On-demand Media

Personalised news is not news

Scott Karp wonders if news is a fundamentally shared experience, and muses that "purely personalised news may be too much control, a slippery slope that leads to solipsism" -- an argument made by Cass Sunstein, on which I have written before. I agree, but I think the risk of this happening is not that great.

The orthodoxy says that we consume news to know facts about the world we live in (or the industry we work in) so that we can act intelligently. But the need to read what others are reading is also a basic, primary need - even if the 'facts' themselves are of no use beyond that.

You could say there are two kinds on news consumption:

1. Informational consumption - e.g. when a stockbroker needs to be across what a firm is doing. Here sharedness is unnecessary and sometimes undesirable (although, interestingly, it is sometimes mandated by law)
2. Ritual consumption - e.g. when I read the paper over coffee to be 'up to date'

Of course all news consumption is a mix of both of these types: we do get the odd useful fact from reading newspapers, and stockbrokers also need to be and feel part of a community of peers. But I suspect the second type is far more important in accounting of everyday general-purpose news.

If right, this means that there is an enduring consumer demand for something shared, and that personalisation will never fully take off in general-purpose news services. It also means that the 'static' nature of much of today's web may be a happy accident and an enduring feature: for all the talk about web 2.0 Ajax magic, the web is a shared space inasmuch as what I see is what you see (see my earlier post on this).

It also means that part of the consumer demand for news is about knowing not what happened, but what are the stories, the headlines that make up today's news. That demand cannot be answered by a linear narrative (a story) but by a selection, a list, a programme that tells us what the news are. Aggregation is content.

News and coffee

In a recent post Tim O'Reilly wonders why short-form content is successful on the web, and offers as explanations that (i) it is easier to read, and (ii) it is more 'modular' - making it easier to build aggregate content collaboratively.

Perhaps. But I would add a third explanation: because the web is PC-based. PCs are unsuitable in those places, physical positions and moods where we like to enjoy a long read; and PCs are often used at the office, where it's not acceptable to spend an hour reading something that is not work-related. As to video, PCs are used by a single person at a time, and so the social inappropriateness of compulsive channel-flicking doesn't apply (but that won't carry over to internet-TV beyond the web).

To keep beating my old drum, people don't want content. They want to read a certain book in bed, watch TV with their partner in a warm room, read the paper over breakfast, watch a movie over Indian food, or retire to the study to browse the web. Content only exists as a (sometimes senior, sometimes junior) partner in a larger dance (the medium) in which such banalities as ergonomics and household routines play a large role. The medium doesn't just change "the format in which content is delivered"; without an appropriate medium to exist in, content would not get made.

(Of course people do consume and want content, but only in the sense that when I want an ice-cream I want sugar; the idea is to sell the ice-cream, not the sugar. And of course you get some people who just need sugar, but you get my point.)

Important stories and good content

Via David Weinberger I read about an interesting discussion at Foocamp. The question was "how to decide which [news] stories are important and interesting without human intervention", and the various answers propose different ways of harnessing the wisdom of the crowds: via voting (as in Digg), by tracking commentary (Techmeme) or watching users reading the news (Newsvine). Never mind that all these approaches involve human intervention: the point, I suppose, is not so much about humans generally as about about editors specifically.

If you've been reading this blog for some time you will know that I question some of the premises behind such attempts: namely, that human editors offer a unique value that machines or crowds can't provide - namely, expression and commitment (I won't repeat my arguments here -- you can read them in an earlier post).

But the aggregators above have an even more fundamental problem: they are not automatic at all. Inasmuch as they rely on popularity they are not autonomous. Nothing would be popular if it wasn't discovered in some way first; and that initial discovery usually happens via an authored, 'manual' aggregate.

 

There is a school of thought (which may or may not be subscribed to by the people in the discussion above) that holds that, in the future, media consumption will only be a matter of finding the 'content that's right for you', and that hence human aggregators will become redundant. Its proponents cite examples like YouTube that can identify 'good' or 'popular' content long before any human aggregator does. These are animals of a different species from Digg or Newsvine and deserve separate treatment (see my later post on this).

For now, I would like to comment on some basic assumptions that inform this second school of thought, assumptions about nature of two types of thing: content and people.

In this view, the world of content is made up of discrete, self-contained things (pieces of 'content'), each with its properties. These properties (subject matter, style, quality, etc) may be factual (e.g. how long a video clip is), matters of unarguable judgement (e.g. that a clip is about George Bush), or matters of subjective preference (e.g. that 80% of people 'like' a clip).

