If it were merely an amplifier, blogging would be uninteresting. But like Wikipedia, blogging harnesses collective intelligence as a kind of filter. What James Suriowecki calls “the wisdom of crowds” comes into play, and much as PageRank produces better results than analysis of any individual document, the collective attention of the blogosphere selects for value.

Other examples which are sometimes cited include digg, Yahoo! Answers, Wikipedia and del.icio.us. People come together to solve problems and their combined effort produces better results than an individual editor or news team could manage.

However, we’re actually smudging together two contrasting decision-making mechanisms here. Henry Jenkins points out in a post related to game design that there’s a significant difference between Pierre Levy’s idea of Collective Intelligence and James Surowiecki’s topic, The Wisdom of Crowds.

The Wisdom of Crowds emerges when data from a number of sources is aggregated. The people contributing need to be acting autonomously according to the best of their ability and in competition with others. The famous example is guessing the weight of the prize bull: the average of people’s guesses turns out to be the correct answer.

Collective Intelligence, on the other hand, emerges through deliberation, where people share, alter and evaluate other’s contributions to arrive at common ground.

As Jenkins notes, Wikipedia is much closer to this second model, Collective Intelligence, than the Wisdom of Crowds approach that finds the mathematical mean of all the suggested ‘answers’. The same would be true of Yahoo! Answers and del.icio.us, and indeed of most Web 2.0 applications that revolve around a community approach.

The Wisdom of Crowds model does in some ways apply, however, to things like the digg front page**, flickr interestingness and Google PageRank, which are algorithmically determined based on the combined anonymous and competitive input of many people.

It isn’t really a question of one of these models being better than the other, Jenkins concludes. It’s more that we’re not going to get very far unless we realise that they are two different things:

Both “collective intelligence” and “the wisdom of crowds” offer productive models for game design but we will get nowhere if we confuse the two. They represent very different accounts for knowledge production in the digital age and they will result in very different design choices.

I’d contend that the approach chosen by an application designer very much depends on the nature of the problem that is being addressed. Both could be correct depending on the situation, and probably one approach would be more sensible than the other for any given application. Completely anonymous postings to Wikipedia with no editing hierarchy whatsoever probably wouldn’t be such a great plan, though it would bring it closer to the wisdom of crowds model. On the other hand, the collective intelligence method of measured deliberation and discussion about which stories to put on the front page of digg or which sites should appear at the top of Google searches probably wouldn’t work out too well either.

[**Actually, digg is interesting in this regard. The submission of stories is not anonymous, nor is the voting. This has led to lots of accusations of bloc voting, allegations of a self-reinforcing elite of top diggers, and adjustments to the promotion algorithm to try to prevent this. It is a strange amalgam of social community and wise-crowds news aggregator. The owners (and presumably enough of the users) want it that way. If the owners didn’t want the social community aspect, and the problems that has created, they’d remove all mention of user names and make voting anonymous. It’s my belief that the gaming aspect to digg is entirely intentional and part of what appears to make it so addictive to its fans.]

I don’t think it is.

While many of the poster children of Web 2.0 are resolutely consumer — digg, youtube, myspace, wikipedia — their approach, and the 2.0 approach generally is about making things better for users, harnessing their input and aggregating them in clever ways. There’s nothing intrinsic to this agenda that segregates it from the business world. What’s sauce for the goose is sauce for the gander.

Users are not just your customers, they are also you and your fellow employees. Practices, tools and technologies which make life better for users are good for everyone. The separation between consumer and business applications is, in some senses, artificial.

If RSS makes it easy for consumers to read their favourite publications, then it is just as easy for employees to pick up the latest company information anywhere, on any device without logging into the intranet. If AJAX allows for more compelling, smoother and more immediate results for website customers then the same might be true of your CRM database. If mash-ups allowing the combination of maps and house prices work well for consumers, then mash-ups that bring together information from sales, accounts and marketing databases into a centralised overview are equally good news for managers.

