that the total range of musical possibilities is actually quite small, and that cultures explore musical space (saturate diversity, if you like) quite quickly? Does this also mean that there's a lot of convergent evolution in music, with the same tricks popping up often in different cultures?

Here's another...

Sounds pretty similar to me.

They also analyse the songs of the Ifugao...

What does the fact that the people behind this arrangement can slather their aboriginal song in a jumbo helping of (country and) western cheese, but still recognize it as something Ifugao, say about the features of music that are most important in placing it in a certain culture? And what does the fact that music hybridizes so easily say about the generation and meaning of musical diversity?

The authors quantify the songs using a technique of their own invention called 'CantoCore', based on the cantometrics approach devised in the 1960s by the mighty Alan Lomax. CantoCore 'codes 26 characters related to song structure, including rhythm, pitch, syllable, texture and form'. They than measure the distance between songs by adapting a technique used to measure genetic distance - borrowing the tools of phylogenetics.

When they do this, they find that there's much more musical variation within cultures than between them. Within culture variation accounts for a whopping 98% of the total, in fact.

The 2% difference between cultures is still statistically significant, though. After all, it's easy to distinguish Cuban music from gypsy music, from Indian music, and so on. I'd guess that some aspects of music must be more diagnostic than others to our ears. (I think that CantoCore only analyses songs, so out in the real world you can also throw in instrumentation as a means of recognition.)

What I want to know is if some cultures have more diverse music than others. That's true of genetic diversity, and it reflects our species' history.

So, just as Africans are most genetically diverse, and the rest of us are, give or take, a subset of them, is the same true of music? In Subsaharan Africa, cultural diversity mirrors biodiversity - that is, the places with the most species per square mile also have the highest density of different languages. So perhaps the same goes for tunes.

Monday, November 07, 2011

Eric B and Rakim explain social learning

OK, People Will Talk is creeping onto the shelves (and whatever the electronic equivalent of a shelf is) and we can begin the task of examining the book's themes through the medium of golden-age hiphop.

Finding a tune for Chapter 1 is easy. The chapter's called 'Follow the leader'. (Although, unlike one later chapter, it's not named specifically after the track in question. I did briefly wonder about naming every chapter after an old-school joint, but never really put in the effort to do so.) You can read this chapter for free.

What a fantastic record. Video hasn't aged so well, though, has it?

Social learning occurs when animals make a decision not by direct experience, trial and error, but by copying what others are doing. It's an important aspect of how many species decide where to eat (in, e.g., sticklebacks), where to nest (many birds), and who to mate with (guppies, humans).

Piggybacking on another like this saves the effort and risk of finding something out for yourself, and is an immensely powerful strategy for exploiting your environment. When different learning strategies are set against one another, social learners outcompete individual learners. (On the other hand, it can also lead to information cascades, if everyone copies everyone else blindly without ever looking at the raw data on which decisions are based.)

Animals don't copy willy nilly. Whether they copy depends on their own internal state: animals who know less, because they are young, for example, or because they failed last time, are more likely to copy others. (It makes perfect sense to be young and impressionable, in other words.) And who they copy depends on the models on offer. The mating and foraging choices of big fish are more likely to be copied, for example, because a big fish is old and experienced; the embodiment of its own good decisions.

Humans are the champion social learners. Where did you get your tastes, your politics, your language and accent, your religion, and so on? By sampling the options and picking the one you liked best? Course not, it was by copying your family and friends. Social learning lies at the heart of culture and technology: it allows us to pass things on, and means that instead of reinventing the wheel, you can invent the bicycle.

The ability to observe and copy others is also one of the pillars of reputation. It's not exactly reputation, at least as far as people are concerned: copying someone is not the same as observing them and using that to plan how you treat them. Social learning is, though, how things get reputation - it's what happens when you choose a restaurant or a movie based on someone else's recommendation. It's how brands get built and broken.

