Category Archives: friction

And I must agree

I’ve been thinking for a while – a phrase which here means “a couple of years” – about adding to my occasional series of book reviews. Something about why Light left me feeling simultaneously awestruck, existentially uprooted and in need of a wash; or how The Star Fraction brought me out in a mild case of conspiracy mania, but Descent didn’t do a thing; or how Jonathan Strange and Mr Norrell only really works if you read it as a secret history of the actual England; or the roles of psychosis, neurosis and therapy in the Frieda Klein books; or the artist’s eye and the problem of genius in The Maker of Swans; or what was actually going on in The Thing Itself (although I might need to re-read it first).

But I’ll start with an easier one: what’s wrong with Paula Hawkins’ phenomenally successful novel The Girl on the Train?

1. The girl on the train

First problem: there is no girl on a train. The main character commutes into London and consequently spends a fair bit of time on trains, and she makes an observation which is central to the main plot (or seems to be) while looking out of a train window; but that’s about it. The plot mutates part of the way through, but in neither of its forms is it actually about her. In any case, she isn’t a girl. Admittedly, different people draw the age boundary between ‘girl’ and ‘woman’ at different points – 16? 18? 21? – but it would be a very odd definition of ‘girl’ that included somebody who had (a) got a job (b) got married (c) bought a house (d) tried and failed to get pregnant (e) suffered from depression triggered by (d) (f) got divorced (g) moved out to live in a friend’s spare room (h) become an alcoholic (i) lost her job and (j) spent two years(!) concealing (h) and (i) from all around her. Not to mention that she describes herself as getting married seven years before the period of the novel and moving into her first house aged 26. Once I’d realised that I was reading Scenes observed by a woman on a train I started to wonder if an actual girl on a train was going to figure in the narrative later on (oh, that‘s what…), but no.

2. The narrative voice(s)

The book has three narrators, whose names are helpfully given at the start of each chapter. Saying that this sort of thing was done better by William Faulkner or Virginia Woolf would be a bit beside the point – their audiences were very different from Hawkins’ – but I will say that John Wain did it a lot better, and nobody even remembers John Wain these days.

Here are our three narrators:

My phone beeps. There’s a message on it, received hours ago. It’s Tom again. I don’t what to hear what he has to say, but I have to, I can’t ignore him.

No one comes. The lights are on, but no one comes. Perhaps he has seen me outside, lurking, perhaps he’s upstairs, just hoping that if he ignores me I’ll go away. I won’t.

I’m upstairs, in the bedroom. Tom’s watching TV with Evie. We’re not talking. It’s my fault. He walked in the door and I just went for him.

As Kevin Rowland might say, their internal monologues are various (various, yeah, various) but they’ve all got one thing in common: they all sound the same. Hawkins’ idea of character differentiation seems to consist of having them talk about different things. So voice 1 (who’s an alcoholic) talks about how delicious her first (or second) cool (or warm) drink tastes; voice 2 (who’s wild and impetuous) talks about feeling frustrated and cooped-up in the hell that is suburbia; and voice 3 (happy young mum) says things like “treated myself to a very cute Max Mara mini dress (Tom will forgive me once he sees me in it)”. But they all talk the same way. Like this, in short sentences. Then longer sentences, ungrammatical (mimicking speech patterns), sometimes creating a sense of urgency by going on just a bit too long. Then short ones again – I mean, obviously.

All this might be bearable if the characters themselves were… well, bearable, or even if they were annoying in interesting ways. (Read Some Effing Dickens! Or listen to the Archers, for that matter.) Even to say that the three narrators talk about different things does Hawkins too much credit; they all spend most of their time talking about the same thing – themselves. Internal monologue is a wonderful thing in the right hands; in the wrong ones it just gets you endless sentences beginning with “I” and “My”.