On the other hand we have another type of thing: people, i.e. those who consume content. They also have their properties: their preferences, the communities they belong to, their demographics, age, sense of taste, interests, past, etc

In the middle we have media, and a challenge: how to pair up these two types of things - people and content - in the best way. The basic assumption is that this best way is a function of the properties of people and content - i.e. a matter of compatibility. To be sure, the picture is sometimes more complicated: to be realistic we also need to consider people's moods, their expressed desires (e.g. via a search query), and time (only fresh news are news). But the basic -compatibility- idea still applies.

Powerful as this view may be, note what it misses:

1. It doesn't account for people who want to see what a trusted human editor may want to offer. This is crucial in the case of news
2. It makes authorship and attribution secondary: good content is good content is good content, quite apart of where it came from
3. Assuming that human editors are here to stay (a point I've argued before), it fails to account for how these will influence popularity, and hence what 'automatic' aggregators will offer. I'll write more about this in a later post
4. It fails to account for content that is consumed because others are consuming it. By presenting consumption as a matter of compatibility, it suggests that media habits are a selfish affair, aimed at personal, self-contained satisfactions and not at communing with others or communicating as part of the ongoing conversation that is social life (even watching Terminator is a conversational act)

I will have more constructive things to say about this in later posts.

Reading Stephen Ward

I've been reading Stephan Ward's excellent book on the history of journalism ethics. From page 118:

Everything stems from periodicity. The publication of a weekly or daily paper has to overcome economic, technical, and editorial problems. But, once established, the periodical then has to maintain a precarious contract of "trust" and presumed reliability with readers. Each week, newsbooks had to attract readers and assure them of their reliable news service and cogent opinion. They had to provide the sorts of information that readers demanded, in the style and tone expected. Commercial competition added to the pressure, prompting editors to characterize competitors as untrustworthy and devoid of editorial standards. Periodic news publishing "set up a continuing relationship... between reader, printer, and the originator of information" (from Smith, p 9). The publisher of the occasional broadside or pamphlet lacked a constant and close relationship with a recognizable readership. [My emphasis]

This resonates with what I've been saying here about aggregates: it is only through periodically-updated media 'properties' that media establish direct relationships with audiences; properties define, and are partly defined by, the consumption communities that make up their audiences. These relationships are formed through habits, which are in turn based on socially-shared practices.

Properties are 'aggregates' in two senses: they are destinations that always offer their visitors a selection of unexpected, surprising content; and they also aggregate in time, giving the recurrent practice a coherence that reflects the curator's voice, over time building a more or less coherent ongoing narrative about the world. It is not that a brand alone can define a personality; it is this voice and this narrative that define a media brand.

What is currently shaking media radically is the sudden ease with which anyone can set up a media destination (in the sense above), and not just the ease in publishing and consuming 'content', or in finding the content one 'wants'. Without destinations content would go unknown, undesired, unlooked-for, and hence not found.

Broccoli content, public media and aggregates

Reflecting on the future of public media, Jeff Jarvis, paraphrasing Scott Karp, writes:

In a one-size-fits-all product world, broccoli news of the sort Scott cites rode along with the bon bons of lifestyle and celebrity coverage. Yet a newspaper with no broccoli would not sell as a newspaper (it instead becomes one of too many magazines like that). So veggies still have value. But when one-size-fits-all media is cut up and unbundled, when advertisers can choose to sponsor this piece of content and not that, then the risk to necessary and unhappy news becomes greater.

Jeff's post is about funding for public-service content, especially for 'brocoli' - i.e. socially worthy but unpopular - content. That's an interesting discussion, but here I want to focus on something else: the operative word above, "along" (my emphasis).

Broccoli content can't work without:

• Non-broccoli content to go on the side

• A 'place' where the two sit side by side.

Those 'places' are what in this blog I've called aggregates. Aggregates can deliver huge public/social value - as big as the content they aggregate. They do this via (i) serendipity and (ii) shared experiences, and are the embodiment of the public sphere.

In his book Republic.com, Cass Sunstein identified serendipity and shared experiences as key values of old media. His concern is that new media, because of personalisation and unbundling, put these values at risk - leading to a society where everybody reads The Daily Me and social cohesion is lost. Jeff's point is related, although it's more about serendipity than shared experiences.