The Wisdom of Crowds approach, for example, is not one that would typically be associated with multinational businesses. However, as I’ve noted before, corporations are already waking up to the idea that decisions and information can be better with input from a wider range of sources than the board room. Google, Microsoft and Eli-Lilly already use prediction markets as internal decision-making tools. Prediction markets are a form of stock exchange in which members might bet on the best-selling products and other strategy decisions. The belief is that if the members of the decision-making pool are autonomous, have a variety of insights and are self-interested, then their collective decision-making power will be extremely successful.

Other companies such as First Direct, BT, Dresdner Kleinwort Wasserstein, Mini and Nokia are using Wikis, real-time collaborative websites that can be generated and updated on-the-fly. Unlike intranets, they don’t require layers of permission and encourage employees to ‘chip-in’, correcting and expanding on other people’s knowledge. Built-in profile pages and complete documentation of all changes encourage the development of relationships between contributors and a sense of responsibility.

These are for used internal communication, to create a knowledge base or replace emails for important information. They’re also used for communications with customers, as a technical support or information tool that is co-created with users. Over 2000 organisations already use the SocialText paid-for Wiki service. Research firm Gartner predicts that Wikis will become mainstream collaboration tools in at least 50% of companies by 2009. In some companies, says SocialText, meeting times and email volume have already been cut by 50% through the use the collaborative tool.

Web 2.0 services such as CRM systems like salesforce.com or office applications like Google Spreadsheets have quite obvious business applications because they directly mimic tools that are already in use in the offline office. However, while switching to web equivalents can help cut costs, they truly become powerful to businesses with new approaches involving mobility, device agnosticism, openness, collaboration and real-time access to developing events and information.

Perhaps it’s the resistance to change that makes a new name, Enterprise 2.0, seem necessary. But perhaps what that really describes is the organisational change that needs to happen in order to get value out of any of these things, rather than the technologies themselves. For many businesses, I suspect, collaboration, lack of hierarchy, customer input and sharing are pretty foreign, scary concepts. In those cases, it’s not what sits on their computers that will be the big change.

The result is the fledgling MoSex Index (NSFW), an attempt to combine the best of social content with “taste networking.”

“Our social content engine is similar to Digg or Reddit, where members can post content and others can rate it. Based on content thresholds, who is doing the rating, karma and reputation, links make it to the homepage,” Daniel [Gluck] says.

“But — and I think this is an extremely important aspect of the project — we’ve also incorporated a collaborative filtering element. On other sites, you rate or (don’t) rate; we have a degree of rating based on a 7-point Likert scale that adds weight to your like or dislike. We take that intelligence and form ‘taste profiles,’ which combine to form ‘taste networks.’ When people in your taste network find content they like, that content gets recommended to you.”

The MoSex Index is barely launched, and like any community-based project, its usefulness will depend on the quality and quantity of its membership. So far, the user group has been small, with 70 testers posting links and adding ratings.

“Even with the small group, we’re seeing semi-accurate predictions, on a broad scale, of what users might like,” Daniel says. “Just with our museum staff, we can see who is most similar to whom. That’s one of the aspects that’s been most interesting to the staff, seeing whose tastes match.“

Update: Realised I hadn’t commented on this — which seems a little smug/prudish/unsatisfactory/lazy in retrospect. What happens is, you do the test and finally get through to the filth index. Then you rate the links given on a 1–7 basis. That’s quite sophisticated compared to digg, etc. where it’s a simple thumbs up. Theoretically, given a large number of users and large amounts of content, it could result in a pretty complex and accurate analysis of your tastes.

Two problems though, as I see it. (a) Digg is already getting lots of spam posts from people desperate for traffic and I can’t help but think this problem will be even worse in this field; (b) I’m not so sure people are very good at analysing (or admitting to) what they like. I’ve just read Blink, and while Gladwell doesn’t cover porn, he does discuss speed-dating and our inability to describe the people we really fancy.

The idea is to draw “smart crowds — groups of people configured to share intelligence”into collaboration at NewAssignment.Net and get stories done that way that aren’t getting done now. By pooling their intelligence and dividing up the work, a network of volunteer users can find things out that the larger public needs to know. I think that’s most likely to happen in collaboration with editors and reporters who are paid to meet deadlines, and to set a consistent standard. Which is the ‘pro-am’ part.