Unlike other species, humans also give something back to those from whom they socially learn. Prestige, Joseph Henrich and Francisco Gil-White argue, is the tribute that the copiers pay to the copied. By honouring those who are excellent at something, the rest of us can get up close to them and hope to learn from them.

The evolution of prestige was something new in the world. Previously, displays had been largely repulsive - designed to show dominance and deter challenges. But prestige is an attractive force, which wins people over rather than putting them off. (Prestige and dominance are the equivalents of soft power and hard power in international relations.)

Chapter 2 is about costly signalling - the idea that conspicuous displays of virtue are also displays of intellectual, material or physical wealth. It's hard to imagine that there might be any hiphop records that touch on the subject of showing off, but I'll try and ferret one out. If you've an interest in evolutionary psychology, behavioural economics, fat basslines and lyrical flow, please feel free to make a suggestion. Late 80s/early 90s preferred, but I'm not going to be too dogmatic about this.

Wednesday, November 02, 2011

Do not trust this post

We've always judged each other by the company we keep, but now, when you meet a stranger via Facebook or Twitter, you can see their social network laid out all at once —how many people they are connected to, and who. And that can form part of the information you use to work out whether you like them, or trust them.

It's like PageRank for people. Unlike websites, however, it seems that people can be too well-connected. In a study published this week, David Westerman and his colleagues showed students mocked up Twitter accounts reporting information on swine flu. They found that tweeters with an intermediate number of followers — 7,000, rather than 70 or 70,000 — were judged as most likely to know what they were talking about and to be telling the truth.

A similar study from 2008 found that strangers on Facebook were ranked as most likeable—and even physically attractive—if they had a middle-ranking number of friends. The sweet spot was around 300, rather than a few tens, or a thousand.

Too few connections, it seems, shows that someone doesn't inspire trust or affection. But too many is the mark of a hoarder, someone who spends their effort on boosting their numbers, not on genuine interaction. It's not hard to estimate from a person's friend or follower count how they spend their online social time, and how much attention you'd get if you joined that person's circle.

Westerman's Twitter study also found that tweeters were rated as most competent when there was a narrow gap between the number of people following them and the number of people they followed. This, the authors suggest, may be a cue that the person behind the account is sociable, inquisitive and a good reciprocator.

Results such as these show how people are adapting to online life — no surprise, given that our species has managed to adapt to every other physical and social environment on the planet. True, the effect of network size on likeability and credibility is pretty small — it's hard to argue that counting someone's followers is a better measure of their credibility than reading their tweets. but then, people have only just begun working out how to parse this information. And as it becomes more common for people's first contacts to be via online social networks, and as networking sites become places people go to for breaking news, this kind of information may become more important.

These findings also hint at what online society might be doing to the size of our social groups. There are various measures and predictions of this, the most famous being 150— Dunbar's number, based on plotting the size of the neocortex against social group size in humans and other primates. (Going by that, a Facebook friend is worth half a real-world friend, which sounds plausible.)

Robin Dunbar's idea (I'm writing off the top of my head here) is that primates without language have to maintain social bonds by grooming. We do so with language, which is less time- and energy- intensive, and can involve more than two people at once. So we can live in larger groups.

But even talking only allows you to connect with so many people at once. In the flesh, large conversational groups fragment naturally, because it's easier than shouting or trying to get a word in edgeways. But online, physical space is not limited, and there's not the same problem with everyone trying to speak at once. It's also much easier, and less awkward, to eavesdrop on an interesting conversation without joining in.

All this should make it possible for people to maintain larger social groups, albeit with weaker connections. On the other hand, a recent study by Dunbar and his colleagues found that people who spent lots of time on social network sites did not have larger offline networks, or closer offline relationships. So maybe what we're seeing is people developing online and offline social worlds that, to a degree, serve different purposes.

These kind of findings also suggests that as norms about online social behaviour develop the upper limit of people we connect to online could be set, not just by the physical demands of keeping in touch, but by the social demands placed on us by the judgment of our peers.

However, my pathetic Twitter numbers (54 followers, 21 following at time of writing) shows that I'm the last person you should trust on this.