While Sunstein's concern is justified, I think the risk may not be as serious. As I've argued before, non-personalised, massively-shared aggregates will never go away (whatever the web 2.0 enthusiasts may say). This is not just because the trend away from one-to-many media is a slow one: it is also because the trend will never take fully over; audiences will always appreciate the value of shared experiences. Just consider the enduring appeal of web portals: non-personalised, massively shared, editorially-curated destinations that are visited by millions as a matter of daily routine: surely a sign of an enduring relationship.

Imagine consuming news without serendipity (the unexpected) and shared experiences ('being in the loop'): what you are left with is just information--which will never do as a replacement for news. Serendipity and shared experiences will endure, quite simply, because they are 'what people want' - often more than they 'want' the content itself.

More on blogging and speech-acts

To continue my earlier post on blogging and JL Austin's theory of speech acts...

To scandalously summarise Austin: it is one thing to read (or hear) the utterance 'it's going to rain', and quite another to understand it as a forecast, a guess, a warning, or a decision that a game of tennis needs to be called off. How the utterance is to be understood (its 'illocutionary force') depends, loosely, on its context--not just in the sense of taking into account the utterances that precede and follow it, but also on who the speaker is, what conventional role he is assuming (e.g. a weather forecaster), the occasion (is the forecaster just guessing when chatting to a friend, or is he announcing his official forecast on TV?), and other factors.

To go back my older theme of linking and content, you could say that a disaggregated, isolated piece of content is like an utterance without its illocutionary force (a mere 'locution', in Austin's parlance). An attributed, branded piece has more illocutionary force, but not quite as much as when it is placed in the bundle (or aggregate) for which it was intended - e.g. a weather forecast is truly official when it is part of, e.g., a news bulletin (Jaron Lanier recently made a related point - see my post on this).

So far so good. About what follows I'm far less clear.

It seems as if the web, with its hyperlinks, has the effect of destabilizing the illocutionary force of content. This is partly because of disaggregation - i.e. because content can be found in isolation, outside the bundles (e.g. a weather site) for which it is produced. But another, more mysterious effect seems to be at play as well. By linking to someone else's disaggregated utterance I am, in some vague way, setting the illocutionary force in which that utterance will be taken.

Perhaps illocutionary force is not the appropriate concept here. Value might be a more appropriate concept - i.e. when I link I make a value statement about what I link to - but this will not do: not all linking is a matter of recommendation; when I link to something I may not be implying that the linked-to content is 'good', or even that you might 'like' it (an interesting word). I may instead be providing evidence ('go see for yourself') or trying to help you discover or find something. And what am I valuing, the content or its author? Relevance might seem to be a better concept.

An unrelated point: To an extent, this is not new to the web. When I review a film, or when I quote someone else, I am somehow setting the context in which the referred-to content will be received (Peter Morville makes a similar point, albeit in a somewhat different context: "when considering a decision, we are unduly influenced by the first information we find"). But hyperlinks are somehow different. This may be to do with immediacy--referencing and referred-to content are consumed in quick succession.

No answers and little clarity. This is work in progress.

Journalists, blogs and speech-acts

A post by Bill Thompson last month triggered an interesting debate. One of the controversial bits:

Unlike amateur bloggers who can rant, comment, express bizarre points of view or promote their latest acquisitions and obsessions with no concern for conflict of interest or even internal consistency, we are not mere citizens in the world of the blog and the MySpace profile, and it is about time we stopped trying to act as if we ever were.

Read also an interesting follow-up by Jeff Jarvis, with lots of good comments.

In my view, the key point here (which Bill may or may not share) is that what journalists say is not just opinion or 'information' (whatever that hideous term means). Their utterances are examples of what JL Austin called speech-acts: real-world, binding actions that just happen to be performed through words.

The archetypal case of speech-acts is that of the "declarative" or "directive" utterances of (respectively) juries or judges, as in "we find the defendant guilty" or "I sentence you to 10 years in prison". These are not just "constative" statements (which can be true or false): they are actions, and the world changes when they are performed successfully. Among the factors that are key to their success (or 'felicity', in Austin's parlance) are that the speaker have the necessary role (e.g. be a judge) and that the utterance be made at the right time in the right conventional setting (e.g. the passing of sentence at a lawfully-convened trial).

Bill Thompson's point, then, is that journalists can never escape their role (they are always journalists) even when they are outside their traditional setting (e.g. a newspaper). The truth of this is debatable, but it is certainly not altogether lacking. Imagine a judge giving a press conference after passing a sentence, saying "I'm not quite sure I should have condemned him... but so be it". It just wouldn't do. A person can't be completely divorced from her role, and a role can't be completely divorced from its official setting.