3. The present tense

As the three extracts above suggest, the book’s written in a breathless present tense, as if spoken by the three narrators. Not only is each chapter headed with the name of its narrator; each chapter – or rather each section within a chapter – also has a date, giving the effect of a series of diary entries. The chapters belonging to one of the narrators take place over a different, longer time period than those of the other two; just to make sure we don’t miss this, the section headed Thursday, 11 July 2013 is followed by a chapter headed with a different name and the words: One year earlier: Wednesday, 16 May 2012. This system makes it very easy for readers to check what’s taking place before what, and lets us know when the earlier timestream starts to catch up with the other two; there’s no other justification for it, though, which is to say that there’s no justification for it at all within the characters’ own reality. For instance, there’s never any suggestion that any one of the narrators is particularly obsessive about dates – or that any of them actually is keeping a diary. It isn’t even suggested that the narrators are reconstructing events after the fact – a plausible framing device that Hawkins could easily have helped herself to, had she not used the ‘diary’ format for all three of the narrators.

Perhaps the thinking was that endless present-tense narration would be disorientating if readers didn’t get regular reminders that time was in fact passing. If so, it would have been much better to ditch the commitment to use the present tense, which Hawkins does in fact abandon whenever it starts to be too much trouble. And it is trouble, writing in the continuous present; that’s why it’s so effective when it’s done well. (Read some effing Silverberg – “Passengers” for a start, and “Sundance” after that; a day or so after, perhaps.) Here are our three narrators again:

‘Did you see this woman on Saturday night?’ he asked. I stared at it for a long time. It seemed so surreal having her presented to me like that, the perfect blonde I’d watched, whose image I’d constructed and deconstructed in my head.

I turned up to my session late, and walked straight into his office without a word to the receptionist. He was sitting at his desk, writing something. He glanced up at me when I walked in, didn’t smile, then looked back down at his papers.

Tom rang me back – he was between meetings, he couldn’t come home. He tried to placate me, he made all the right noises, he told me it was probably a load of rubbish anyway.

Present-tense narration, indeed. Our narrators talk to themselves at great length and in what’s basically the same voice, beginning each day’s ruminations with a quick timestamp for ease of reference. Then later – when the plot starts to pick up – they tell themselves the story of what’s been happening to them and what they’ve been doing, in the past tense. Some events are still narrated as taking place in the present, but there doesn’t seem to be any reason for this, other than to make the present-tense sections seem more gripping. As with the diary dates, it’s a completely spurious device.

4. Women and men

All this wouldn’t be so bad if it was fun to listen in on our narrators chatting to themselves. Unfortunately, it’s tough spending time with these people. The main cast is tiny – three women (the narrators), three men. The women aren’t women, they’re airheads: a drunken, neurotic airhead, a scatty, hippyish airhead and a smug young-married airhead. The men are drawn in less detail, but they’re all alike: they’re all strong, silent and secretive, and they’re all devastatingly attractive. One’s a nice guy – although he’s also a therapist, so it may just be the job talking – and the other two are jealous, possessive bullies and thugs; devastatingly attractive bullies and thugs, mind you. The two women who aren’t drunk are pretty hot stuff themselves – one of them comes close to seducing the therapist, which is of course at the top of page one on the Therapist’s Big Book Of Things You Must Absolutely Not Do. But I guess he’s too weak to resist her, what with being a nice guy and not a thug… ugh. It’s like a Mills and Boon with the masochism turned up to 11.

Each narrator goes in for a lot of flappy will-I-won’t-I-yes-I-will-damn-it equivocation, treading irresolute water for a while and eliciting our sympathy for her indecision, before deciding on a course of action and eliciting our sympathy all over again; it’s the kind of thing you used to see in short fiction in women’s magazines, like Bella‘s “Tales with a twist”. To be fair, when it’s done sparingly, and when you’ve invested in the character, getting inside her head in this way can work really effectively. When it’s done over and over again – over the length of a novel – it just seems mechanical and exploitative. It also uses up a lot more words than just getting on with it would do – which may also have been a factor in those Bella stories, of course. And how our narrators like using up words! Was a character thinking about work as he put on his shirt and tie? Yes, he probably was. More specifically, he was “probably running through his schedule for the day – meetings, appointments, who, what, where.” Oh, that kind of schedule.