Rosen has already thought through and answered a lot of the immediate objections that might spring to mind (interest groups manipulating stories, sponsors balking at ‘inconvenient truths’ & local stories, volunteers will be nutjobs with an agenda).

Clever stuff. But … there’s a cute parallel here with the ongoing debate about roles of experts and citizen contributors in the proposed Citizendium projects. In some respect, NewAssignment sounds like a Citizendium for news.

What I think the difficulties experienced at Wikipedia and those at digg.com show us is that collective intelligence and wise crowds sound great. Really, I am a believer. But that they are very difficult to orchestrate. One difficulty with news that doesn’t appear in mainstream papers is that it’s often very contentious. What will motivate the unpaid contributors to NewAssignment, if it isn’t their own nutty agendas? If the editors refuse to follow the promptings of the crowd, because they’re all nuts, what will happen then?

The examples Rosen gives of editor bloggers developing a huge following only partly help explain what will happen:

Part of it is the example now being set by liberal journalist and blogger Josh Marshall. His Talking Points Memo blog is invaluable if you follow national politics; he’s widely read on Capital Hill. During the 2004 campaign he raised money for a trip to New Hampshire to hear and question the candidates. He told readers why he wanted to go, what he thought he could accomplish.

The essential transaction I’m counting on is right there. Users fund an act of journalism because they have confidence — a lot– in who’s doing it and why; the chances of getting something really good back seem pretty good.

So the editor needs to be a maven. S/he develops a cult of personality around what they do, strong enough to win the trust and support of large numbers of readers. NewAssignment editors will also have financial backing from sponsors and so won’t need to pass the cup around their readers.

Instead, the readers suggest topics for investigation, and help provide data. Presumably, the editors then use their discretion to choose the non-nutty options. So this is a news site that might cover the plight of illegal torrent sites one day, dental amalgam the next and Panda Baiting the day after. And at the same time, despite flipping from subject to subject, the editor is developing a cult following.

No. That wouldn’t work. You’d have a number of sites. Each of them would only investigate subjects around one quite narrow area, an area the editor is already passionate and knowledgeable about. Probably with an established audience. S/he would get tips and ideas from readers which would decide the topics within the subject that get written about.… Err sounds quite a lot like a blog network, eh.

NB: there are a huge number of alternative news sites already in operation. If you’re like me, you’ve only ever heard of a handful. Why? Well, I expect you haven’t got time or you don’t find them trustworthy.

In 2003, for example, the Pentagon set up a prediction market for world events including terrorist activity, FutureMAP. It was rapidly closed down again when the press discovered this “People Betting on Terrorism Outrage!!!” However, as Time magazine reported last year, this wasn’t a clever move:

…the fact is, betting on terrorism actually makes sense. Consider the investigation just launched in Washington over an apparent leak at the Department of Homeland Security in which insiders seem to have tipped off relatives about an alleged threat to the New York City subway system. Outrageous behavior? Perhaps. But get those “insider traders” into a market, and everyone will have access to that information. Insiders have a motivation — money, at the basest level — to distribute their knowledge.

The Hollywood Stock Exchange (HSX.com) uses pretend money to buy and sell the “stocks” of movies and stars. The predictions made by the exchange are so accurate that the company sells its valuations and opening weekend predictions to film production studios. Popex (now defunct) did the same for the fortunes of pop acts. The Iowa Electronic Market’s election predictions have outperformed major national polls, even months ahead of the election. The IEM’s predictions have been out by just 1.37% in presidential elections, 3.43% in other US elections, and 2.12% in foreign elections. This is despite the fact that members of the Exchange have mainly been men (viz. not women) from Iowa, and so not representative of the voting public.

So, interesting to hear about the launch of Inkling on Techcrunch. Basically, it’s a site for betting on future events in a stock-market simulation using pretend money, inkies. More innovatively, the service offers Enterprise and Small Business versions of the service. The idea is that businesses create stock exchanges within their companies to facilitate decision-making. Google, Eli-Lilly and Microsoft already use prediction markets internally for this purpose.