So what kinds of speech-acts is it that journalists make? At least three:

1. They make "comissive" statements of fact: they promise that something is true, and stake their reputations on this

2. They (at least editors) make the "directive" recommendation that you should be aware of what is being said (see my post on aggregates)

3. They (at least mainstream-media editors) make the "declarative" utterance that this story (as distinct from countless others that could have been written about countless other events) is from now on "news" and part of the national conversation.

Two questions remain:

First, do you have to be a journalist to make a valid speech act about current events? Certainly not, but unless you own a powerful aggregate (like the Daily Kos) your statement doesn't have the force of (3). Most bloggers' opinions even fail at point (1).

Second, are blogs changing the conventions that dictate what makes a valid journalistic speech-act? Certainly, but exactly in what way is something that escapes me.

If these arguments seem persuasive, you will agree with me that people like Adrian Holovaty who claim that journalism is about "gathering", "distilling" and "presenting" "information" are either hopelessly miguided or being generic to the point of meaninglessness.

 

Later: See this post by Clay Shirky (18/09/06)

News on Internet TV: shaping the medium

As David Jay Bolter and Richard Grusin argue in their book, all successful media have started by repackaging existing content borrowed from older media; and they all eventually produced new types of content which, although owing much to older formats, constituted new genres in their own right.

News on Internet TV will be no exception. Much of what we see today is just repackaged broadcast content: we see TV channels available as streams on the internet, or news sites that package clips from TV news bulletins in navigable ways.

In the early days of the web, newspaper websites started by reproducing their print editions. Eventually a new form of journalism, one more suited to the new medium, developed, and and only then did online journalism take off.

What did web journalists have to do to achieve this? At least four things:

• Time: the daily deadlines had to be eliminated. The web is a 24-hour medium where people expect to see something current. Surprisingly, broadcasters (e.g. CNN, BBC) were better prepared for this than newspapers, even though text was not an expertise they had initially.

• Writing style: Writing style had to evolve into a web-friendly form. People in the web don't read long articles; this was partly because it's just unpleasant to stare at text on the screen for too long, and partly because of the way that computers are embedded in our daily life (most internet news reading happens at work, where it is not acceptable to read the news for more than a few minutes).

• Design conventions: A new medium had to be designed. All news sites now have the basic structure of  (i) a constantly updated front-page, (ii) subject indices accessed through a navigation bar, and (iii) story pages. It is not just that this design is a good one; what is crucial is that it is a standard one that users don't need to learn as they go from site to site.

• Links: You are failing to exploit the power of the web if you don't use hyperlinks. Users expect links to related and unrelated content, and sites that fail to deliver this look at best poor and at worst neglectful.

A similar evolution is now beginning with video, and some areas are more advanced than others:

• Time: Many sites have this right already. The video services of all the major broadcasters are updated around the clock

• Style: This is harder. Although there are some notable exceptions, for cost reasons it is hard to create video content that is not meant for broadcast distribution. But as on-demand platforms  gain audiences, this will become more and more possible. What the new style will be is another question, but it will certainly involve long-form content (e.g. full interviews and speeches) that is unsuitable for broadcast

• Design conventions: Some conventions are developing: Many sites have dedicated video services that usually play in a dedicated window where text is less prominent than elsewhere and some navigation and promotion is provided. Still, there is a lot more variety here than in the text web.

• Links: This is a very undeveloped area. Apart from promotions, navigation and the occasional link to related clips, linking from video products is limited. There is much to be developed here, and it will not all be of the form that we know from the (text) web: with video, links need not be associated to a piece of video in its entirety; a link can be relevant to only a segment, or even to only a part of the screen for a few seconds.

New ways of producing aggregates will also play a key role. In the text web aggregates are shown as lists of links: they make up front pages (where each link has an assigned prominence and text) and they also appear whenever related content need to be shown. With video, aggregates need not take this form; they can be concatenation of clips (which is what a TV bulletin is), which could be set editorially but also influenced by user's actions--resulting in a mutation of TV's sense of 'flow'.

In this post I've focused on the web--i.e. on video meant to be watched on a computer with a web browser, a keyboard, etc. Of course this is not where all of Internet TV will happen (indeed it may end up being a minor platform), but, because of the low costs of experimenting on it, the web offers the best window into what may be commonplace in five years' time.