People do go to work – or rather, the men do; not one of the women has a job. But for the most part work is something that people just disappear off to and come back from. Even outside work, nobody has any hinterland at all – nobody reads books, nobody watches films, nobody even watches TV as far as we can tell. They go shopping, they go to the gym; the alcoholic goes to the pub, the young mum goes for coffee with “the NCT girls”[sic]; and, er, that’s it. About two thirds of the way through somebody mentions a band she used to listen to, and it’s like opening a window in a stuffy room.

And then there’s

5. The plot

Are you sitting comfortably? Then I’ll begin. An awful man was rejected by the Army and trained to be a surveyor instead (he just did, all right), then conned lots of money out of his parents and got married to a woman who worked in PR (no, I don’t know what he spent it all on). She couldn’t get pregnant, got depressed and started drinking heavily, and he started an affair with a woman who worked in a gallery. He also started threatening his wife and beating her up, but only when she was too drunk to remember what happened the next morning. Then he divorced her and married Gallery Woman, had a child with her and started an affair with the (married) childminder, who promptly got fed up with being a childminder (but carried on with the affair). Meanwhile PR Woman had taken to getting horribly drunk and harassing, or stalking, Awful Man – although very often when she did this she was too drunk to remember what had happened the next morning. When less drunk, during the day, she caught the train into London – where she worked until she was sacked for being drunk – which she could afford to do because reasons. ♠She got into a habit of looking at her old house, and then looking at the house a few doors down where Former Childminder lived with her husband, who was in fact a Very Slightly Less Awful Man. One day she saw FC kissing another man. In fact, Former Childminder had been seeing a Nice Therapist about her traumatic (but unrelated) backstory, and had fallen for him horribly, as patients in therapy often do; it was innocent! In any case, she’d decided to break off the affair with Awful Man. She then told her husband about it; being awful himself, he attacked her violently. She escaped and met Awful Man, who took her away in his car and (being a lot more Awful than she, or indeed we, had realised) killed her. PR Woman, who was in the area at the time stalking Awful Man, saw them in the car, but she forgot all about it the next day, due to being drunk. She then read about FC’s disappearance, put the police on to Nice Therapist, got him arrested and ruined his life, went and introduced herself to VSLAM under false pretences, tried to put the police on to VSLAM instead, and generally got in everyone’s way. (You can’t help feeling there’s a decent black comedy in here somewhere.) Eventually she remembered what she’d really seen and tried to rescue Gallery Woman from Awful Man. He admitted everything, there was a confrontation, and the two of them killed him, in self defence, sort of, not really. At the end of the story PR Woman still hasn’t got a job, but she’s given up drinking and her mother’s given her a lot of money (“Mum was quite generous when she discovered everything I’d been through”), so it’s a kind of happy ending.

It’s pants. The plot is driven by two completely unbelievable characters, one of them the main villain: a failed squaddie turned surveyor(!), a con artist, rapist and murderer, who is described as irresistibly charming but consistently portrayed as a thug. It only works at all because it starts in medias res – at the point marked ♠ above – and fills in the backstory afterwards. The dice are loaded further by the interleaving of the victim’s story, told over a longer timeframe; this makes it possible to withhold the (crucial) information about her and the therapist for a long time, and postpone the murder itself for even longer. The interleaving and the timelag don’t seem to have any other justification, though; once again, a technical device is used for no other reason than to create suspense. Not too much suspense, though. If, for instance, the entire story had been told through the eyes of PR Woman, without the interleaved chapters from FC’s point of view, the truth about the murder would have been deferred even further, leaving us in some doubt right to the end of the story. Running a supposedly unreliable narrative alongside a perfectly truthful account of the same events lets readers feel a frisson of doubt without ever really losing the comfort of omniscience.