You might set up a market for example, on which of your four products are going to be best-sellers. You open that up to as many people as you like — arguably your whole company, or even your customer base. The market should involve real rewards for correct answers. This would be the best approach because you need people to care about the answers they give. Because you are getting varied, independent, self-interested perspectives, the answer given by the market is likely to reveal the truth.

This is better than a poll, because (a) it blends different answers to set a price for the likelihood of various alternatives and (b) it gives people a reason to use their best intelligence, intuition and insider knowledge because there’s a reward for them at the end for betting on what they really think will happen.

In a technology company for example, the sales people probably know your customers best, but the engineers know the real capabilities and limitations of the product. Marketing, on the other hand, has an idea on what people like and what attracts press coverage. The secretary may have a bit of insider knowledge. The cleaner may not know technology, but he knows what sounds good. Combine all those perspectives in the correct way and you end up with a far more complete picture than other methods are likely to achieve.

From the FAQ:

What is a Prediction Market?

Prediction markets allow a group of people to express an opinion over a period of time about the probability of an event occurring. A question is posed and people buy and sell shares in stocks representing possible answers to that question. The highest priced stock at the end of a period of time is the group’s prediction.

Prediction markets are not only useful at forecasting the future, they also provide invaluable insight in to the perception of a group of people, information that can be just as valuable as input to an upcoming decision or action as an accurate forecast.

Inkling is not the first prediction market, but we believe it sets the standard for ease of use and setup, thus capturing the true ethos of what prediction markets are all about: wide participation of a diverse group of people to collectively state an opinion.

Interesting news for digg:

As usual, the criticism is based on a misunderstanding of what the book actually says. It does not say that big groups of people make the best decisions. It says that they are likely to, under the correct conditions. The crowd needs to consist of people who are:

diverse
qualified
independent
self-interested

The interactions between the crowd needs to be carefully managed to avoid social factors distorting an individual’s best judgement. In addition, some problems — crossword puzzles, guess the weight of a prize bull, sports results, open-source software — are a lot more tractable to the approach than others — the most interesting news or the best pop album.

The old digg — which allowed bloc votes from groups of friends and pressure groups — fell down on more than one of these criteria. The book’s author, James Surowiecki, comments: “The thing that makes the wisdom of crowds work is lots of diverse opinions and independent judgments … Digg acknowledged it wanted more diversity of input.”

Personally, I think that at best digg can produce a front page that’s interesting to its typical user, which is fine. Similarly, the hit charts are only going to show what most people like, not necessarily what’s best for everyone. Problems like these involving qualitative judgements can only be solved when you have a crowd of people with similar tastes to you — which is why last.fm works for music recommendations and why a different news-voting community such as reddit or a CrispyNews group might be better for you.

Wikipedia is an interesting case and whatever problems it has aren’t going to be solved by tweaking an algorithm. I liked Jimmy Wales’ comment in the Wall Street Journal, though, that “…it is a misunderstanding to think of ‘openness’ as antithetical to quality. ‘Openness’ is going to be necessary in order to reach the highest levels of quality.”

I’m a fan of both digg and reddit, and it’s interesting to see the differences between what both crowds consider to be the day’s top stories. Such differences speak volumes about the wisdom of crowds, which both sites tout as their driving force. The answers found to problems (i.e. “what are the most interesting news stories today?”) depend considerably on who is in the crowd and the norms set for what makes an something interesting. Perhaps such a subjective issue is not really amenable to the approach except with groups that already have a general consensus about such matters. If that is the case, then perhaps the approach of Crispy News, which allows users to form their own news-voting communities, might offer an interesting alternative.

University of Virginia graduates Steve Huffman and Alexis Ohanian started the website in June 2005 with the help of the Y Combinator venture firm in Cambridge, followed by an unnamed angel investor who funded the site through August.

Ohanian said the [PC World] magazine’s approval will give his team added motivation. “It’s mind-blowing for us,” he said.

Reddit, which attracts 50,000 visitors a day, reached profitability in April, Ohanian said. The company doesn’t disclose revenue figures.

Y Combinator partner Paul Graham said Reddit’s users circumvent the traditional news model, choosing the most interesting news stories of the day. “It lets the wisdom of crowds do what editors used to do,” he said.

Reddit hits profit as its technology nets praise in market.