6. Boozing, bloody well boozing

And what about that unreliable narrative? Everyone says this book’s got an unreliable narrator. It hasn’t. An unreliable narrator is a voice that tells you what the character believes (s)he’s seeing – or wants to believe (s)he’s seeing – but also lets you know that that’s happening, and leaves you not quite knowing what you can believe. (Read “Sundance”. Read some Ruth Rendell, come to that.) What we’ve got here is a narrator who drinks till she blacks out and wakes up not remembering the night before. Sometimes the memory comes back later, and she tells us what it was. Sometimes another memory comes back and she decides the first one was wrong, and she tells us that too. We know we’re reading a narrator who doesn’t know the whole story – and by the end of the book we know we’ve just read a narrator who was sometimes mistaken – but at no point do we know that what we’re being told isn’t reliable. Once again – yet again – a technical device which can be used to do amazing things is being used, clumsily, for no other reason than to postpone a crucial revelation.

I referred to two completely unbelievable characters, and the main narrator is the second; she’s nothing like any alcoholic I’ve ever read about, or any drinker I’ve ever known. She doesn’t seem to have an addict’s craving for alcohol; the way that she describes her longing for a drink, it’s not something she needs to make her feel normal or get her through the day. It’s just… a drink; she just really, really likes a drink.

The pre-mixed gin and tonic fizzes up over the lip of the can as I bring it to my mouth and sip.

I open one of the little bottles of Chenin Blanc I purchased from the Whistlestop at Euston. It’s not cold, but it’ll do.

I finish the second can and make a start on the third. The blissful rush of alcohol hitting my bloodstream lasts only a few minutes and then I feel sick. I’m going too fast, even for me, I need to slow down

(Halfway through the third drink? Never mind an alcoholic, she isn’t even a seasoned drinker.)

The pleasure she takes in the drink – in every drink – seems all wrong, as does the variety: red wine, white wine, pre-mixed G&Ts, shots of Jack Daniels (none of your rubbish)… It’s like a non-drinker’s idea of what it would be like to have a drink problem: alcoholism as a really extreme form of self-indulgence.

But it doesn’t have to be realistic as long as it serves its purpose. The booze, like most other things in this novel, is only a means to an end: the blackout. She drinks, she gets drunk, she remembers nothing. In a suspense novel that derives most of its suspense from withholding information, this is mightily convenient.

7. Tales with a twist

I suppose I should conclude by saying something positive about the book, but – without recourse to lying, distortion or cheating – I’m not sure what it would be. I wanted to carry on reading until I’d finished it, I’ll say that much. But that’s not so much praise as testimony to the effectiveness of its narrative devices. Beyond that, I don’t think it has any value at all. The characters are paper-thin, the story’s ridiculous, the gender politics is truly awful, and everything else in the book is a manipulative and rather badly-executed contrivance with no internal justification – the multiple narrators, the multiple time frames, the ‘diary’ structure, the amnesia.

But I guess the success of the book does tell us something interesting. You’d think that, as between the proverbial “good story well told” and a rather poor story dolled up with literary tricks and twists – it’s being told backwards! she’s a boy! they’re all ghosts! – the latter would only work on the jaded palates of the literary fiction crowd. The success of The Woman Who Saw Something From A Train – a novel which is nothing but literary devices, to the point that it would fall apart without them – suggests that there’s a real appetite for writing that takes readers outside their narrative comfort zone, even if it only goes a little way outside and doesn’t end up anywhere interesting. That’s encouraging news, at least for lovers of literary technique.

Advertisements

Soft enough for you

Anne:

Is it reasonable to have to learn to ride a bike but expect a computer to be as simple to figure out as a toaster? (Not the perfect analogy I know, but you know what I’m getting at…) Some days I think that user-friendliness was/is a really bad idea, not least because it’s obdurate, so hard to change.

If you have to work at using a technology, in other words, you necessarily end up working with it and through it. You work to adapt it to your needs – and you adapt it. Technologies which offer ease of use, by contrast, make it easy to work in certain pre-defined ways – and resist adaptation by the individual user. (There are, of course, technologies which are both easy to use and flexible – ask any Flickr user. But I think the ‘user-friendliness’ Anne is talking about here is more like the comment a tutor of mine once made on the BBC and ‘open access’ broadcasting: “They say they’ll come and help you, show you how to do it. They don’t, of course – what they do is show you how to do what you do because that’s how you do it.” User-friendliness is very often a matter of HTDWYDBTHYDI.)

But there’s more to it than that. What is this thing called obduracy? Anne again:

[Anique Hommels] argues that one way to emphasise the material aspects [of technologies in society] is to focus on their obduracy or resistance to change. (Imagine what it would *actually* take to replace the infrastructure that currently provides our electricity with something more sustainable.) The notion of obduracy is inextricably connected to embeddedness – a matter of interest to any kind of computing that seeks to become part of something else, be it an event, a habit, a skirt, a chair, a building, a street, a city. As Hommels reminds us, obduracy (or embeddedness) is a relational concept: “Because the elements of a network are closely interrelated, the changing of one element requires the adaptation of other elements. The extent to which an artifact has become embedded determines its resistance to efforts aimed at changing it.”

An embedded technology, then, would be one which has behind it a community of people who do a certain thing in a certain way. Becoming a user entails enrolment in that community. In short, the technology adapts you.

Where does this leave user-friendliness? Perhaps we could think of the embedding of a new technology as a process, which can continue to the point of the collapse of the possible ends and uses inherent in the technology and its reduction to the status of tool: a toaster, not a bicycle. And perhaps a ‘user-friendly’ technology – at least in the HTDWYD sense – is one designed to enlist a tool-using community and collapse its own potential into instrumentality.

(Relatedly, from Dan Hills’ essential critique of digital music: “there is a powerful necessity to think long term; to not take such short cuts which may inadvertently delete possible outcomes; to enable the flexibility and endless modifications seen in previous generations of music devices”. Dan has a lovely quote from William Gibson: “That which is overdesigned, too highly specific, anticipates outcome; the anticipation of outcome guarantees, if not failure, the absence of grace.”)

More broadly, what all this highlights is the value of difficulty, incompatibility, misunderstanding. Dan also led me (indirectly) to this quote from the late Derek Bailey:

There has to be some degree, not just of unfamiliarity, but incompatibility [with a partner]. Otherwise, what are you improvising for? What are you improvising with or around? You’ve got to find somewhere where you can work. If there are no difficulties, it seems to me that there’s pretty much no point in playing. I find that the things that excite me are trying to make something work. And when it does work, it’s the most fantastic thing.

One of the great frustrations in my work with ontologies and e-social science is the recurrent assumption that the concepts used in social science data can be documented cleanly and consistently – or, conversely, that if they can’t be documented cleanly and consistently they’re not worth documenting. The point, surely, is to find ways of recording both the logic of individual classifications and the incompatibilities between them – and the (qualified, partial) correspondences between them. And, of course, to make this documentation changeable over time, without effacing the historical traces which contribute to its meaning. Parenthetically, it’s worth noting here that preservation of historical data has nothing to do with obduracy. History is not obdurate, having no power to resist and (by and large) no enrolled community; the erasure of history can facilitate embeddedness and instrumentality, while the preservation of an artifact’s history may actually preserve resources of flexibility. (That’s enough abstractions – Ed.)

We are your friends

I recently attended an “e-Government Question Time” session, organised in connection with this conference. There were some good points made: one speaker stressed the importance of engaging with the narratives which people build rather than assuming that the important facts can be read off from an accumulation of data; one questioner called the whole concept of ‘e-government’ into question, pointing out that the stress seemed to be entirely on using the Web/email/digital TV/texting/etc as a mechanism for delivering services rather than as a medium for democratic exchanges. Much more typical, though, was the spin which the mediator put on this question as he passed it on to the panel:

That’s a very good question – what about democracy? And conversely, if it’s all democracy where does that leave leadership?

The evening was much less about democracy than it was about leadership – or rather, management. This starting-point produced some strikingly fallacious arguments, particularly in the field of privacy. The following statements were all made by one panellist; I won’t single him out, as they were all endorsed by other panellists – and, in some cases, members of the audience. (And no, identifying him as male doesn’t narrow it down a great deal. The people in the hall were 3:1 male to female (approximately), the people on stage 6:1 (precisely).)

I like to protect my own privacy, but I’m in the privileged position of having assets to protect. When you’re looking at people who have got nothing, and in many cases aren’t claiming benefits to which they’re entitled, I don’t think safeguarding their privacy should be our main concern.

At first blush this argument echoes the classic Marxist critique of the bourgeois definition of human rights – if we have the right to privacy, what about the right to a living wage? But instead of going from universalism to a broader (and hence more genuine) universalism, we’ve ended up with the opposite of universalism: you and I can worry about privacy, but it doesn’t apply to them. Superficially radical, or at least populist – you can just hear David Blunkett coming out with something similar – this is actually a deeply reactionary argument: it treats the managed as a different breed from the people who manage them (I like to protect my own privacy, but…). Management Fallacy 1: ‘they’re not like us’.

We’re talking about improving people’s life chances. We need to make personal information more accessible – to put more access to personal information in the hands of the people who can change people’s lives for the better.

Management Fallacy 2: ‘we mean well’. If every intervention by a public servant were motivated by the best interests of the citizens, safeguards against improper intervention would not be required. And if police officers never stepped out of line, there’d be no need for a Police Complaints Commission. In reality, good intentions cannot be assumed: partly because the possibility of a corrupt or malicious individual getting at your data cannot be ruled out; partly because government agencies have other functions as well as safeguarding the citizen’s interests, and those priorities may occasionally come into conflict; and partly because a government agency’s idea of your best interests may not be the same as yours (see Fallacy 3). All of which means that the problem needs to be addressed at the other end, by protecting your data from people who don’t have a specific reason to use it – however well-intentioned those people may be. One questioner spoke wistfully of the Data Protection Act getting in the way of creative, innovative uses of data. It’s true that data mining technology now makes it possible to join the dots in some very creative and innovative ways. But if it’s data about me, I don’t think prior consent is too much to ask – and I don’t think other people are all that different (see Fallacy 1).

I’ve got no objection to surrendering some of my civil liberties, so-called

Have to stop you there. Management Fallacy 3: ‘it looks all right to me’. The speaker was a local government employee: a private individual. His policy for handling his own private data doesn’t concern me. But I would hope that, before he came to apply that policy more generally, he would reflect on how the people who would be affected might feel about surrendering their civil liberties, so-called. (Perhaps he could consult them, even.)

Carry on:

I’ve got no objection to surrendering some of my civil liberties, so-called, if it’s going to prevent another Victoria Climbie case.

Management Fallacy 4: ‘numbers don’t lie’. (Or: ‘Everything is measurable and what can be measured can be managed’.) This specific example is a common error with statistics, which can be illustrated with the example of a hypothetical test for the AIDS virus. Let’s say that you’ve got an HIV test which is 95% accurate – that is, out of every 100 people with HIV it will correctly identify 95 and mis-identify 5, and similarly for people who do not carry the virus. And let’s say that you believe, from other sources, that 1,000 people in a town of 100,000 carry the virus. You administer the test to the entire town. If your initial assumption is correct, how many positive results will you get? And how confident can you be, in percentage terms, that someone who tests positive is actually HIV-positive?

The answers are 5900 and 16.1%. The test would identify 950 of the 1000 people with the virus, but it would also misidentify 4950 people who did not have it: consequently, anyone receiving a positive test result would have a five in six chance of actually being HIV-negative. What this points to is a fundamental problem with any attempt to identify rare phenomena in large volumes of data. If the frequency of the phenomenon you’re looking for is, in effect, lower than the predictable rate of error, any positive result is more likely to be an error than not.

Contra McKinsey, I would argue that not everything can or should be measured, let alone managed on the basis of measurement. (If the data-driven approach to preventing another Climbie case sounds bad, imagine it with the addition of performance targets.) Some phenomena – particularly social phenomena – are not amenable to being captured through the collection of quantitative data, and shouldn’t be treated as if they were.

What all these fallacies have in common is a self-enclosed, almost solipsistic conception of the task of management. With few exceptions, the speakers (and the questioners) talked in terms of meeting people’s needs by delivering a pre-defined service with pre-defined goals, pre-defined techniques, pre-defined identities (me service provider, you service recipient). There were only occasional references to the exploratory, dialogic approach of asking people what their needs were and how they would like them to be met – despite the possibilities for work in this area which new technologies have created. But then, management is not dialogue.

Social software may start with connecting data, but what it’s really about is connecting people – and connecting them in dialogue, on a basis of equality. If this goal gets lost, joining the dots may do more harm than good.

Know what I mean

Back here, I wrote:

Tagging, I’m suggesting, isn’t there to tell us about stuff: it’s there to tell us about what people say about stuff. As such, it performs rather poorly when you’re asking “where is X?” or “what is X?”, and it comes into its own when you’re asking “what are people saying about X?”

This relates back to my earlier argument that all knowledge is cloud-shaped, and that tagging is simply giving us a live demonstration of how the social mind works. In other words, all there is is “what people are saying about X” – but some conversations have been going on longer than others. Some conversations, in fact, have developed assumptions, artefacts, structures and systems within and around which the conversation has to take place. The conversation carried on in the medium of tagging isn’t at that stage yet, perhaps, but it will be – the interesting question is about the nature of those artefacts and structures.

Now (with thanks to Anne Galloway) over to Dan Sperber.

When say, vervet monkeys communicate among themselves, one vervet monkey might spot a leopard and emit an alarm cry that indicates to the other monkeys in his group that there’s a leopard around. The other vervet monkeys are informed by this alarm cry of the presence of a leopard, but they’re not particularly informed of the mental state of the communicator, and they don’t give a damn about it. The signal puts them in a cognitive state of knowledge about the presence of a leopard, similar to that of the communicating monkey — here you really have a smooth coding-decoding system.In the case of humans, when we speak we’re not interested per se in the meaning of the words, we register what the word means as a way to find out what the speaker means. Speaker’s meaning is what’s involved. Speaker’s meaning is a mental state of the speaker, an intention he or she has to share with us some content. Human communication is based on the ability we have to attribute mental state to others, to want to change the mental states of others, and to accept that others change ours.

When I communicate with you I am trying to change your mind. I am trying to act on your mental state. I’m not just putting out a kind of signal for you to decode. And I do that by providing you with evidence of a mental state in which I want to put you in and evidence of my intention to do so. The role of what is often known in cognitive science as “theory of mind,” that is the uniquely human ability to attribute complex mental states to others, is as much a basis of human communication as is language itself.

I am full of admiration for the mathematical theory of information and communication, the work of Shannon, Weaver, and others, and it does give a kind of very general conceptual framework which we might take advantage of. But if you apply it directly to human communication, what you get is a mistaken picture, because the general model of communication you find is a coding-decoding model of communication, as opposed to this more constructive and inferential form of communication which involves inferring the mental state of others, and that’s really characteristic of humans.
[…]
For Dawkins, you can take the Darwinian model of selection and apply it almost as is to culture. Why? Because the basic idea is that, just as genes are replicators, bits of culture that Dawkins called “memes” are replicators too. If you take the case of population genetics, the causal mechanisms involved split into two subsets. You have the genes, which are extremely reliable mechanisms of replication. On the other hand, you have a great variety of environmental factors — including organisms which are both expression of genes and part of their environment — environmental factors that affect the relative reproductive success of the genes. You have then on one side this extremely robust replication mechanism, and on the other side a huge variety of other factors that make these competing replication devices more or less successful. Translate this into the cultural domain, and you’ll view memes, bits of culture, as again very strong replication devices, and all the other factors, historical, ecological, and so on, as contributing to the relative success of the memes.

What I’m denying, and I’ve mentioned this before, is that there is a basis for a strong replication mechanism either in cognition or in communication. It’s much weaker than that. As I said, preservative processes are always partly constructive processes. When they don’t replicate, this does not mean that they make an error of copying. Their goal is not to copy. There are transformation in the process of transmission all the time, and also in the process of remembering and retrieving past, stored information, and these transformations are part of the efficient working of these mechanisms. In the case of cultural evolution, this yields a kind of paradox. On the one hand, of course, we have macro cultural stability — we do see the same dish being cooked, the same ideologies being adopted, the same words being used, the same song being sung. Without some relatively high degree of cultural stability — which was even exaggerated in classical anthropology — the very notion of culture wouldn’t make sense.

How then do we reconcile this relative macro stability at the cultural level, with a lack of fidelity at the micro level? … The answer, I believe, is linked precisely to the fact that in human, transmission is achieved not just by replication, but also by construction. … Although indeed when things get transmitted they tend to vary with each episode of transmission, these variations tend to gravitate around what I call “cultural attractors”, which are, if you look at the dynamics of cultural transmission, points or regions in the space of possibilities, towards which transformations tend to go. The stability of cultural phenomena is not provided by a robust mechanism of replication. It’s given in part, yes, by a mechanism of preservation which is not very robust, not very faithful (and it’s not its goal to be so). And it’s given in part by a strong tendency for the construction — in every mind at every moment — of new ideas, new uses of words, new artifacts, new behaviors, to go not in a random direction, but towards attractors. And, by the way, these cultural attractors themselves have a history.

There’s more – much more – but what I’ve quoted brings out two key points. Firstly, communication is not replication: in conversation, there is no smooth transmission of information from speaker to listener, but a continuing collaborative effort to present, construct, re-present and reconstruct shared mental models. The overlap between this and the ‘knowledge cloud’ model is evident. Secondly, construction has a context: the process of model-building (or ‘thinking’ as we scientists sometimes call it) is always creative, always innovative, and always framed by pre-existing cultural ‘attractors’. And these cultural attractors themselves have a history – you could say that people make their own mental history, but they do not do so in circumstances of their own choosing…

This is tremendously powerful stuff – from my (admittedly idiosyncratic) philosophical standpoint it suggests a bridge between Schutz, Merleau-Ponty and Bourdieu (and I’ve been looking for one of those for ages). My only reservation relates to Sperber’s stress on speaker’s meaning … a mental state of the speaker. I think it would enhance Sperber’s model, rather than marring it, to focus on mental models as they are constructed within communication rather than as they exist within the speaker’s skull – in other words, to bracket the existence of mental states external to communicative social experience. On this point Schutz converges, oddly, with Wittgenstein.

Sperber’s argument tends to underpin my intuition on tagging and knowledge clouds: if all communication is constructive – if there is no simple transmission or replication of information – then conversation really is where knowledge develops, or more precisely where knowledge resides. Sperber also helps explain the process by which some conversations become better-established than others; we can see this as a feedback process, involving the development of a domain-specific set of ‘attractors’. These would perhaps serve as a version of Rorty’s ‘final vocabulary’: a shared and unquestionable set of assumptions, a domain-specific backdrop without which the conversation would make no sense.

One final thought from Sperber:

The idea of God isn’t a supernatural idea. If the idea of God were supernatural, then religion would be true.

Well, I liked it.

%d bloggers like this: